CDO User Guide

Climate Data Operator

Version 1.9.6
February 2019

Uwe Schulzweida – MPI for Meteorology

Contents
1. Introduction 6
1.1. Building from sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.1.1. Compilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.2. Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2. Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.1. Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.2. Environment variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.3. Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.4. Operator chaining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.5. Parallelized operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.6. Operator parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3. Horizontal grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.1. Grid area weights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.2. Grid description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4. Z-axis description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.5. Time axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.1. Absolute time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.2. Relative time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.3. Conversion of the time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.6. Parameter table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.7. Missing values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.7.1. Mean and average . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.8. Percentile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.8.1. Percentile over timesteps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2. Reference manual 18
2.1. Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.1.1. INFO - Information and simple statistics . . . . . . . . . . . . . . . . . . . . . . . . 20
2.1.2. SINFO - Short information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.3. DIFF - Compare two datasets field by field . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.4. NINFO - Print the number of parameters, levels or times . . . . . . . . . . . . . . . 23
2.1.5. SHOWINFO - Show variables, levels or times . . . . . . . . . . . . . . . . . . . . . . 24
2.1.6. SHOWATTRIBUTE - Show attributes . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.7. FILEDES - Dataset description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2. File operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2.1. COPY - Copy datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.2.2. TEE - Duplicate a data stream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.2.3. REPLACE - Replace variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.4. DUPLICATE - Duplicates a dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.5. MERGEGRID - Merge grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.6. MERGE - Merge datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.2.7. SPLIT - Split a dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.8. SPLITTIME - Split timesteps of a dataset . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.9. SPLITSEL - Split selected timesteps . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.2.10. DISTGRID - Distribute horizontal grid . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.2.11. COLLGRID - Collect horizontal grid . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.3. Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.3.1. SELECT - Select fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.3.2. SELMULTI - Select multiple fields via GRIB1 parameters . . . . . . . . . . . . . . . 40
2.3.3. SELVAR - Select fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

2
Contents Contents

2.3.4. SELTIME - Select timesteps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

2.3.5. SELBOX - Select a box of a field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.3.6. SELGRIDCELL - Select grid cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.3.7. SAMPLEGRID - Resample grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.4. Conditional selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.4.1. COND - Conditional select one field . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.4.2. COND2 - Conditional select two fields . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.4.3. CONDC - Conditional select a constant . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.4.4. MAPREDUCE - Reduce fields to user-defined mask . . . . . . . . . . . . . . . . . . 51
2.5. Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.5.1. COMP - Comparison of two fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.5.2. COMPC - Comparison of a field with a constant . . . . . . . . . . . . . . . . . . . . 54
2.6. Modification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.6.1. SETATTRIBUTE - Set attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.6.2. SETPARTAB - Set parameter table . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.6.3. SET - Set field info . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.6.4. SETTIME - Set time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.6.5. CHANGE - Change field header . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.6.6. SETGRID - Set grid information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.6.7. SETZAXIS - Set z-axis information . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.6.8. INVERT - Invert latitudes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.6.9. INVERTLEV - Invert levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.6.10. SHIFTXY - Shift field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.6.11. MASKREGION - Mask regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2.6.12. MASKBOX - Mask a box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.6.13. SETBOX - Set a box to constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.6.14. ENLARGE - Enlarge fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.6.15. SETMISS - Set missing value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.7. Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.7.1. EXPR - Evaluate expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.7.2. MATH - Mathematical functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
2.7.3. ARITHC - Arithmetic with a constant . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2.7.4. ARITH - Arithmetic on two datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
2.7.5. MONARITH - Monthly arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
2.7.6. YHOURARITH - Multi-year hourly arithmetic . . . . . . . . . . . . . . . . . . . . . 84
2.7.7. YDAYARITH - Multi-year daily arithmetic . . . . . . . . . . . . . . . . . . . . . . . 85
2.7.8. YMONARITH - Multi-year monthly arithmetic . . . . . . . . . . . . . . . . . . . . . 86
2.7.9. YSEASARITH - Multi-year seasonal arithmetic . . . . . . . . . . . . . . . . . . . . . 87
2.7.10. ARITHDAYS - Arithmetic with days . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2.8. Statistical values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
2.8.1. TIMCUMSUM - Cumulative sum over all timesteps . . . . . . . . . . . . . . . . . . 95
2.8.2. CONSECSTAT - Consecute timestep periods . . . . . . . . . . . . . . . . . . . . . . 95
2.8.3. VARSSTAT - Statistical values over all variables . . . . . . . . . . . . . . . . . . . . 96
2.8.4. ENSSTAT - Statistical values over an ensemble . . . . . . . . . . . . . . . . . . . . . 97
2.8.5. ENSSTAT2 - Statistical values over an ensemble . . . . . . . . . . . . . . . . . . . . 99
2.8.6. ENSVAL - Ensemble validation tools . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
2.8.7. FLDSTAT - Statistical values over a field . . . . . . . . . . . . . . . . . . . . . . . . 102
2.8.8. ZONSTAT - Zonal statistical values . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
2.8.9. MERSTAT - Meridional statistical values . . . . . . . . . . . . . . . . . . . . . . . . 105
2.8.10. GRIDBOXSTAT - Statistical values over grid boxes . . . . . . . . . . . . . . . . . . 107
2.8.11. VERTSTAT - Vertical statistical values . . . . . . . . . . . . . . . . . . . . . . . . . 108
2.8.12. TIMSELSTAT - Time range statistical values . . . . . . . . . . . . . . . . . . . . . . 109
2.8.13. TIMSELPCTL - Time range percentile values . . . . . . . . . . . . . . . . . . . . . . 110
2.8.14. RUNSTAT - Running statistical values . . . . . . . . . . . . . . . . . . . . . . . . . . 111
2.8.15. RUNPCTL - Running percentile values . . . . . . . . . . . . . . . . . . . . . . . . . 112
2.8.16. TIMSTAT - Statistical values over all timesteps . . . . . . . . . . . . . . . . . . . . . 113
2.8.17. TIMPCTL - Percentile values over all timesteps . . . . . . . . . . . . . . . . . . . . 114
2.8.18. HOURSTAT - Hourly statistical values . . . . . . . . . . . . . . . . . . . . . . . . . 115

3
Contents Contents

2.8.19. HOURPCTL - Hourly percentile values . . . . . . . . . . . . . . . . . . . . . . . . . 116

2.8.20. DAYSTAT - Daily statistical values . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
2.8.21. DAYPCTL - Daily percentile values . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
2.8.22. MONSTAT - Monthly statistical values . . . . . . . . . . . . . . . . . . . . . . . . . 119
2.8.23. MONPCTL - Monthly percentile values . . . . . . . . . . . . . . . . . . . . . . . . . 120
2.8.24. YEARMONSTAT - Yearly mean from monthly data . . . . . . . . . . . . . . . . . . 121
2.8.25. YEARSTAT - Yearly statistical values . . . . . . . . . . . . . . . . . . . . . . . . . . 122
2.8.26. YEARPCTL - Yearly percentile values . . . . . . . . . . . . . . . . . . . . . . . . . . 123
2.8.27. SEASSTAT - Seasonal statistical values . . . . . . . . . . . . . . . . . . . . . . . . . 124
2.8.28. SEASPCTL - Seasonal percentile values . . . . . . . . . . . . . . . . . . . . . . . . . 125
2.8.29. YHOURSTAT - Multi-year hourly statistical values . . . . . . . . . . . . . . . . . . 126
2.8.30. YDAYSTAT - Multi-year daily statistical values . . . . . . . . . . . . . . . . . . . . 128
2.8.31. YDAYPCTL - Multi-year daily percentile values . . . . . . . . . . . . . . . . . . . . 130
2.8.32. YMONSTAT - Multi-year monthly statistical values . . . . . . . . . . . . . . . . . . 131
2.8.33. YMONPCTL - Multi-year monthly percentile values . . . . . . . . . . . . . . . . . . 133
2.8.34. YSEASSTAT - Multi-year seasonal statistical values . . . . . . . . . . . . . . . . . . 134
2.8.35. YSEASPCTL - Multi-year seasonal percentile values . . . . . . . . . . . . . . . . . . 136
2.8.36. YDRUNSTAT - Multi-year daily running statistical values . . . . . . . . . . . . . . . 137
2.8.37. YDRUNPCTL - Multi-year daily running percentile values . . . . . . . . . . . . . . 139
2.9. Correlation and co. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
2.9.1. FLDCOR - Correlation in grid space . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
2.9.2. TIMCOR - Correlation over time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
2.9.3. FLDCOVAR - Covariance in grid space . . . . . . . . . . . . . . . . . . . . . . . . . 142
2.9.4. TIMCOVAR - Covariance over time . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
2.10. Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
2.10.1. REGRES - Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
2.10.2. DETREND - Detrend time series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
2.10.3. TREND - Trend of time series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
2.10.4. SUBTREND - Subtract a trend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
2.11. EOFs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
2.11.1. EOFS - Empirical Orthogonal Functions . . . . . . . . . . . . . . . . . . . . . . . . . 147
2.11.2. EOFCOEFF - Principal coefficients of EOFs . . . . . . . . . . . . . . . . . . . . . . 149
2.12. Interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
2.12.1. REMAPBIL - Bilinear interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
2.12.2. REMAPBIC - Bicubic interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
2.12.3. REMAPNN - Nearest neighbor remapping . . . . . . . . . . . . . . . . . . . . . . . . 153
2.12.4. REMAPDIS - Distance-weighted average remapping . . . . . . . . . . . . . . . . . . 154
2.12.5. REMAPCON - First order conservative remapping . . . . . . . . . . . . . . . . . . . 155
2.12.6. REMAPCON2 - Second order conservative remapping . . . . . . . . . . . . . . . . . 157
2.12.7. REMAPLAF - Largest area fraction remapping . . . . . . . . . . . . . . . . . . . . . 159
2.12.8. REMAP - Grid remapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
2.12.9. REMAPETA - Remap vertical hybrid level . . . . . . . . . . . . . . . . . . . . . . . 161
2.12.10.VERTINTML - Vertical interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . 163
2.12.11.VERTINTAP - Vertical interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
2.12.12.INTLEVEL - Linear level interpolation . . . . . . . . . . . . . . . . . . . . . . . . . 164
2.12.13.INTLEVEL3D - Linear level interpolation from/to 3d vertical coordinates . . . . . . 166
2.12.14.INTTIME - Time interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
2.12.15.INTYEAR - Year interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
2.13. Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
2.13.1. SPECTRAL - Spectral transformation . . . . . . . . . . . . . . . . . . . . . . . . . . 170
2.13.2. SPECCONV - Spectral conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
2.13.3. WIND2 - D and V to velocity potential and stream function . . . . . . . . . . . . . 171
2.13.4. WIND - Wind transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
2.13.5. FOURIER - Fourier transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
2.14. Import/Export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
2.14.1. IMPORTBINARY - Import binary data sets . . . . . . . . . . . . . . . . . . . . . . 175
2.14.2. IMPORTCMSAF - Import CM-SAF HDF5 files . . . . . . . . . . . . . . . . . . . . 176
2.14.3. IMPORTAMSR - Import AMSR binary files . . . . . . . . . . . . . . . . . . . . . . 177

4
Contents Contents

2.14.4. INPUT - Formatted input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

2.14.5. OUTPUT - Formatted output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
2.14.6. OUTPUTTAB - Table output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
2.14.7. OUTPUTGMT - GMT output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
2.15. Miscellaneous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
2.15.1. GRADSDES - GrADS data descriptor file . . . . . . . . . . . . . . . . . . . . . . . . 185
2.15.2. AFTERBURNER - ECHAM standard post processor . . . . . . . . . . . . . . . . . 186
2.15.3. FILTER - Time series filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
2.15.4. GRIDCELL - Grid cell quantities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
2.15.5. SMOOTH - Smooth grid points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
2.15.6. REPLACEVALUES - Replace variable values . . . . . . . . . . . . . . . . . . . . . . 191
2.15.7. TIMSORT - Timsort . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
2.15.8. VARGEN - Generate a field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
2.15.9. WINDTRANS - Wind Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . 194
2.15.10.ROTUVB - Rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
2.15.11.MASTRFU - Mass stream function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
2.15.12.DERIVEPAR - Sea level pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
2.15.13.ADISIT - Potential temperature to in-situ temperature and vice versa . . . . . . . . 196
2.15.14.RHOPOT - Calculates potential density . . . . . . . . . . . . . . . . . . . . . . . . . 196
2.15.15.HISTOGRAM - Histogram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
2.15.16.SETHALO - Set the left and right bounds of a field . . . . . . . . . . . . . . . . . . 197
2.15.17.WCT - Windchill temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
2.15.18.FDNS - Frost days where no snow index per time period . . . . . . . . . . . . . . . . 198
2.15.19.STRWIN - Strong wind days index per time period . . . . . . . . . . . . . . . . . . . 198
2.15.20.STRBRE - Strong breeze days index per time period . . . . . . . . . . . . . . . . . . 199
2.15.21.STRGAL - Strong gale days index per time period . . . . . . . . . . . . . . . . . . . 199
2.15.22.HURR - Hurricane days index per time period . . . . . . . . . . . . . . . . . . . . . 199
2.15.23.CMORLITE - CMOR lite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

3. Contributors 202
3.1. History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
3.2. External sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
3.3. Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

A. Environment Variables 206

B. Parallelized operators 207

C. Standard name table 208

D. Grid description examples 209

D.1. Example of a curvilinear grid description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
D.2. Example description for an unstructured grid . . . . . . . . . . . . . . . . . . . . . . . . . . 210

Operator index 211

5
1. Introduction
The Climate Data Operator (CDO) software is a collection of many operators for standard processing of
climate and forecast model data. The operators include simple statistical and arithmetic functions, data
selection and subsampling tools, and spatial interpolation. CDO was developed to have the same set of
processing functions for GRIB [GRIB] and NetCDF [NetCDF] datasets in one package.
The Climate Data Interface [CDI] is used for the fast and file format independent access to GRIB and
NetCDF datasets. The local MPI-MET data formats SERVICE, EXTRA and IEG are also supported.
There are some limitations for GRIB and NetCDF datasets. A GRIB dataset has to be consistent, similar
to NetCDF. That means all time steps need to have the same variables, and within a time step each variable
may occur only once. NetCDF datasets are only supported for the classic data model and arrays up to 4
dimensions. These dimensions should only be used by the horizontal and vertical grid and the time. The
NetCDF attributes should follow the GDT, COARDS or CF Conventions.
The user interface and some operators are similar to the PINGO [PINGO] package.
The main CDO features are:
• More than 700 operators available
• Modular design and easily extendable with new operators
• Very simple UNIX command line interface
• A dataset can be processed by several operators, without storing the interim results in files
• Most operators handle datasets with missing values
• Fast processing of large datasets
• Support of many different grid types
• Tested on many UNIX/Linux systems, Cygwin, and MacOS-X

1.1. Building from sources

This section describes how to build CDO from the sources on a UNIX system. CDO uses the GNU configure
and build system for compilation. The only requirement is a working ISO C++11 and ANSI C99 compiler.
First go to the download page (https://code.mpimet.mpg.de/projects/cdo) to get the latest distribu-
tion, if you do not have it yet.
To take full advantage of CDO features the following additional libraries should be installed:

• Unidata NetCDF library (https://www.unidata.ucar.edu/software/netcdf) version 3 or higher.

This library is needed to process NetCDF [NetCDF] files with CDO.
• ECMWF ecCodes library (https://software.ecmwf.int/wiki/display/ECC/ecCodes+Home) ver-
sion 2.3.0 or higher. This library is needed to process GRIB2 files with CDO.
• HDF5 szip library (http://www.hdfgroup.org/doc_resource/SZIP) version 2.1 or higher.
This library is needed to process szip compressed GRIB [GRIB] files with CDO.
• HDF5 library (http://www.hdfgroup.org/HDF5) version 1.6 or higher.
This library is needed to import CM-SAF [CM-SAF] HDF5 files with the CDO operator im-
port_cmsaf.

6
Introduction Usage

• PROJ.4 library (http://trac.osgeo.org/proj) version 4.6 or higher.

This library is needed to convert Sinusoidal and Lambert Azimuthal Equal Area coordinates to
geographic coordinates, for e.g. remapping.
• Magics library (https://software.ecmwf.int/wiki/display/MAGP/Magics) version 2.18 or higher.
This library is needed to create contour, vector and graph plots with CDO.
CDO is a multi-threaded application. Therefor all the above libraries should be compiled thread safe.
Using non-threadsafe libraries could cause unexpected errors!

1.1.1. Compilation

Compilation is done by performing the following steps:

1. Unpack the archive, if you haven’t done that yet:
gunzip cdo-$VERSION.tar.gz # uncompress the archive
tar xf cdo-$VERSION.tar # unpack it
cd cdo-$VERSION
2. Run the configure script:
./configure
• Optionaly with NetCDF [NetCDF] support:
./configure --with-netcdf=<NetCDF root directory>
• and with ecCodes:
./configure --with-eccodes=<ecCodes root directory>
For an overview of other configuration options use
./configure --help
3. Compile the program by running make:
make
The program should compile without problems and the binary (cdo) should be available in the src directory
of the distribution.

1.1.2. Installation

After the compilation of the source code do a make install, possibly as root if the destination permissions
require that.

make install

The binary is installed into the directory <prefix>/bin. <prefix> defaults to /usr/local but can be
changed with the –prefix option of the configure script.
Alternatively, you can also copy the binary from the src directory manually to some bin directory in your
search path.

1.2. Usage
This section descibes how to use CDO. The syntax is:

cdo [ Options ] Operator1 [ -Operator2 [ -OperatorN ] ]

7
Usage Introduction

1.2.1. Options

All options have to be placed before the first operator. The following options are available for all operators:

-a Generate an absolute time axis.

-b <nbits> Set the number of bits for the output precision. The valid precisions depend
on the file format:
<format> <nbits>
grb1, grb2 P1 - P24
nc1, nc2, nc4, nc4c I8/I16/I32/F32/F64
grb2, srv, ext, ieg F32/F64
For srv, ext and ieg format the letter L or B can be added to set the byteorder
to Little or Big endian.
- -cmor CMOR conform NetCDF output.
-C, - -color Colorized output messages.
- -eccodes Use ecCodes to decode/encode GRIB1 messages.
-f <format> Set the output file format. The valid file formats are:
File format <format>
GRIB version 1 grb1/grb
GRIB version 2 grb2
NetCDF nc1
NetCDF version 2 (64-bit offset) nc2/nc
NetCDF-4 (HDF5) nc4
NetCDF-4 classic nc4c
NetCDF version 5 (64-bit data) nc5
SERVICE srv
EXTRA ext
IEG ieg
GRIB2 is only available if CDO was compiled with ecCodes support and all
NetCDF file types are only available if CDO was compiled with NetCDF support!
-g <grid> Define the default grid description by name or from file (see chapter 1.3 on page 11).
Available grid names are: r<NX>x<NY>, lon=<LON>/lat=<LAT>, n<N>, gme<NI>
-h, - -help Help information for the operators.
- -no_history Do not append to NetCDF history global attribute.
- -netcdf_hdr_pad, - -hdr_pad, - -header_pad <nbr>
Pad NetCDF output header with nbr bytes.
-k <chunktype> NetCDF4 chunk type: auto, grid or lines.
-L Lock I/O (sequential access).
-M Switch to indicate that the I/O streams have missing values.
-m <missval> Set the missing value of non NetCDF files (default: -9e+33).
-O Overwrite existing output file, if checked.
Existing output file is checked only for: ens<STAT>, merge, mergetime
- -operators List of all operators.
-P <nthreads> Set number of OpenMP threads (Only available if OpenMP support was compiled in).
- -percentile <method>
Percentile method: nrank nist numpy numpy_lower numpy_higher numpy_nearest
- -reduce_dim Reduce NetCDF dimensions.
-R, - -regular Convert GRIB1 data from global reduced to regular Gaussian grid (only with cgribex lib).
-r Generate a relative time axis.
-S Create an extra output stream for the module TIMSTAT. This stream contains
the number of non missing values for each output period.
-s, - -silent Silent mode.
- -sortname Alphanumeric sorting of NetCDF parameter names.
-t <partab> Set the GRIB1 (cgribex) default parameter table name or file (see chapter 1.6 on page 16).
Predefined tables are: echam4 echam5 echam6 mpiom1 ecmwf remo

8
Introduction Usage

- -timestat_date <srcdate>
Target timestamp (temporal statistics): first, middle, midhigh or last source timestep.
-V, - -version Print the version number.
-v, - -verbose Print extra details for some operators.
-W Print extra warning messages.
-z szip SZIP compression of GRIB1 records.
jpeg JPEG compression of GRIB2 records.
zip[_1-9] Deflate compression of NetCDF4 variables.

1.2.2. Environment variables

There are some environment variables which influence the behavior of CDO. An incomplete list can be
found in Appendix A.
Here is an example to set the envrionment variable CDO_RESET_HISTORY for different shells:
Bourne shell (sh): CDO_RESET_HISTORY=1 ; export CDO_RESET_HISTORY
Korn shell (ksh): export CDO_RESET_HISTORY=1
C shell (csh): setenv CDO_RESET_HISTORY 1

1.2.3. Operators

There are more than 700 operators available. A detailed description of all operators can be found in the
Reference Manual section.

1.2.4. Operator chaining

All operators with a fixed number of input streams and one output stream can pipe the result directly to
an other operator. The operator must begin with "–", in order to combine it with others. This can improve
the performance by:
• reducing unnecessary disk I/O
• parallel processing
Use

cdo sub -dayavg infile2 -timavg infile1 outfile

instead of

cdo timavg infile1 tmp1

cdo dayavg infile2 tmp2
cdo sub tmp2 tmp1 outfile
rm tmp1 tmp2

All operators with one input stream will process only one input stream! You need to take care when mixing
those operators with an operator with an arbitrary number of input streams. The following examples
illustrate this problem.
1. cdo info -timavg infile?
2. cdo info -timavg infile1 infile2
3. cdo timavg infile1 tmpfile
cdo info tmpfile infile2

9
Usage Introduction

All three examples produce identical results. The time average will be computed only on the first input
file.
All operators with an arbitrary number of input streams (infiles) can’t be combined with other operators
if these operators are used with more than one input stream. Here is an incomplete list of these operators:
copy, cat, merge, mergetime, select, ens< ST AT >
Use single quotes if the input stream names matched to a single wildcard expression. In this case CDO
will do the pattern matching and the output can be combined with other operators. Here is an example
for this feature:

cdo timavg -select,name=temperature ’infile?’ outfile

The CDO internal wildcard expansion is using the glob() function. Therefore internal wildcard expansion
is not available on operating systems without the glob() function! The following wildcards are supported:
∗, ?, []
Note: Operator chaining is implemented over POSIX Threads (pthreads). Therefore this CDO feature is
not available on operating systems without POSIX Threads support!

1.2.5. Parallelized operators

Some of the CDO operators are shared memory parallelized with OpenMP. An OpenMP-enabled C compiler
is needed to use this feature. Users may request a specific number of OpenMP threads nthreads with the
’ -P’ switch.
Here is an example to distribute the bilinear interpolation on 8 OpenMP threads:
cdo -P 8 remapbil,targetgrid infile outfile

Many CDO operators are I/O-bound. This means most of the time is spend in reading and writing the
data. Only compute intensive CDO operators are parallelized. An incomplete list of OpenMP parallelized
operators can be found in Appendix B.

1.2.6. Operator parameter

Some operators need one or more parameter. A list of parameter is indicated by the seperator ’,’.
• STRING
Unquoted characters without blanks and tabs. The following command select variables with the
name pressure and tsurf:
cdo selvar,pressure,tsurf infile outfile
• FLOAT
Floating point number in any representation. The following command sets the range between 0 and
273.15 of all fields to missing value:
cdo setrtomiss,0,273.15 infile outfile
• BOOL
Boolean parameter in the following representation TRUE/FALSE, T/F or 0/1. To disable the weight-
ing by grid cell area in the calculation of a field mean, use:
cdo fldmean,weights=FALSE infile outfile
• INTEGER
A range of integer parameter can be specified by first/last[/inc]. To select the days 5, 6, 7, 8 and 9
use:
cdo selday,5/9 infile outfile
The result is the same as:
cdo selday,5,6,7,8,9 infile outfile

10
Introduction Horizontal grids

1.3. Horizontal grids

Physical quantities of climate models are typically stored on a horizonal grid. CDO supports structured
grids like regular lon/lat or curvilinear grids and also unstructured grids.

1.3.1. Grid area weights

One single point of a horizontal grid represents the mean of a grid cell. These grid cells are typically of
different sizes, because the grid points are of varying distance.
Area weights are individual weights for each grid cell. They are needed to compute the area weighted
mean or variance of a set of grid cells (e.g. fldmean - the mean value of all grid cells). In CDO the area
weights are derived from the grid cell area. If the cell area is not available then it will be computed from
the geographical coordinates via spherical triangles. This is only possible if the geographical coordinates of
the grid cell corners are available or derivable. Otherwise CDO gives a warning message and uses constant
area weights for all grid cells.
The cell area is read automatically from a NetCDF input file if a variable has the corresponding “cell_measures”
attribute, e.g.:

var : c e l l _ m e a s u r e s = " a r e a : c e l l _ a r e a " ;

If the computed cell area is not desired then the CDO operator setgridarea can be used to set or overwrite
the grid cell area.

1.3.2. Grid description

In the following situations it is necessary to give a description of a horizontal grid:

• Changing the grid description (operator: setgrid)
• Horizontal interpolation (all remapping operators)
• Generating of variables (operator: const, random)
As now described, there are several possibilities to define a horizontal grid.

1.3.2.1. Predefined grids

Predefined grids are available for global regular, gaussian or icosahedral-hexagonal GME grids.

Global regular grid: global_<DXY>

global_<DXY> defines a global regular lon/lat grid. The grid increment <DXY> can be selected at will. The
longitudes start at <DXY>/2 - 180◦ and the latitudes start at <DXY>/2 - 90◦ .

Global regular grid: r<NX>x<NY>

r<NX>x<NY> defines a global regular lon/lat grid. The number of the longitudes <NX> and the latitudes
<NY> can be selected at will. The longitudes start at 0◦ with an increment of (360/<NX>)◦ . The latitudes
go from south to north with an increment of (180/<NY>)◦ .

One grid point: lon=<LON>/lat=<LAT>

lon=<LON>/lat=<LAT> defines a lon/lat grid with only one grid point.

11
Horizontal grids Introduction

Global Gaussian grid: n<N>

n<N> defines a global Gaussian grid. N specifies the number of latitudes lines between the Pole and the
Equator. The longitudes start at 0◦ with an increment of (360/nlon)◦ . The gaussian latitudes go from
north to south.

Global icosahedral-hexagonal GME grid: gme<NI>

gme<NI> defines a global icosahedral-hexagonal GME grid. NI specifies the number of intervals on a main
triangle side.

1.3.2.2. Grids from data files

You can use the grid description from an other datafile. The format of the datafile and the grid of the data
field must be supported by CDO . Use the operator ’sinfo’ to get short informations about your variables
and the grids. If there are more then one grid in the datafile the grid description of the first variable will
be used. Add the extension :N to the name of the datafile to select grid number N.

1.3.2.3. SCRIP grids

SCRIP (Spherical Coordinate Remapping and Interpolation Package) uses a common grid description for
curvilinear and unstructured grids. For more information about the convention see [SCRIP]. This grid
description is stored in NetCDF. Therefor it is only available if CDO was compiled with NetCDF support!

SCRIP grid description example of a curvilinear MPIOM [MPIOM] GROB3 grid (only the NetCDF header):
netcdf grob3s {
dimensions :
g r i d _ s i z e = 12120 ;
g r i d _ x s i z e = 120 ;
g r i d _ y s i z e = 101 ;
grid_corners = 4 ;
grid_rank = 2 ;
variables :
i n t grid_dims ( g r i d _ r a n k ) ;
f l o a t grid_center_lat ( grid_ysize , grid_xsize ) ;
grid_center_lat : units = " degrees " ;
g r i d _ c e n t e r _ l a t : bounds = " g r i d _ c o r n e r _ l a t " ;
f l o a t grid_center_lon ( grid_ysize , grid_xsize ) ;
grid_center_lon : units = " degrees " ;
g r i d _ c e n t e r _ l o n : bounds = " g r i d _ c o r n e r _ l o n " ;
i n t grid_imask ( g r i d _ y s i z e , g r i d _ x s i z e ) ;
grid_imask : u n i t s = " u n i t l e s s " ;
grid_imask : c o o r d i n a t e s = " g r i d _ c e n t e r _ l o n g r i d _ c e n t e r _ l a t " ;
f l o a t grid_corner_lat ( grid_ysize , grid_xsize , grid_corners ) ;
grid_corner_lat : units = " degrees " ;
f l o a t grid_corner_lon ( grid_ysize , grid_xsize , grid_corners ) ;
grid_corner_lon : units = " degrees " ;

// g l o b a l a t t r i b u t e s :
: t i t l e = " grob3s " ;
}

1.3.2.4. CDO grids

All supported grids can also be described with the CDO grid description. The following keywords can be
used to describe a grid:

12
Introduction Horizontal grids

Keyword Datatype Description

gridtype STRING Type of the grid (gaussian, lonlat, curvilinear, unstructured).
gridsize INTEGER Size of the grid.
xsize INTEGER Size in x direction (number of longitudes).
ysize INTEGER Size in y direction (number of latitudes).
xvals FLOAT ARRAY X values of the grid cell center.
yvals FLOAT ARRAY Y values of the grid cell center.
nvertex INTEGER Number of the vertices for all grid cells.
xbounds FLOAT ARRAY X bounds of each gridbox.
ybounds FLOAT ARRAY Y bounds of each gridbox.
xfirst, xinc FLOAT, FLOAT Macros to define xvals with a constant increment,
xfirst is the x value of the first grid cell center.
yfirst, yinc FLOAT, FLOAT Macros to define yvals with a constant increment,
yfirst is the y value of the first grid cell center.
xunits STRING units of the x axis
yunits STRING units of the y axis

Which keywords are necessary depends on the gridtype. The following table gives an overview of the
default values or the size with respect to the different grid types.

gridtype lonlat gaussian projection curvilinear unstructured

gridsize xsize*ysize xsize*ysize xsize*ysize xsize*ysize ncell
xsize nlon nlon nx nlon gridsize
ysize nlat nlat ny nlat gridsize
xvals xsize xsize xsize gridsize gridsize
yvals ysize ysize ysize gridsize gridsize
nvertex 2 2 2 4 nv
xbounds 2*xsize 2*xsize 2*xsize 4*gridsize nv*gridsize
ybounds 2*ysize 2*ysize 2*xsize 4*gridsize nv*gridsize
xunits degrees degrees m degrees degrees
yunits degrees degrees m degrees degrees

The keywords nvertex, xbounds and ybounds are optional if area weights are not needed. The grid cell
corners xbounds and ybounds have to rotate counterclockwise.

CDO grid description example of a T21 gaussian grid:

gridtype = gaussian
xsize = 64
ysize = 32
xfirst = 0
xinc = 5.625
yvals = 85.76 80.27 74.75 69.21 63.68 58.14 52.61 47.07
41.53 36.00 30.46 24.92 19.38 13.84 8.31 2.77
−2.77 −8.31 −13.84 −19.38 −24.92 −30.46 −36.00 −41.53
−47.07 −52.61 −58.14 −63.68 −69.21 −74.75 −80.27 −85.76

CDO grid description example of a global regular grid with 60x30 points:
gridtype = lonlat
xsize = 60
ysize = 30
xfirst = −177
xinc = 6
yfirst = −87
yinc = 6

13
Z-axis description Introduction

The description for a projection is somewhat more complicated. Use the first section to describe the
coordinates of the projection with the above keywords. Add the keyword grid_mapping_name to
descibe the mapping between the given coordinates and the true latitude and longitude coordinates.
grid_mapping_name takes a string value that contains the name of the projection. A list of attributes
can be added to define the mapping. The name of the attributes depend on the projection. The valid
names of the projection and there attributes follow the NetCDF CF-Convention.
CDO supports the special grid mapping attribute proj4_params. These parameter will be passed directly
to the proj4 library to generate the geographic coordinates if needed.
The geographic coordinates of the following projections can be generated without the attribute proj4_params,
if all other attributes are available:

• rotated_latitude_longitude
• lambert_conformal_conic
• lambert_azimuthal_equal_area
• sinusoidal
• polar_stereographic

CDO grid description example of a regional rotated lon/lat grid:

gridtype = projection
xsize = 81
ysize = 91
xunits = " degrees "
yunits = " degrees "
xfirst = −19.5
xinc = 0.5
yfirst = −25.0
yinc = 0.5
grid_mapping_name = r o t a t e d _ l a t i t u d e _ l o n g i t u d e
g r i d _ n o r t h _ p o l e _ l o n g i t u d e = −170
grid_north_pole_latitude = 32.5

Example CDO descriptions of a curvilinear and an unstructured grid can be found in Appendix D.

1.4. Z-axis description

Sometimes it is necessary to change the description of a z-axis. This can be done with the operator setzaxis.
This operator needs an ASCII formatted file with the description of the z-axis. The following keywords
can be used to describe a z-axis:

Keyword Datatype Description

zaxistype STRING type of the z-axis
size INTEGER number of levels
levels FLOAT ARRAY values of the levels
lbounds FLOAT ARRAY lower level bounds
ubounds FLOAT ARRAY upper level bounds
vctsize INTEGER number of vertical coordinate parameters
vct FLOAT ARRAY vertical coordinate table

The keywords lbounds and ubounds are optional. vctsize and vct are only necessary to define hybrid
model levels.
Available z-axis types:

14
Introduction Time axis

Z-axis type Description Units

surface Surface
pressure Pressure level pascal
hybrid Hybrid model level
height Height above ground meter
depth_below_sea Depth below sea level meter
depth_below_land Depth below land surface centimeter
isentropic Isentropic (theta) level kelvin

Z-axis description example for pressure levels 100, 200, 500, 850 and 1000 hPa:
zaxistype = pressure
size = 5
levels = 10000 20000 50000 85000 100000

Z-axis description example for ECHAM5 L19 hybrid model levels:

zaxistype = hybrid
size = 19
levels = 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
vctsize = 40
vct = 0 2000 4000 6 0 4 6 . 1 0 9 3 8 8 2 6 7 . 9 2 5 7 8 1 0 6 0 9 . 5 1 1 7 1 2 8 5 1 . 1 0 1 6 1 4 6 9 8 . 5
15861.125 16116.2383 15356.9258 13621.4609 11101.5625 8127.14453
5125.14062 2549.96875 783.195068 0 0 0
0 0 0 0.000338993268 0.00335718691 0.0130700432 0.0340771675
0.0706498027 0.12591666 0.201195419 0.295519829 0.405408859
0.524931908 0.646107674 0.759697914 0.856437683 0.928747177
0.972985268 0.992281914 1

Note that the vctsize is twice the number of levels plus two and the vertical coordinate table must be
specified for the level interfaces.

1.5. Time axis

A time axis describes the time for every timestep. Two time axis types are available: absolute time and
relative time axis. CDO tries to maintain the actual type of the time axis for all operators.

1.5.1. Absolute time

An absolute time axis has the current time to each time step. It can be used without knowledge of the
calendar. This is preferably used by climate models. In NetCDF files the absolute time axis is represented
by the unit of the time: "day as %Y%m%d.%f".

1.5.2. Relative time

A relative time is the time relative to a fixed reference time. The current time results from the reference time
and the elapsed interval. The result depends on the calendar used. CDO supports the standard Gregorian,
proleptic Gregorian, 360 days, 365 days and 366 days calendars. The relative time axis is preferably used
by numerical weather prediction models. In NetCDF files the relative time axis is represented by the unit
of the time: "time-units since reference-time", e.g "days since 1989-6-15 12:00".

1.5.3. Conversion of the time

Some programs which work with NetCDF data can only process relative time axes. Therefore it may be
necessary to convert from an absolute into a relative time axis. This conversion can be done for each
operator with the CDO option ’-r’. To convert a relative into an absolute time axis use the CDO option
’-a’.

15
Parameter table Introduction

1.6. Parameter table

A parameter table is an ASCII formated file to convert code numbers to variable names. Each variable
has one line with its code number, name and a description with optional units in a blank separated list. It
can only be used for GRIB, SERVICE, EXTRA and IEG formated files. The CDO option ’-t <partab>’
sets the default parameter table for all input files. Use the operator ’setpartab’ to set the parameter table
for a specific file.

Example of a CDO parameter table:

134 aps s u r f a c e p r e s s u r e [ Pa ]
141 sn snow depth [m]
147 ahfl l a t e n t h e a t f l u x [W/m∗ ∗ 2 ]
172 slm l a n d s e a mask
175 albedo s u r f a c e albedo
211 siced i c e depth [m]

1.7. Missing values

Most operators can handle missing values. The default missing value for GRIB, SERVICE, EXTRA and
IEG files is −9.e33 . The CDO option ’-m <missval>’ overwrites the default missing value. In NetCDF
files the variable attribute ’_FillValue’ is used as a missing value. The operator ’setmissval’ can be used
to set a new missing value.
The CDO use of the missing value is shown in the following tables, where one table is printed for each
operation. The operations are applied to arbitrary numbers a, b, the special case 0, and the missing value
miss. For example the table named "addition" shows that the sum of an arbitrary number a and the
missing value is the missing value, and the table named "multiplication" shows that 0 multiplied by missing
value results in 0.

addition b miss
a a+b miss
miss miss miss
subtraction b miss
a a−b miss
miss miss miss
multiplication b 0 miss
a a∗b 0 miss
0 0 0 0
miss miss 0 miss
division b 0 miss
a a/b miss miss
0 0 miss miss
miss miss miss miss
maximum b miss
a max(a, b) a
miss b miss
minimum b miss
a min(a, b) a
miss b miss
sum b miss
a a+b a
miss b miss

16
Introduction Percentile

The handling of missing values by the operations "minimum" and "maximum" may be surprising, but the
definition given here is more consistent with that expected in practice. Mathematical functions (e.g. log,
sqrt, etc.) return the missing value if an argument is the missing value or an argument is out of range.
All statistical functions ignore missing values, treading them as not belonging to the sample, with the
side-effect of a reduced sample size.

1.7.1. Mean and average

An artificial distinction is made between the notions mean and average. The mean is regarded as a
statistical function, whereas the average is found simply by adding the sample members and dividing the
result by the sample size. For example, the mean of 1, 2, miss and 3 is (1 + 2 + 3)/3 = 2, whereas the
average is (1 + 2 + miss + 3)/4 = miss/4 = miss. If there are no missing values in the sample, the average
and mean are identical.

1.8. Percentile
There is no standard definition of percentile. All definitions yield to similar results when the number of
values is very large. The following percentile methods are available in CDO:

Percentile
Description
method
nrank Nearest Rank method, the default method used in CDO
nist The primary method recommended by NIST
numpy numpy.percentile with the option interpolation set to ’linear’
numpy_lower numpy.percentile with the option interpolation set to ’lower’
numpy_higher numpy.percentile with the option interpolation set to ’higher’
numpy_nearest numpy.percentile with the option interpolation set to ’nearest’

The percentile method can be selected with the CDO option - -percentile. The Nearest Rank method
is the default percentile method in CDO.
The different percentile methods can lead to different results, especially for small number of data values.
Consider the ordered list {15, 20, 35, 40, 50, 55}, which contains six data values. Here is the result for the
30th, 40th, 50th, 75th and 100th percentiles of this list using the different percentile methods:

Percentile numpy numpy numpy

nrank nist numpy
P lower higher nearest
30th 20 21.5 27.5 20 35 35
40th 35 32 35 35 35 35
50th 35 37.5 37.5 35 40 40
75th 50 51.25 47.5 40 50 50
100th 55 55 55 55 55 55

1.8.1. Percentile over timesteps

The amount of data for time series can be very large. All data values need to held in memory to calculate
the percentile. The percentile over timesteps uses a histogram algorithm, to limit the amount of required
memory. The default number of histogram bins is 101. That means the histogram algorithm is used,
when the dataset has more than 101 time steps. The default can be overridden by setting the environment
variable CDO_PCTL_NBINS to a different value. The histogram algorithm is implemented only for the Nearest
Rank method.

17
2. Reference manual
This section gives a description of all operators. Related operators are grouped to modules. For easier
description all single input files are named infile or infile1, infile2, etc., and an arbitrary number of
input files are named infiles. All output files are named outfile or outfile1, outfile2, etc. Further
the following notion is introduced:
i(t) Timestep t of infile
i(t, x) Element number x of the field at timestep t of infile
o(t) Timestep t of outfile
o(t, x) Element number x of the field at timestep t of outfile

18
Reference manual Information

2.1. Information
This section contains modules to print information about datasets. All operators print there results to
standard output.
Here is a short overview of all operators in this section:

info Dataset information listed by parameter identifier

infon Dataset information listed by parameter name
map Dataset information and simple map

sinfo Short information listed by parameter identifier

sinfon Short information listed by parameter name

diff Compare two datasets listed by parameter id

diffn Compare two datasets listed by parameter name

npar Number of parameters

nlevel Number of levels
nyear Number of years
nmon Number of months
ndate Number of dates
ntime Number of timesteps
ngridpoints Number of gridpoints
ngrids Number of horizontal grids

showformat Show file format

showcode Show code numbers
showname Show variable names
showstdname Show standard names
showatts Show all attributes
showattsglob Show all global attributes
showlevel Show levels
showltype Show GRIB level types
showyear Show years
showmon Show months
showdate Show date information
showtime Show time information
showtimestamp Show timestamp

showattribute Show a global attribute or a variable attribute

showattsvar Show all variable attributes.

partab Parameter table

codetab Parameter code table
griddes Grid description
zaxisdes Z-axis description
vct Vertical coordinate table

19
Information Reference manual

2.1.1. INFO - Information and simple statistics

Synopsis

< operator > infiles

Description
This module writes information about the structure and contents of all input files to standard output.
All input files need to have the same structure with the same variables on different timesteps. The
information displayed depends on the chosen operator.

Operators

info Dataset information listed by parameter identifier

Prints information and simple statistics for each field of all input datasets. For each field
the operator prints one line with the following elements:
• Date and Time
• Level, Gridsize and number of Missing values
• Minimum, Mean and Maximum
The mean value is computed without the use of area weights!
• Parameter identifier
infon Dataset information listed by parameter name
The same as operator info but using the name instead of the identifier to label the param-
eter.
map Dataset information and simple map
Prints information, simple statistics and a map for each field of all input datasets. The
map will be printed only for fields on a regular lon/lat grid.

Example
To print information and simple statistics for each field of a dataset use:
cdo infon infile

This is an example result of a dataset with one 2D parameter over 12 timesteps:

−1 : Date Time L e v e l Size Miss : Minimum Mean Maximum : Name
1 : 1987−01−31 12:00:00 0 2048 1361 : 232.77 2 6 6 . 6 5 3 0 5 . 3 1 : SST
2 : 1987−02−28 12:00:00 0 2048 1361 : 233.64 2 6 7 . 1 1 3 0 7 . 1 5 : SST
3 : 1987−03−31 12:00:00 0 2048 1361 : 225.31 2 6 7 . 5 2 3 0 7 . 6 7 : SST
4 : 1987−04−30 12:00:00 0 2048 1361 : 215.68 2 6 8 . 6 5 3 1 0 . 4 7 : SST
5 : 1987−05−31 12:00:00 0 2048 1361 : 215.78 2 7 1 . 5 3 3 1 2 . 4 9 : SST
6 : 1987−06−30 12:00:00 0 2048 1361 : 212.89 2 7 2 . 8 0 3 1 4 . 1 8 : SST
7 : 1987−07−31 12:00:00 0 2048 1361 : 209.52 2 7 4 . 2 9 3 1 6 . 3 4 : SST
8 : 1987−08−31 12:00:00 0 2048 1361 : 210.48 2 7 4 . 4 1 3 1 5 . 8 3 : SST
9 : 1987−09−30 12:00:00 0 2048 1361 : 210.48 2 7 2 . 3 7 3 1 2 . 8 6 : SST
10 : 1987−10−31 12:00:00 0 2048 1361 : 219.46 2 7 0 . 5 3 3 0 9 . 5 1 : SST
11 : 1987−11−30 12:00:00 0 2048 1361 : 230.98 2 6 9 . 8 5 3 0 8 . 6 1 : SST
12 : 1987−12−31 12:00:00 0 2048 1361 : 241.25 2 6 9 . 9 4 3 0 9 . 2 7 : SST

20
Reference manual Information

2.1.2. SINFO - Short information

Synopsis
< operator > infiles

Description
This module writes information about the structure of infiles to standard output. infiles is an
arbitrary number of input files. All input files need to have the same structure with the same variables
on different timesteps. The information displayed depends on the chosen operator.

Operators
sinfo Short information listed by parameter identifier
Prints short information of a dataset. The information is divided into 4 sections. Section
1 prints one line per parameter with the following information:
• institute and source
• time c=constant v=varying
• type of statistical processing
• number of levels and z-axis number
• horizontal grid size and number
• data type
• parameter identifier
Section 2 and 3 gives a short overview of all grid and vertical coordinates. And the last
section contains short information of the time coordinate.
sinfon Short information listed by parameter name
The same as operator sinfo but using the name instead of the identifier to label the
parameter.

Example
To print short information of a dataset use:
cdo sinfon infile

This is the result of an ECHAM5 dataset with 3 parameter over 12 timesteps:

−1 : I n s t i t u t S o u r c e T S t e p t y p e L e v e l s Num P o i n t s Num Dtype : Name
1 : MPIMET ECHAM5 c i n s t a n t 1 1 2048 1 F32 : GEOSP
2 : MPIMET ECHAM5 v i n s t a n t 4 2 2048 1 F32 : T
3 : MPIMET ECHAM5 v i n s t a n t 1 1 2048 1 F32 : TSURF
Grid coordinates :
1 : gaussian : p o i n t s =2048 ( 6 4 x32 ) np=16
l o n g i t u d e : 0 t o 3 5 4 . 3 7 5 by 5 . 6 2 5 d e g r e e s _ e a s t circular
l a t i t u d e : 8 5 . 7 6 0 6 t o −85.7606 d e g r e e s _ n o r t h
Vertical coordinates :
1 : surface : l e v e l s =1
2 : pressure : l e v e l s =4
l e v e l : 92500 t o 20000 Pa
Time c o o r d i n a t e : 12 s t e p s
YYYY−MM−DD hh :mm: s s YYYY−MM−DD hh :mm: s s YYYY−MM−DD hh :mm: s s YYYY−MM−DD hh :mm: s s
1987−01−31 1 2 : 0 0 : 0 0 1987−02−28 1 2 : 0 0 : 0 0 1987−03−31 1 2 : 0 0 : 0 0 1987−04−30 1 2 : 0 0 : 0 0
1987−05−31 1 2 : 0 0 : 0 0 1987−06−30 1 2 : 0 0 : 0 0 1987−07−31 1 2 : 0 0 : 0 0 1987−08−31 1 2 : 0 0 : 0 0
1987−09−30 1 2 : 0 0 : 0 0 1987−10−31 1 2 : 0 0 : 0 0 1987−11−30 1 2 : 0 0 : 0 0 1987−12−31 1 2 : 0 0 : 0 0

21
Information Reference manual

2.1.3. DIFF - Compare two datasets field by field

Synopsis

< operator >[,options] infile1 infile2

Description
Compares the contents of two datasets field by field. The input datasets need to have the same
structure and its fields need to have the same header information and dimensions. Try the option
names if the number of variables differ. Exit status is 0 if inputs are the same and 1 if they differ.

Operators

diff Compare two datasets listed by parameter id

Provides statistics on differences between two datasets. For each pair of fields the operator
prints one line with the following information:
• Date and Time
• Level, Gridsize and number of Missing values
• Number of different values
• Occurrence of coefficient pairs with different signs (S)
• Occurrence of zero values (Z)
• Maxima of absolute difference of coefficient pairs
• Maxima of relative difference of non-zero coefficient pairs with equal signs
• Parameter identifier

Absdif f (t, x) = |i1 (t, x) − i2 (t, x)|

|i1 (t, x) − i2 (t, x)|

Reldif f (t, x) =
max(|i1 (t, x)| , |i2 (t, x)|)

diffn Compare two datasets listed by parameter name

The same as operator diff. Using the name instead of the identifier to label the parameter.

Parameter
abslim FLOAT Limit of the maximum absolute difference (default: 0)
rellim FLOAT Limit of the maximum relative difference (default: 1)
names STRING Consideration of the variable names of only one input file (left/right) or the
intersection of both (intersect).

Example
To print the difference for each field of two datasets use:
cdo diffn infile1 infile2

This is an example result of two datasets with one 2D parameter over 12 timesteps:

22
Reference manual Information

Date Time L e v e l Size Miss D i f f : S Z Max_Absdiff Max_Reldiff : Name

1 : 1987−01−31 1 2 : 0 0 : 0 0 0 2048 1361 273 : F F 0 . 0 0 0 1 0 6 8 1 4 . 1 6 6 0 e −07 : SST
2 : 1987−02−28 1 2 : 0 0 : 0 0 0 2048 1361 309 : F F 6 . 1 0 3 5 e −05 2 . 3 7 4 2 e −07 : SST
3 : 1987−03−31 1 2 : 0 0 : 0 0 0 2048 1361 292 : F F 7 . 6 2 9 4 e −05 3 . 3 7 8 4 e −07 : SST
4 : 1987−04−30 1 2 : 0 0 : 0 0 0 2048 1361 183 : F F 7 . 6 2 9 4 e −05 3 . 5 1 1 7 e −07 : SST
5 : 1987−05−31 1 2 : 0 0 : 0 0 0 2048 1361 207 : F F 0 . 0 0 0 1 0 6 8 1 4 . 0 3 0 7 e −07 : SST
7 : 1987−07−31 1 2 : 0 0 : 0 0 0 2048 1361 317 : F F 9 . 1 5 5 3 e −05 3 . 5 6 3 4 e −07 : SST
8 : 1987−08−31 1 2 : 0 0 : 0 0 0 2048 1361 219 : F F 7 . 6 2 9 4 e −05 2 . 8 8 4 9 e −07 : SST
9 : 1987−09−30 1 2 : 0 0 : 0 0 0 2048 1361 188 : F F 7 . 6 2 9 4 e −05 3 . 6 1 6 8 e −07 : SST
10 : 1987−10−31 1 2 : 0 0 : 0 0 0 2048 1361 297 : F F 9 . 1 5 5 3 e −05 3 . 5 0 0 1 e −07 : SST
11 : 1987−11−30 1 2 : 0 0 : 0 0 0 2048 1361 234 : F F 6 . 1 0 3 5 e −05 2 . 3 8 3 9 e −07 : SST
12 : 1987−12−31 1 2 : 0 0 : 0 0 0 2048 1361 267 : F F 9 . 3 5 5 3 e −05 3 . 7 6 2 4 e −07 : SST
11 o f 12 r e c o r d s d i f f e r

2.1.4. NINFO - Print the number of parameters, levels or times

Synopsis

< operator > infile

Description
This module prints the number of variables, levels or times of the input dataset.

Operators

npar Number of parameters

Prints the number of parameters (variables).
nlevel Number of levels
Prints the number of levels for each variable.
nyear Number of years
Prints the number of different years.
nmon Number of months
Prints the number of different combinations of years and months.
ndate Number of dates
Prints the number of different dates.
ntime Number of timesteps
Prints the number of timesteps.
ngridpoints Number of gridpoints
Prints the number of gridpoints for each variable.
ngrids Number of horizontal grids
Prints the number of horizontal grids.

Example
To print the number of parameters (variables) in a dataset use:
cdo npar infile

To print the number of months in a dataset use:

cdo nmon infile

23
Information Reference manual

2.1.5. SHOWINFO - Show variables, levels or times

Synopsis

< operator > infile

Description
This module prints the format, variables, levels or times of the input dataset.

Operators

showformat Show file format

Prints the file format of the input dataset.
showcode Show code numbers
Prints the code number of all variables.
showname Show variable names
Prints the name of all variables.
showstdname Show standard names
Prints the standard name of all variables.
showatts Show all attributes
Prints all variable and global attributes.
showattsglob Show all global attributes
Prints all global attributes.
showlevel Show levels
Prints all levels for each variable.
showltype Show GRIB level types
Prints the GRIB level type for all z-axes.
showyear Show years
Prints all years.
showmon Show months
Prints all months.
showdate Show date information
Prints date information of all timesteps (format YYYY-MM-DD).
showtime Show time information
Prints time information of all timesteps (format hh:mm:ss).
showtimestamp Show timestamp
Prints timestamp of all timesteps (format YYYY-MM-DDThh:mm:ss).

Example
To print the code number of all variables in a dataset use:
cdo showcode infile

This is an example result of a dataset with three variables:

129 130 139

To print all months in a dataset use:

24
Reference manual Information

cdo showmon infile

This is an examples result of a dataset with an annual cycle:

1 2 3 4 5 6 7 8 9 10 11 12

2.1.6. SHOWATTRIBUTE - Show attributes

Synopsis

showattribute,attribute infile
showattsvar[,var_nm] infile

Description
This operator prints attributes of a dataset. If a global attribute should be printed, the attribute
name can be specified as a parameter directly. If a variable attribute should be printed, the following
format is requested:
var_nm@att_nm

var_nm Variable name. Example: pressure

att_nm Attribute name. Example: units

Operators

showattribute Show a global attribute or a variable attribute

showattsvar Show all variable attributes.

If var_nm is specified, only for a subset of variables.

Parameter
attribute STRING Attribute in the format [var_nm@]att_nm
var_nm STRING Variable name

25
Information Reference manual

2.1.7. FILEDES - Dataset description

Synopsis

< operator > infile

Description
This module provides operators to print meta information about a dataset. The printed meta-data
depends on the chosen operator.

Operators

partab Parameter table

Prints all available meta information of the variables.
codetab Parameter code table
Prints a code table with a description of all variables. For each variable the operator
prints one line listing the code, name, description and units.
griddes Grid description
Prints the description of all grids.
zaxisdes Z-axis description
Prints the description of all z-axes.
vct Vertical coordinate table
Prints the vertical coordinate table.

Example
Assume all variables of the dataset are on a Gausssian N16 grid. To print the grid description of this
dataset use:
cdo griddes infile

Result:
gridtype : gaussian
gridsize : 2048
xname : lon
xlongname : longitude
xunits : degrees_east
yname : lat
ylongname : latitude
yunits : degrees_north
xsize : 64
ysize : 32
xfirst : 0
xinc : 5.625
yvals : 85.76058 80.26877 74.74454 69.21297 63.67863 58.1429 52.6065
47.06964 41.53246 35.99507 30.4575 24.91992 19.38223 13.84448
8 . 3 0 6 7 0 2 2 . 7 6 8 9 0 3 −2.768903 −8.306702 −13.84448 −19.38223
−24.91992 −30.4575 −35.99507 −41.53246 −47.06964 −52.6065
−58.1429 −63.67863 −69.21297 −74.74454 −80.26877 −85.76058

26
Reference manual File operations

2.2. File operations

This section contains modules to perform operations on files.
Here is a short overview of all operators in this section:

copy Copy datasets

cat Concatenate datasets

tee Duplicate a data stream

replace Replace variables

duplicate Duplicates a dataset

mergegrid Merge grid

merge Merge datasets with different fields

mergetime Merge datasets sorted by date and time

splitcode Split code numbers

splitparam Split parameter identifiers
splitname Split variable names
splitlevel Split levels
splitgrid Split grids
splitzaxis Split z-axes
splittabnum Split parameter table numbers

splithour Split hours

splitday Split days
splitseas Split seasons
splityear Split years
splityearmon Split in years and months
splitmon Split months

splitsel Split time selection

distgrid Distribute horizontal grid

collgrid Collect horizontal grid

27
File operations Reference manual

2.2.1. COPY - Copy datasets

Synopsis

< operator > infiles outfile

Description
This module contains operators to copy or concatenate datasets. infiles is an arbitrary number
of input files. All input files need to have the same structure with the same variables on different
timesteps.

Operators

copy Copy datasets

Copies all input datasets to outfile.
cat Concatenate datasets
Concatenates all input datasets and appends the result to the end of outfile. If outfile
does not exist it will be created.

Example
To change the format of a dataset to NetCDF use:
cdo -f nc copy infile outfile.nc

Add the option ’-r’ to create a relative time axis, as is required for proper recognition by GrADS or
Ferret:
cdo -r -f nc copy infile outfile.nc

To concatenate 3 datasets with different timesteps of the same variables use:

cdo copy infile1 infile2 infile3 outfile

If the output dataset already exists and you wish to extend it with more timesteps use:
cdo cat infile1 infile2 infile3 outfile

2.2.2. TEE - Duplicate a data stream

Synopsis

tee infile outfile1 outfile2

Description
This operator copies the input datasets to outfile1 and outfile2. It can be used to store interme-
diate results to a file.

Example
To compute the daily and monthy average of a dataset use:
cdo monavg -tee dayavg infile outfile_dayavg outfile_monavg

28
Reference manual File operations

2.2.3. REPLACE - Replace variables

Synopsis

replace infile1 infile2 outfile

Description
The replace operator replaces variables in infile1 by variables from infile2 and write the result to
outfile. Both input datasets need to have the same number of timesteps.

Example
Assume the first input dataset infile1 has three variables with the names geosp, t and tslm1 and the
second input dataset infile2 has only the variable tslm1. To replace the variable tslm1 in infile1
by tslm1 from infile2 use:
cdo replace infile1 infile2 outfile

2.2.4. DUPLICATE - Duplicates a dataset

Synopsis

duplicate[,ndup] infile outfile

Description
This operator duplicates the contents of infile and writes the result to outfile. The optional
parameter sets the number of duplicates, the default is 2.

Parameter
ndup INTEGER Number of duplicates, default is 2.

2.2.5. MERGEGRID - Merge grid

Synopsis

mergegrid infile1 infile2 outfile

Description
Merges grid points of all variables from infile2 to infile1 and write the result to outfile. Only
the non missing values of infile2 will be used. The horizontal grid of infile2 should be smaller
or equal to the grid of infile1 and the resolution must be the same. Only rectilinear grids are
supported. Both input files need to have the same variables and the same number of timesteps.

29
File operations Reference manual

2.2.6. MERGE - Merge datasets

Synopsis

< operator > infiles outfile

Description
This module reads datasets from several input files, merges them and writes the resulting dataset to
outfile.

Operators

merge Merge datasets with different fields

Merges time series of different fields from several input datasets. The number of fields
per timestep written to outfile is the sum of the field numbers per timestep in all
input datasets. The time series on all input datasets are required to have different
fields and the same number of timesteps. The fields in each different input file either
have to be different variables or different levels of the same variable. A mixture of
different variables on different levels in different input files is not allowed.
mergetime Merge datasets sorted by date and time
Merges all timesteps of all input files sorted by date and time. All input files need
to have the same structure with the same variables on different timesteps. After this
operation every input timestep is in outfile and all timesteps are sorted by date
and time.

Environment
SKIP_SAME_TIME If set to 1, skips all consecutive timesteps with a double entry of the same
timestamp.

Note
The operators in this module need to open all input files simultaneously. The maximum number of
open files depends on the operating system!

Example
Assume three datasets with the same number of timesteps and different variables in each dataset. To
merge these datasets to a new dataset use:
cdo merge infile1 infile2 infile3 outfile

Assume you split a 6 hourly dataset with splithour. This produces four datasets, one for each hour.
The following command merges them together:
cdo mergetime infile1 infile2 infile3 infile4 outfile

30
Reference manual File operations

2.2.7. SPLIT - Split a dataset

Synopsis

< operator >[,params] infile obase

Description
This module splits infile into pieces. The output files will be named <obase><xxx><suffix>
where suffix is the filename extension derived from the file format. xxx and the contents of the output
files depends on the chosen operator. params is a comma separated list of processing parameters.

Operators

splitcode Split code numbers

Splits a dataset into pieces, one for each different code number. xxx will have three
digits with the code number.
splitparam Split parameter identifiers
Splits a dataset into pieces, one for each different parameter identifier. xxx will be
a string with the parameter identifier.
splitname Split variable names
Splits a dataset into pieces, one for each variable name. xxx will be a string with
the variable name.
splitlevel Split levels
Splits a dataset into pieces, one for each different level. xxx will have six digits
with the level.
splitgrid Split grids
Splits a dataset into pieces, one for each different grid. xxx will have two digits
with the grid number.
splitzaxis Split z-axes
Splits a dataset into pieces, one for each different z-axis. xxx will have two digits
with the z-axis number.
splittabnum Split parameter table numbers
Splits a dataset into pieces, one for each GRIB1 parameter table number. xxx will
have three digits with the GRIB1 parameter table number.

Parameter
swap STRING Swap the position of obase and xxx in the output filename
uuid=<attname> STRING Add a UUID as global attribute <attname> to each output file

Environment
CDO_FILE_SUFFIX Set the default file suffix. This suffix will be added to the output file names
instead of the filename extension derived from the file format. Set this variable
to NULL to disable the adding of a file suffix.

Note
The operators in this module need to open all output files simultaneously. The maximum number of
open files depends on the operating system!

31
File operations Reference manual

Example
Assume an input GRIB1 dataset with three variables, e.g. code number 129, 130 and 139. To split
this dataset into three pieces, one for each code number use:
cdo splitcode infile code

Result of ’dir code*’:

code129 . grb code130 . grb code139 . grb

32
Reference manual File operations

2.2.8. SPLITTIME - Split timesteps of a dataset

Synopsis

< operator > infile obase

splitmon[,format] infile obase

Description
This module splits infile into timesteps pieces. The output files will be named <obase><xxx><suffix>
where suffix is the filename extension derived from the file format. xxx and the contents of the out-
put files depends on the chosen operator.

Operators

splithour Split hours

Splits a file into pieces, one for each different hour. xxx will have two digits with
the hour.
splitday Split days
Splits a file into pieces, one for each different day. xxx will have two digits with
the day.
splitseas Split seasons
Splits a file into pieces, one for each different season. xxx will have three characters
with the season.
splityear Split years
Splits a file into pieces, one for each different year. xxx will have four digits with
the year (YYYY).
splityearmon Split in years and months
Splits a file into pieces, one for each different year and month. xxx will have six
digits with the year and month (YYYYMM).
splitmon Split months
Splits a file into pieces, one for each different month. xxx will have two digits with
the month.

Parameter
format STRING C-style format for strftime() (e.g. %B for the full month name)

Environment
CDO_FILE_SUFFIX Set the default file suffix. This suffix will be added to the output file names
instead of the filename extension derived from the file format. Set this variable
to NULL to disable the adding of a file suffix.

Note
The operators in this module need to open all output files simultaneously. The maximum number of
open files depends on the operating system!

33
File operations Reference manual

Example
Assume the input GRIB1 dataset has timesteps from January to December. To split each month with
all variables into one separate file use:
cdo splitmon infile mon

Result of ’dir mon*’:

mon01 . grb mon02 . grb mon03 . grb mon04 . grb mon05 . grb mon06 . grb
mon07 . grb mon08 . grb mon09 . grb mon10 . grb mon11 . grb mon12 . grb

2.2.9. SPLITSEL - Split selected timesteps

Synopsis

splitsel,nsets[,noffset[,nskip]] infile obase

Description
This operator splits infile into pieces, one for each adjacent sequence t_1, ...., t_n of timesteps of
the same selected time range. The output files will be named <obase><nnnnnn><suffix> where
nnnnnn is the sequence number and suffix is the filename extension derived from the file format.

Parameter
nsets INTEGER Number of input timesteps for each output file
noffset INTEGER Number of input timesteps skipped before the first timestep range (optional)
nskip INTEGER Number of input timesteps skipped between timestep ranges (optional)

Environment
CDO_FILE_SUFFIX Set the default file suffix. This suffix will be added to the output file names
instead of the filename extension derived from the file format. Set this variable
to NULL to disable the adding of a file suffix.

34
Reference manual File operations

2.2.10. DISTGRID - Distribute horizontal grid

Synopsis

distgrid,nx[,ny] infile obase

Description
This operator distributes a dataset into smaller pieces. Each output file contains a different region of
the horizontal source grid. A target grid region contains a structured longitude/latitude box of the
source grid. Only rectilinear and curvilinear source grids are supported by this operator. The number
of different regions can be specified with the parameter nx and ny. The output files will be named
<obase><xxx><suffix> where suffix is the filename extension derived from the file format. xxx will
have five digits with the number of the target region.

Parameter
nx INTEGER Number of regions in x direction
ny INTEGER Number of regions in y direction [default: 1]

Note
This operator needs to open all output files simultaneously. The maximum number of open files
depends on the operating system!

Example
Distribute a file into 6 smaller files, each output file receives one half of x and a third of y of the
source grid:
cdo distgrid,2,3 infile.nc obase

Below is a schematic illustration of this example:

−20 20 20 60
−20 20 60 45 45
45 45

15 15
15 15 15 15

−15 −15 −15 −15

−15 −15

−45 −45
−20 20 60 −45 −45
−20 20 20 60

On the left side is the data of the input file and on the right side is the data of the six output files.

35
File operations Reference manual

2.2.11. COLLGRID - Collect horizontal grid

Synopsis

collgrid[,nx[,names]] infiles outfile

Description
This operator collects the data of the input files to one output file. All input files need to have
the same variables and the same number of timesteps on a different horizonal grid region. A source
region must be a structured longitude/latitude grid box. And all regions together muss give a new
structured longitude/latitude grid box. The parameter nx needs to be specified only for non regular
lon/lat grids.

Parameter
nx INTEGER Number of regions in x direction [default: number of input files]
names STRING Comma separated list of variable names [default: all variables]

Note
This operator needs to open all input files simultaneously. The maximum number of open files depends
on the operating system!

Example
Collect the horizonal grid of 6 input files. Each input file contains a lon/lat region of the target grid:
cdo collgrid infile[1-6] outfile

Below is a schematic illustration of this example:

−20 20 20 60
45 45 −20 20 60
45 45

15 15
15 15 15 15

−15 −15 −15 −15

−15 −15

−45 −45
−45 −45 −20 20 60
−20 20 20 60

On the left side is the data of the six input files and on the right side is the collected data of the
output file.

36
Reference manual Selection

2.3. Selection
This section contains modules to select time steps, fields or a part of a field from a dataset.
Here is a short overview of all operators in this section:

select Select fields

delete Delete fields

selmulti Select multiple fields

delmulti Delete multiple fields
changemulti Change identication of multiple fields

selparam Select parameters by identifier

delparam Delete parameters by identifier
selcode Select parameters by code number
delcode Delete parameters by code number
selname Select parameters by name
delname Delete parameters by name
selstdname Select parameters by standard name
sellevel Select levels
sellevidx Select levels by index
selgrid Select grids
selzaxis Select z-axes
selzaxisname Select z-axes by name
selltype Select GRIB level types
seltabnum Select parameter table numbers

seltimestep Select timesteps

seltime Select times
selhour Select hours
selday Select days
selmonth Select months
selyear Select years
selseason Select seasons
seldate Select dates
selsmon Select single month

sellonlatbox Select a longitude/latitude box

selindexbox Select an index box

selgridcell Select grid cells

delgridcell Delete grid cells

samplegrid Resample grid

37
Selection Reference manual

2.3.1. SELECT - Select fields

Synopsis
< operator >,params infiles outfile

Description
This module selects some fields from infiles and writes them to outfile. infiles is an arbitrary
number of input files. All input files need to have the same structure with the same variables on
different timesteps. The fields selected depends on the chosen parameters. Parameter is a comma
separated list of "key=value" pairs. Wildcards can be used for string parameter.

Operators
select Select fields
Selects all fields with parameters in a user given list.
delete Delete fields
Deletes all fields with parameters in a user given list.

Parameter
name STRING Comma separated list of variable names.
param STRING Comma separated list of parameter identifiers.
code INTEGER Comma separated list of code numbers.
level FLOAT Comma separated list of vertical levels.
levidx INTEGER Comma separated list of index of levels.
zaxisname STRING Comma separated list of zaxis names.
zaxisnum INTEGER Comma separated list of zaxis numbers.
ltype INTEGER Comma separated list of GRIB level types.
gridname STRING Comma separated list of grid names.
gridnum INTEGER Comma separated list of grid numbers.
steptype STRING Comma separated list of timestep types.
date STRING Comma separated list of dates (format YYYY-MM-DDThh:mm:ss).
startdate STRING Start date (format YYYY-MM-DDThh:mm:ss).
enddate STRING End date (format YYYY-MM-DDThh:mm:ss).
minute INTEGER Comma separated list of minutes.
hour INTEGER Comma separated list of hours.
day INTEGER Comma separated list of days.
month INTEGER Comma separated list of months.
season STRING Comma separated list of seasons (substring of DJFMAMJJA-
SOND or ANN).
year INTEGER Comma separated list of years.
timestep INTEGER Comma separated list of timesteps. Negative values selects timesteps
from the end (NetCDF only).
timestep_of_year INTEGER Comma separated list of timesteps of year.
timestepmask STRING Read timesteps from a mask file.

38
Reference manual Selection

Example
Assume you have 3 inputfiles. Each inputfile contains the same variables for a different time period.
To select the variable T,U and V on the levels 200, 500 and 850 from all 3 input files, use:
cdo select,name=T,U,V,level=200,500,850 infile1 infile2 infile3 outfile

39
Selection Reference manual

2.3.2. SELMULTI - Select multiple fields via GRIB1 parameters

Synopsis
< operator >,selection-specification infile outfile

Description
This module selects multiple fields from infile and writes them to outfile. selection-specification
is a filename or in-place string with the selection specification. Each selection-specification has the
following compact notation format:

<type>(parameters; leveltype(s); levels)

type sel for select or del for delete (optional)

parameters GRIB1 parameter code number
leveltype GRIB1 level type
levels value of each level
Examples:

( 1 ; 103; 0)
( 3 3 , 3 4 ; 105; 10)
( 1 1 , 1 7 ; 105; 2)
(71 ,73 ,74 ,75 ,61 ,62 ,65 ,117 ,67 ,122 ,121 ,11 ,131 ,66 ,84 ,111 ,112; 105; 0)

The following descriptive notation can also be used for selection specification from a file:

SELECT/DELETE, PARAMETER=parameters, LEVTYPE=leveltye(s), LEVEL=levels

Examples:

SELECT, PARAMETER=1 , LEVTYPE=103 , LEVEL=0

SELECT, PARAMETER=33/34 , LEVTYPE=105 , LEVEL=10
SELECT, PARAMETER=11/17 , LEVTYPE=105 , LEVEL=2
SELECT, PARAMETER=71/73/74/75/61/62/65/117/67/122 , LEVTYPE=105 , LEVEL=0
DELETE, PARAMETER=128 , LEVTYPE=109 , LEVEL=∗

The following will convert Pressure from Pa into hPa; Temp from Kelvin to Celsius:
SELECT, PARAMETER=1 , LEVTYPE= 1 0 3 , LEVEL=0, SCALE=0.01
SELECT, PARAMETER=11 , LEVTYPE=105 , LEVEL=2 , OFFSET=273.15

If SCALE and/or OFFSET are defined, then the data values are scaled as SCALE*(VALUE-OFFSET).

Operators
selmulti Select multiple fields

delmulti Delete multiple fields

changemulti Change identication of multiple fields

40
Reference manual Selection

Example
Change ECMWF GRIB code of surface pressure to Hirlam notation:
cdo changemulti,’{(134;1;*|1;105;*)}’ infile outfile

41
Selection Reference manual

2.3.3. SELVAR - Select fields

Synopsis

< operator >,params infile outfile

selcode,codes infile outfile
delcode,codes infile outfile
selname,names infile outfile
delname,names infile outfile
selstdname,stdnames infile outfile
sellevel,levels infile outfile
sellevidx,levidx infile outfile
selgrid,grids infile outfile
selzaxis,zaxes infile outfile
selzaxisname,zaxisnames infile outfile
selltype,ltypes infile outfile
seltabnum,tabnums infile outfile

Description
This module selects some fields from infile and writes them to outfile. The fields selected depends
on the chosen operator and the parameters.

Operators

selparam Select parameters by identifier

Selects all fields with parameter identifiers in a user given list.
delparam Delete parameters by identifier
Deletes all fields with parameter identifiers in a user given list.
selcode Select parameters by code number
Selects all fields with code numbers in a user given list.
delcode Delete parameters by code number
Deletes all fields with code numbers in a user given list.
selname Select parameters by name
Selects all fields with parameter names in a user given list.
delname Delete parameters by name
Deletes all fields with parameter names in a user given list.
selstdname Select parameters by standard name
Selects all fields with standard names in a user given list.
sellevel Select levels
Selects all fields with levels in a user given list.
sellevidx Select levels by index
Selects all fields with index of levels in a user given list.
selgrid Select grids
Selects all fields with grids in a user given list.

42
Reference manual Selection

selzaxis Select z-axes

Selects all fields with z-axes in a user given list.
selzaxisname Select z-axes by name
Selects all fields with z-axis names in a user given list.
selltype Select GRIB level types
Selects all fields with GRIB level type in a user given list.
seltabnum Select parameter table numbers
Selects all fields with parameter table numbers in a user given list.

Parameter
params INTEGER Comma separated list of parameter identifiers
codes INTEGER Comma separated list of code numbers
names STRING Comma separated list of variable names
stdnames STRING Comma separated list of standard names
levels FLOAT Comma separated list of vertical levels
levidx INTEGER Comma separated list of index of levels
ltypes INTEGER Comma separated list of GRIB level types
grids STRING Comma separated list of grid names or numbers
zaxes STRING Comma separated list of z-axis types or numbers
zaxisnames STRING Comma separated list of z-axis names
tabnums INTEGER Comma separated list of parameter table numbers

Example
Assume an input dataset has three variables with the code numbers 129, 130 and 139. To select the
variables with the code number 129 and 139 use:
cdo selcode,129,139 infile outfile

You can also select the code number 129 and 139 by deleting the code number 130 with:
cdo delcode,130 infile outfile

43
Selection Reference manual

2.3.4. SELTIME - Select timesteps

Synopsis

seltimestep,timesteps infile outfile

seltime,times infile outfile
selhour,hours infile outfile
selday,days infile outfile
selmonth,months infile outfile
selyear,years infile outfile
selseason,seasons infile outfile
seldate,startdate[,enddate] infile outfile
selsmon,month[,nts1[,nts2]] infile outfile

Description
This module selects user specified timesteps from infile and writes them to outfile. The timesteps
selected depends on the chosen operator and the parameters.

Operators

seltimestep Select timesteps

Selects all timesteps with a timestep in a user given list.
seltime Select times
Selects all timesteps with a time in a user given list.
selhour Select hours
Selects all timesteps with a hour in a user given list.
selday Select days
Selects all timesteps with a day in a user given list.
selmonth Select months
Selects all timesteps with a month in a user given list.
selyear Select years
Selects all timesteps with a year in a user given list.
selseason Select seasons
Selects all timesteps with a month of a season in a user given list.
seldate Select dates
Selects all timesteps with a date in a user given range.
selsmon Select single month
Selects a month and optional an arbitrary number of timesteps before and after this
month.

Parameter
timesteps INTEGER Comma separated list of timesteps. Negative values selects timesteps from
the end (NetCDF only).
times STRING Comma separated list of times (format hh:mm:ss).
hours INTEGER Comma separated list of hours.

44
Reference manual Selection

days INTEGER Comma separated list of days.

months INTEGER Comma separated list of months.
years INTEGER Comma separated list of years.
seasons STRING Comma separated list of seasons (substring of DJFMAMJJASOND or
ANN).
startdate STRING Start date (format YYYY-MM-DDThh:mm:ss).
enddate STRING End date (format YYYY-MM-DDThh:mm:ss) [default: startdate].
nts1 INTEGER Number of timesteps before the selected month [default: 0].
nts2 INTEGER Number of timesteps after the selected month [default: nts1].

45
Selection Reference manual

2.3.5. SELBOX - Select a box of a field

Synopsis

sellonlatbox,lon1,lon2,lat1,lat2 infile outfile

selindexbox,idx1,idx2,idy1,idy2 infile outfile

Description
Selects a box of the rectangularly understood field.

Operators

sellonlatbox Select a longitude/latitude box

Selects a regular longitude/latitude box. The user has to give the longitudes and
latitudes of the edges of the box. Considered are only those grid cells with the grid
center inside the lon/lat box. For rotated lon/lat grids the parameter needs to be
rotated coordinates.
selindexbox Select an index box
Selects an index box. The user has to give the indexes of the edges of the box. The
index of the left edge may be greater then that of the right edge.

Parameter
lon1 FLOAT Western longitude
lon2 FLOAT Eastern longitude
lat1 FLOAT Southern or northern latitude
lat2 FLOAT Northern or southern latitude
idx1 INTEGER Index of first longitude (1 - nlon)
idx2 INTEGER Index of last longitude (1 - nlon)
idy1 INTEGER Index of first latitude (1 - nlat)
idy2 INTEGER Index of last latitude (1 - nlat)

Example
To select the region with the longitudes from 30W to 60E and latitudes from 30N to 80N from all
input fields use:
cdo sellonlatbox,-30,60,30,80 infile outfile

If the input dataset has fields on a Gaussian N16 grid, the same box can be selected with selindexbox
by:
cdo selindexbox,60,11,3,11 infile outfile

46
Reference manual Selection

2.3.6. SELGRIDCELL - Select grid cells

Synopsis

< operator >,indexes infile outfile

Description
Selects grid cells of all fields from infile. The user has to give the indexes of each grid cell. The
resulting grid in outfile is unstructured.

Operators

selgridcell Select grid cells

delgridcell Delete grid cells

Parameter
indexes INTEGER Comma separated list of indexes

2.3.7. SAMPLEGRID - Resample grid

Synopsis

samplegrid,factor infile outfile

Description
This is a special operator for resampling the horizontal grid. No interpolation takes place. Resample
factor=2 means every second grid point is removed. Only rectilinear and curvilinear source grids are
supported by this operator.

Parameter
factor INTEGER Resample factor, typically 2, which will half the resolution

47
Conditional selection Reference manual

2.4. Conditional selection

This section contains modules to conditional select field elements. The fields in the first input file are
handled as a mask. A value not equal to zero is treated as "true", zero is treated as "false".
Here is a short overview of all operators in this section:

ifthen If then
ifnotthen If not then

ifthenelse If then else

ifthenc If then constant

ifnotthenc If not then constant

reducegrid Reduce input file variables to locations, where mask is non-zero.

48
Reference manual Conditional selection

2.4.1. COND - Conditional select one field

Synopsis

< operator > infile1 infile2 outfile

Description
This module selects field elements from infile2 with respect to infile1 and writes them to outfile.
The fields in infile1 are handled as a mask. A value not equal to zero is treated as "true", zero is
treated as "false". The number of fields in infile1 has either to be the same as in infile2 or the
same as in one timestep of infile2 or only one. The fields in outfile inherit the meta data from
infile2.

Operators

ifthen If then 
i2 (t, x) if i1 ([t, ]x) 6= 0 ∧ i1 ([t, ]x) 6= miss
o(t, x) =
miss if i1 ([t, ]x) = 0 ∨ i1 ([t, ]x) = miss
ifnotthen If not then

i2 (t, x) if i1 ([t, ]x) = 0 ∧ i1 ([t, ]x) 6= miss
o(t, x) =
miss if i1 ([t, ]x) 6= 0 ∨ i1 ([t, ]x) = miss

Example
To select all field elements of infile2 if the corresponding field element of infile1 is greater than
0 use:
cdo ifthen infile1 infile2 outfile

2.4.2. COND2 - Conditional select two fields

Synopsis

ifthenelse infile1 infile2 infile3 outfile

Description
This operator selects field elements from infile2 or infile3 with respect to infile1 and writes
them to outfile. The fields in infile1 are handled as a mask. A value not equal to zero is treated
as "true", zero is treated as "false". The number of fields in infile1 has either to be the same as in
infile2 or the same as in one timestep of infile2 or only one. infile2 and infile3 need to have
the same number of fields. The fields in outfile inherit the meta data from infile2.

 i2 (t, x) if i1 ([t, ]x) 6= 0 ∧ i1 ([t, ]x) 6= miss
o(t, x) = i3 (t, x) if i1 ([t, ]x) = 0 ∧ i1 ([t, ]x) 6= miss

miss if i1 ([t, ]x) = miss

Example
To select all field elements of infile2 if the corresponding field element of infile1 is greater than
0 and from infile3 otherwise use:
cdo ifthenelse infile1 infile2 infile3 outfile

49
Conditional selection Reference manual

2.4.3. CONDC - Conditional select a constant

Synopsis

< operator >,c infile outfile

Description
This module creates fields with a constant value or missing value. The fields in infile are handled
as a mask. A value not equal to zero is treated as "true", zero is treated as "false".

Operators

ifthenc If then constant


c if i(t, x) 6= 0 ∧ i(t, x) 6= miss
o(t, x) =
miss if i(t, x) = 0 ∨ i(t, x) = miss
ifnotthenc If not then
 constant
c if i(t, x) = 0 ∧ i(t, x) 6= miss
o(t, x) =
miss if i(t, x) 6= 0 ∨ i(t, x) = miss

Parameter
c FLOAT Constant

Example
To create fields with the constant value 7 if the corresponding field element of infile is greater than
0 use:
cdo ifthenc,7 infile outfile

50
Reference manual Conditional selection

2.4.4. MAPREDUCE - Reduce fields to user-defined mask

Synopsis

reducegrid,mask[,limitCoordsOutput] infile outfile

Description
This module holds an operator for data reduction based on a user defined mask. The output grid
is unstructured and includes coordinate bounds. Bounds can be avoided by using the additional
’nobounds’ keyword. With ’nocoords’ given, coordinates a completely suppressed.

Parameter
mask STRING file which holds the mask field
limitCoordsOutput STRING optional parameter to limit coordinates output: ’nobounds’ dis-
ables coordinate bounds, ’nocoords’ avoids all coordinate information

Example
To limit data fields to land values, a mask has to be created first with
cdo -gtc,0 -topo,ni96 lsm_gme96.grb

Here a GME grid is used. Say temp_gme96.grb contains a global temperture field. The following
command limits the global grid to landpoints.
cdo -f nc reduce,lsm_gme96.grb temp_gme96.grb tempOnLand_gme96.nc

Note that output file type is NetCDF, because unstructured grids cannot be stored in GRIB format.

51
Comparison Reference manual

2.5. Comparison
This section contains modules to compare datasets. The resulting field is a mask containing 1 if the
comparison is true and 0 if not.
Here is a short overview of all operators in this section:

eq Equal
ne Not equal
le Less equal
lt Less than
ge Greater equal
gt Greater than

eqc Equal constant

nec Not equal constant
lec Less equal constant
ltc Less than constant
gec Greater equal constant
gtc Greater than constant

52
Reference manual Comparison

2.5.1. COMP - Comparison of two fields

Synopsis

< operator > infile1 infile2 outfile

Description
This module compares two datasets field by field. The resulting field is a mask containing 1 if the
comparison is true and 0 if not. The number of fields in infile1 should be the same as in infile2.
One of the input files can contain only one timestep or one field. The fields in outfile inherit the
meta data from infile1 or infile2. The type of comparison depends on the chosen operator.

Operators

eq Equal 
 1 if i1 (t, x) = i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
o(t, x) = 0 if i1 (t, x) 6= i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
∨ i2 (t, x) = miss

miss if i1 (t, x) = miss
ne Not equal
 1 if i1 (t, x) 6= i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
o(t, x) = 0 if i1 (t, x) = i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
∨ i2 (t, x) = miss

miss if i1 (t, x) = miss
le Less equal
 1 if i1 (t, x) ≤ i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
o(t, x) = 0 if i1 (t, x) > i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
∨ i2 (t, x) = miss

miss if i1 (t, x) = miss
lt Less than
 1 if i1 (t, x) < i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
o(t, x) = 0 if i1 (t, x) ≥ i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
∨ i2 (t, x) = miss

miss if i1 (t, x) = miss
ge Greater equal

 1 if i1 (t, x) ≥ i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
o(t, x) = 0 if i1 (t, x) < i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
∨ i2 (t, x) = miss

miss if i1 (t, x) = miss
gt Greater than

 1 if i1 (t, x) > i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
o(t, x) = 0 if i1 (t, x) ≤ i2 (t, x) ∧ i1 (t, x), i2 (t, x) 6= miss
∨ i2 (t, x) = miss

miss if i1 (t, x) = miss

Example
To create a mask containing 1 if the elements of two fields are the same and 0 if the elements are
different use:
cdo eq infile1 infile2 outfile

53
Comparison Reference manual

2.5.2. COMPC - Comparison of a field with a constant

Synopsis

< operator >,c infile outfile

Description
This module compares all fields of a dataset with a constant. The resulting field is a mask containing
1 if the comparison is true and 0 if not. The type of comparison depends on the chosen operator.

Operators

eqc Equal constant


 1 if i(t, x) = c ∧ i(t, x), c 6= miss
o(t, x) = 0 if i(t, x) 6= c ∧ i(t, x), c 6= miss
miss if i(t, x) = miss ∨ c = miss


nec Not equalconstant

 1 if i(t, x) 6= c ∧ i(t, x), c 6= miss
o(t, x) = 0 if i(t, x) = c ∧ i(t, x), c 6= miss
miss if i(t, x) = miss ∨ c = miss


lec Less equalconstant

 1 if i(t, x) ≤ c ∧ i(t, x), c 6= miss
o(t, x) = 0 if i(t, x) > c ∧ i(t, x), c 6= miss
miss if i(t, x) = miss ∨ c = miss


ltc Less thanconstant

 1 if i(t, x) < c ∧ i(t, x), c 6= miss
o(t, x) = 0 if i(t, x) ≥ c ∧ i(t, x), c 6= miss
miss if i(t, x) = miss ∨ c = miss


gec Greater equal

 constant
 1 if i(t, x) ≥ c ∧ i(t, x), c 6= miss
o(t, x) = 0 if i(t, x) < c ∧ i(t, x), c 6= miss
miss if i(t, x) = miss ∨ c = miss


gtc Greater than

 constant
 1 if i(t, x) > c ∧ i(t, x), c 6= miss
o(t, x) = 0 if i(t, x) ≤ c ∧ i(t, x), c 6= miss
miss if i(t, x) = miss ∨ c = miss


Parameter
c FLOAT Constant

Example
To create a mask containing 1 if the field element is greater than 273.15 and 0 if not use:
cdo gtc,273.15 infile outfile

54
Reference manual Modification

2.6. Modification

This section contains modules to modify the metadata, fields or part of a field in a dataset.
Here is a short overview of all operators in this section:

setattribute Set attributes

setpartabp Set parameter table

setpartabn Set parameter table

setcodetab Set parameter code table

setcode Set code number
setparam Set parameter identifier
setname Set variable name
setunit Set variable unit
setlevel Set level
setltype Set GRIB level type

setdate Set date

settime Set time of the day
setday Set day
setmon Set month
setyear Set year
settunits Set time units
settaxis Set time axis
settbounds Set time bounds
setreftime Set reference time
setcalendar Set calendar
shifttime Shift timesteps

chcode Change code number

chparam Change parameter identifier
chname Change variable or coordinate name
chunit Change variable unit
chlevel Change level
chlevelc Change level of one code
chlevelv Change level of one variable

setgrid Set grid

setgridtype Set grid type
setgridarea Set grid cell area
setgridmask Set grid mask

setzaxis Set z-axis

genlevelbounds Generate level bounds

invertlat Invert latitudes

invertlev Invert levels

shiftx Shift x
shifty Shift y

maskregion Mask regions

55
Modification Reference manual

masklonlatbox Mask a longitude/latitude box

maskindexbox Mask an index box

setclonlatbox Set a longitude/latitude box to constant

setcindexbox Set an index box to constant

enlarge Enlarge fields

setmissval Set a new missing value

setctomiss Set constant to missing value
setmisstoc Set missing value to constant
setrtomiss Set range to missing value
setvrange Set valid range
setmisstonn Set missing value to nearest neighbor
setmisstodis Set missing value to distance-weighted average

56
Reference manual Modification

2.6.1. SETATTRIBUTE - Set attributes

Synopsis
setattribute,attributes infile outfile

Description
This operator sets attributes of a dataset and writes the result to outfile. The new attributes are
only available in outfile if the file format supports attributes.
Each attribute has the following structure:
[var_nm@]att_nm=att_val
var_nm Variable name (optional). Example: pressure
att_nm Attribute name. Example: units
att_val Comma separated list of attribute values. Example: pascal
The value of var_nm is the name of the variable containing the attribute (named att_nm) that
you want to set. Use wildcards to set the attribute att_nm to more than one variable. A value
of var_nm of ’*’ will set the attribute att_nm to all data variables. If var_nm is missing then
att_nm refers to a global attribute.
The value of att_nm is the name of the attribute you want to set.
The value of att_val is the contents of the attribute att_nm. att_val may be a single value or
one-dimensional array of elements. The type and the number of elements of an attribute will be
detected automaticly from the contents of the values.
A special meaning has the attribute name FILE. If this is the 1st attribute then all attributes are
read from a file specified in the value of att_val.

Parameter
attributes STRING Comma separated list of attributes.

Note
Attributes are evaluated by CDO when opening infile. Therefor the result of this operator is not
available for other operators when this operator is used in chaining operators.

Example
To set the units of the variable pressure to pascal use:
cdo setattribute,pressure@units=pascal infile outfile

To set the global text attribute "my_att" to "my contents", use:

cdo setattribute,my_att="my contents" infile outfile

Result of ’ncdump -h outfile’:

netcdf o u t f i l e {
dimensions : . . .

variables : ...

// g l o b a l a t t r i b u t e s :
: my_att = "my c o n t e n t s " ;
}

57
Modification Reference manual

2.6.2. SETPARTAB - Set parameter table

Synopsis

< operator >,table[,convert] infile outfile

Description
This module transforms data and metadata of infile via a parameter table and writes the result
to outfile. A parameter table is an ASCII formatted file with a set of parameter entries for each
variable. Each new set have to start with "&parameter" and to end with "/".
The following parameter table entries are supported:

Entry Type Description

name WORD Name of the variable
out_name WORD New name of the variable
param WORD Parameter identifier (GRIB1: code[.tabnum]; GRIB2: num[.cat[.dis]])
out_param WORD New parameter identifier
type WORD Data type (real or double)
standard_name WORD As defined in the CF standard name table
long_name STRING Describing the variable
units STRING Specifying the units for the variable
comment STRING Information concerning the variable
cell_methods STRING Information concerning calculation of means or climatologies
cell_measures STRING Indicates the names of the variables containing cell areas and volumes
missing_value FLOAT Specifying how missing data will be identified
valid_min FLOAT Minimum valid value
valid_max FLOAT Maximum valid value
ok_min_mean_abs FLOAT Minimum absolute mean
ok_max_mean_abs FLOAT Maximum absolute mean
factor FLOAT Scale factor
delete INTEGER Set to 1 to delete variable
convert INTEGER Set to 1 to convert the unit if necessary

Unsupported parameter table entries are stored as variable attributes. The search key for the variable
depends on the operator. Use setpartabn to search variables by the name. This is typically used for
NetCDF datasets. The operator setpartabp searches variables by the parameter ID.

Operators

setpartabp Set parameter table

Search variables by the parameter identifier.
setpartabn Set parameter table
Search variables by name.

Parameter
table STRING Parameter table file or name
convert STRING Converts the units if necessary

58
Reference manual Modification

Example
Here is an example of a parameter table for one variable:
prompt> cat mypartab
&parameter
name = t
out_name = ta
standard_name = air_temperature
units = "K"
missing_value = 1e+20
valid_min = 157.1
valid_max = 336.3
/

To apply this parameter table to a dataset use:

cdo setpartabn,mypartab,convert infile outfile

This command renames the variable t to ta. The standard name of this variable is set to air_temperature
and the unit is set to [K] (converts the unit if necessary). The missing value will be set to 1e+20.
In addition it will be checked whether the values of the variable are in the range of 157.1 to 336.3.

59
Modification Reference manual

2.6.3. SET - Set field info

Synopsis

setcodetab,table infile outfile

setcode,code infile outfile
setparam,param infile outfile
setname,name infile outfile
setunit,unit infile outfile
setlevel,level infile outfile
setltype,ltype infile outfile

Description
This module sets some field information. Depending on the chosen operator the parameter table,
code number, parameter identifier, variable name or level is set.

Operators

setcodetab Set parameter code table

Sets the parameter code table for all variables.
setcode Set code number
Sets the code number for all variables to the same given value.
setparam Set parameter identifier
Sets the parameter identifier of the first variable.
setname Set variable name
Sets the name of the first variable.
setunit Set variable unit
Sets the unit of the first variable.
setlevel Set level
Sets the first level of all variables.
setltype Set GRIB level type
Sets the GRIB level type of all variables.

Parameter
table STRING Parameter table file or name
code INTEGER Code number
param STRING Parameter identifier (GRIB1: code[.tabnum]; GRIB2: num[.cat[.dis]])
name STRING Variable name
level FLOAT New level
ltype INTEGER GRIB level type

60
Reference manual Modification

2.6.4. SETTIME - Set time

Synopsis

setdate,date infile outfile

settime,time infile outfile
setday,day infile outfile
setmon,month infile outfile
setyear,year infile outfile
settunits,units infile outfile
settaxis,date,time[,inc] infile outfile
settbounds,frequency infile outfile
setreftime,date,time[,units] infile outfile
setcalendar,calendar infile outfile
shifttime,sval infile outfile

Description
This module sets the time axis or part of the time axis. Which part of the time axis is overwrit-
ten/created depends on the chosen operator.

Operators

setdate Set date

Sets the date in every timestep to the same given value.
settime Set time of the day
Sets the time in every timestep to the same given value.
setday Set day
Sets the day in every timestep to the same given value.
setmon Set month
Sets the month in every timestep to the same given value.
setyear Set year
Sets the year in every timestep to the same given value.
settunits Set time units
Sets the base units of a relative time axis.
settaxis Set time axis
Sets the time axis.
settbounds Set time bounds
Sets the time bounds.
setreftime Set reference time
Sets the reference time of a relative time axis.
setcalendar Set calendar
Sets the calendar of a relative time axis.
shifttime Shift timesteps
Shifts all timesteps by the parameter sval.

61
Modification Reference manual

Parameter
day INTEGER Value of the new day
month INTEGER Value of the new month
year INTEGER Value of the new year
units STRING Base units of the time axis (seconds, minutes, hours, days, months, years)
date STRING Date (format: YYYY-MM-DD)
time STRING Time (format: hh:mm:ss)
inc STRING Optional increment (seconds, minutes, hours, days, months, years) [de-
fault: 1hour]
frequency STRING Frequency of the time series (hour, day, month, year)
calendar STRING Calendar (standard, proleptic_gregorian, 360_day, 365_day, 366_day)
sval STRING Shift value (e.g. -3hour)

Example
To set the time axis to 1987-01-16 12:00:00 with an increment of one month for each timestep use:
cdo settaxis,1987-01-16,12:00:00,1mon infile outfile

Result of ’cdo showdate outfile’ for a dataset with 12 timesteps:

1987−01−16 1987−02−16 1987−03−16 1987−04−16 1987−05−16 1987−06−16 \
1987−07−16 1987−08−16 1987−09−16 1987−10−16 1987−11−16 1987−12−16

To shift this time axis by -15 days use:

cdo shifttime,-15days infile outfile

Result of ’cdo showdate outfile’:

1987−01−01 1987−02−01 1987−03−01 1987−04−01 1987−05−01 1987−06−01 \
1987−07−01 1987−08−01 1987−09−01 1987−10−01 1987−11−01 1987−12−01

62
Reference manual Modification

2.6.5. CHANGE - Change field header

Synopsis
chcode,oldcode,newcode[,...] infile outfile
chparam,oldparam,newparam,... infile outfile
chname,oldname,newname,... infile outfile
chunit,oldunit,newunit,... infile outfile
chlevel,oldlev,newlev,... infile outfile
chlevelc,code,oldlev,newlev infile outfile
chlevelv,name,oldlev,newlev infile outfile

Description
This module reads fields from infile, changes some header values and writes the results to outfile.
The kind of changes depends on the chosen operator.

Operators
chcode Change code number
Changes some user given code numbers to new user given values.
chparam Change parameter identifier
Changes some user given parameter identifiers to new user given values.
chname Change variable or coordinate name
Changes some user given variable or coordinate names to new user given names.
chunit Change variable unit
Changes some user given variable units to new user given units.
chlevel Change level
Changes some user given levels to new user given values.
chlevelc Change level of one code
Changes one level of a user given code number.
chlevelv Change level of one variable
Changes one level of a user given variable name.

Parameter
code INTEGER Code number
oldcode,newcode,... INTEGER Pairs of old and new code numbers
oldparam,newparam,... STRING Pairs of old and new parameter identifiers
name STRING Variable name
oldname,newname,... STRING Pairs of old and new variable names
oldlev FLOAT Old level
newlev FLOAT New level
oldlev,newlev,... FLOAT Pairs of old and new levels

Example
To change the code number 98 to 179 and 99 to 211 use:
cdo chcode,98,179,99,211 infile outfile

63
Modification Reference manual

2.6.6. SETGRID - Set grid information

Synopsis

setgrid,grid infile outfile

setgridtype,gridtype infile outfile
setgridarea,gridarea infile outfile
setgridmask,gridmask infile outfile

Description
This module modifies the metadata of the horizontal grid. Depending on the chosen operator a new
grid description is set, the coordinates are converted or the grid cell area is added.

Operators

setgrid Set grid

Sets a new grid description. The input fields need to have the same grid size as the
size of the target grid description.
setgridtype Set grid type
Sets the grid type of all input fields. The following grid types are available:
curvilinear Converts a regular grid to a curvilinear grid
unstructured Converts a regular or curvilinear grid to an unstructured grid
dereference Dereference a reference to a grid
regular Linear interpolation of a reduced Gaussian grid to a regular Gaus-
sian grid
regularnn Nearest neighbor interpolation of a reduced Gaussian grid to a
regular Gaussian grid
lonlat Converts a regular lonlat grid stored as a curvilinear grid back to
a lonlat grid
setgridarea Set grid cell area
Sets the grid cell area. The parameter gridarea is the path to a data file, the first
field is used as grid cell area. The input fields need to have the same grid size as the
grid cell area. The grid cell area is used to compute the weights of each grid cell if
needed by an operator, e.g. for fldmean.
setgridmask Set grid mask
Sets the grid mask. The parameter gridmask is the path to a data file, the first field
is used as the grid mask. The input fields need to have the same grid size as the
grid mask. The grid mask is used as the target grid mask for remapping, e.g. for
remapbil.

Parameter
grid STRING Grid description file or name
gridtype STRING Grid type (curvilinear, unstructured, regular, lonlat or dereference)
gridarea STRING Data file, the first field is used as grid cell area
gridmask STRING Data file, the first field is used as grid mask

64
Reference manual Modification

Example
Assuming a dataset has fields on a grid with 2048 elements without or with wrong grid description.
To set the grid description of all input fields to a Gaussian N32 grid (8192 gridpoints) use:
cdo setgrid,n32 infile outfile

2.6.7. SETZAXIS - Set z-axis information

Synopsis

setzaxis,zaxis infile outfile

genlevelbounds[,zbot[,ztop]] infile outfile

Description
This module modifies the metadata of the vertical grid.

Operators

setzaxis Set z-axis

This operator sets the z-axis description of all variables with the same number
of level as the new z-axis.
genlevelbounds Generate level bounds
Generates the layer bounds of the z-axis.

Parameter
zaxis STRING Z-axis description file or name of the target z-axis
zbot FLOAT Specifying the bottom of the vertical column. Must have the same units as
z-axis.
ztop FLOAT Specifying the top of the vertical column. Must have the same units as z-axis.

65
Modification Reference manual

2.6.8. INVERT - Invert latitudes

Synopsis

invertlat infile outfile

Description
This operator inverts the latitudes of all fields on a rectilinear grid.

Example
To invert the latitudes of a 2D field from N->S to S->N use:
cdo invertlat infile outfile

2.6.9. INVERTLEV - Invert levels

Synopsis

invertlev infile outfile

Description
This operator inverts the levels of all 3D variables.

66
Reference manual Modification

2.6.10. SHIFTXY - Shift field

Synopsis

< operator >,<nshift>,<cyclic>,<coord> infile outfile

Description
This module contains operators to shift all fields in x or y direction. All fields need to have the same
horizontal rectilinear or curvilinear grid.

Operators

shiftx Shift x
Shifts all fields in x direction.
shifty Shift y
Shifts all fields in y direction.

Parameter
nshift INTEGER Number of grid cells to shift (default: 1)
cyclic STRING If set, cells are filled up cyclic (default: missing value)
coord STRING If set, coordinates are also shifted

Example
To shift all input fields in the x direction by +1 cells and fill the new cells with missing value, use:
cdo shiftx infile outfile

To shift all input fields in the x direction by +1 cells and fill the new cells cyclic, use:
cdo shiftx,1,cyclic infile outfile

67
Modification Reference manual

2.6.11. MASKREGION - Mask regions

Synopsis

maskregion,regions infile outfile

Description
Masks different regions of fields with a regular lon/lat grid. The elements inside a region are un-
touched, the elements outside are set to missing value. Considered are only those grid cells with the
grid center inside the regions. All input fields must have the same horizontal grid. The user has to
give ASCII formatted files with different regions. A region is defined by a polygon. Each line of a
polygon description file contains the longitude and latitude of one point. Each polygon description
file can contain one or more polygons separated by a line with the character &.

Parameter
regions STRING Comma separated list of ASCII formatted files with different regions

Example
To mask the region with the longitudes from 120E to 90W and latitudes from 20N to 20S on all input
fields use:
cdo maskregion,myregion infile outfile

For this example the polygon description file myregion should contain the following four coordinates:
120 20
120 −20
270 −20
270 20

68
Reference manual Modification

2.6.12. MASKBOX - Mask a box

Synopsis

masklonlatbox,lon1,lon2,lat1,lat2 infile outfile

maskindexbox,idx1,idx2,idy1,idy2 infile outfile

Description
Masked a box of the rectangularly understood field. The elements inside the box are untouched, the
elements outside are set to missing value. All input fields need to have the same horizontal grid. Use
sellonlatbox or selindexbox if only the data inside the box are needed.

Operators

masklonlatbox Mask a longitude/latitude box

Masked a regular longitude/latitude box. The user has to give the longitudes
and latitudes of the edges of the box. Considered are only those grid cells with
the grid center inside the lon/lat box.
maskindexbox Mask an index box
Masked an index box. The user has to give the indexes of the edges of the box.
The index of the left edge can be greater then the one of the right edge.

Parameter
lon1 FLOAT Western longitude
lon2 FLOAT Eastern longitude
lat1 FLOAT Southern or northern latitude
lat2 FLOAT Northern or southern latitude
idx1 INTEGER Index of first longitude
idx2 INTEGER Index of last longitude
idy1 INTEGER Index of first latitude
idy2 INTEGER Index of last latitude

Example
To mask the region with the longitudes from 120E to 90W and latitudes from 20N to 20S on all input
fields use:
cdo masklonlatbox,120,-90,20,-20 infile outfile

If the input dataset has fields on a Gaussian N16 grid, the same box can be masked with maskindexbox
by:
cdo maskindexbox,23,48,13,20 infile outfile

69
Modification Reference manual

2.6.13. SETBOX - Set a box to constant

Synopsis

setclonlatbox,c,lon1,lon2,lat1,lat2 infile outfile

setcindexbox,c,idx1,idx2,idy1,idy2 infile outfile

Description
Sets a box of the rectangularly understood field to a constant value. The elements outside the box
are untouched, the elements inside are set to the given constant. All input fields need to have the
same horizontal grid.

Operators

setclonlatbox Set a longitude/latitude box to constant

Sets the values of a longitude/latitude box to a constant value. The user has to
give the longitudes and latitudes of the edges of the box.
setcindexbox Set an index box to constant
Sets the values of an index box to a constant value. The user has to give the
indexes of the edges of the box. The index of the left edge can be greater than
the one of the right edge.

Parameter
c FLOAT Constant
lon1 FLOAT Western longitude
lon2 FLOAT Eastern longitude
lat1 FLOAT Southern or northern latitude
lat2 FLOAT Northern or southern latitude
idx1 INTEGER Index of first longitude
idx2 INTEGER Index of last longitude
idy1 INTEGER Index of first latitude
idy2 INTEGER Index of last latitude

Example
To set all values in the region with the longitudes from 120E to 90W and latitudes from 20N to 20S
to the constant value -1.23 use:
cdo setclonlatbox,-1.23,120,-90,20,-20 infile outfile

If the input dataset has fields on a Gaussian N16 grid, the same box can be set with setcindexbox by:
cdo setcindexbox,-1.23,23,48,13,20 infile outfile

70
Reference manual Modification

2.6.14. ENLARGE - Enlarge fields

Synopsis

enlarge,grid infile outfile

Description
Enlarge all fields of infile to a user given horizontal grid. Normally only the last field element is
used for the enlargement. If however the input and output grid are regular lon/lat grids, a zonal or
meridional enlargement is possible. Zonal enlargement takes place, if the xsize of the input field is 1
and the ysize of both grids are the same. For meridional enlargement the ysize have to be 1 and the
xsize of both grids should have the same size.

Parameter
grid STRING Target grid description file or name

Example
Assumed you want to add two datasets. The first dataset is a field on a global grid (n field elements)
and the second dataset is a global mean (1 field element). Before you can add these two datasets the
second dataset have to be enlarged to the grid size of the first dataset:
cdo enlarge,infile1 infile2 tmpfile
cdo add infile1 tmpfile outfile

Or shorter using operator piping:

cdo add infile1 -enlarge,infile1 infile2 outfile

71
Modification Reference manual

2.6.15. SETMISS - Set missing value

Synopsis
setmissval,newmiss infile outfile
setctomiss,c infile outfile
setmisstoc,c infile outfile
setrtomiss,rmin,rmax infile outfile
setvrange,rmin,rmax infile outfile
setmisstonn infile outfile
setmisstodis[,neighbors] infile outfile

Description
This module sets part of a field to missing value or missing values to a constant value. Which part of
the field is set depends on the chosen operator.

Operators
setmissval Set a newmissing value
newmiss if i(t, x) = miss
o(t, x) =
i(t, x) if i(t, x) 6= miss
setctomiss Set constant
 to missing value
miss if i(t, x) = c
o(t, x) =
i(t, x) if i(t, x) 6= c
setmisstoc Set missing
 value to constant
c if i(t, x) = miss
o(t, x) =
i(t, x) if i(t, x) 6= miss
setrtomiss Set rangeto missing value
miss if i(t, x) ≥ rmin ∧ i(t, x) ≤ rmax
o(t, x) =
i(t, x) if i(t, x) < rmin ∨ i(t, x) > rmax
setvrange Set valid 
range
miss if i(t, x) < rmin ∨ i(t, x) > rmax
o(t, x) =
i(t, x) if i(t, x) ≥ rmin ∧ i(t, x) ≤ rmax
setmisstonn Set missing value to nearest neighbor
Set all missing values to the nearest non missing value.

i(t, y) if i(t, x) = miss ∧ i(t, y) 6= miss
o(t, x) =
i(t, x) if i(t, x) 6= miss
setmisstodis Set missing value to distance-weighted average
Set all missing values to the distance-weighted average of the nearest non missing
values. The default number of nearest neighbors is 4.

Parameter
neighbors INTEGER Number of nearest neighbors
newmiss FLOAT New missing value
c FLOAT Constant
rmin FLOAT Lower bound
rmax FLOAT Upper bound

72
Reference manual Modification

Example
setrtomiss
Assume an input dataset has one field with temperatures in the range from 246 to 304 Kelvin. To set
all values below 273.15 Kelvin to missing value use:
cdo setrtomiss,0,273.15 infile outfile

Result of ’cdo info infile’:

−1 : Date Time Code L e v e l Size Miss : Minimum Mean Maximum
1 : 1987−12−31 1 2 : 0 0 : 0 0 139 0 2048 0 : 246.27 276.75 303.71

Result of ’cdo info outfile’:

−1 : Date Time Code L e v e l Size Miss : Minimum Mean Maximum
1 : 1987−12−31 1 2 : 0 0 : 0 0 139 0 2048 871 : 273.16 287.08 303.71

setmisstonn
Set all missing values to the nearest non missing value:
cdo setmisstonn infile outfile

Below is a schematic illustration of this example:

−20˚ 20˚ 60˚ −20˚ 20˚ 60˚
45˚ 45˚

15˚ 15˚

−15˚ −15˚

−45˚ −45˚
−20˚ 20˚ 60˚ −20˚ 20˚ 60˚

On the left side is input data with missing values in grey and on the right side the result with the
filled missing values.

73
Arithmetic Reference manual

2.7. Arithmetic

This section contains modules to arithmetically process datasets.

Here is a short overview of all operators in this section:

expr Evaluate expressions

exprf Evaluate expressions script
aexpr Evaluate expressions and append results
aexprf Evaluate expression script and append results

abs Absolute value

int Integer value
nint Nearest integer value
pow Power
sqr Square
sqrt Square root
exp Exponential
ln Natural logarithm
log10 Base 10 logarithm
sin Sine
cos Cosine
tan Tangent
asin Arc sine
acos Arc cosine
atan Arc tangent
reci Reciprocal value
not Logical NOT

addc Add a constant

subc Subtract a constant
mulc Multiply with a constant
divc Divide by a constant
minc Minimum of a field and a constant
maxc Maximum of a field and a constant

add Add two fields

sub Subtract two fields
mul Multiply two fields
div Divide two fields
min Minimum of two fields
max Maximum of two fields
atan2 Arc tangent of two fields

monadd Add monthly time series

monsub Subtract monthly time series
monmul Multiply monthly time series
mondiv Divide monthly time series

yhouradd Add multi-year hourly time series

yhoursub Subtract multi-year hourly time series
yhourmul Multiply multi-year hourly time series
yhourdiv Divide multi-year hourly time series

74
Reference manual Arithmetic

ydayadd Add multi-year daily time series

ydaysub Subtract multi-year daily time series
ydaymul Multiply multi-year daily time series
ydaydiv Divide multi-year daily time series

ymonadd Add multi-year monthly time series

ymonsub Subtract multi-year monthly time series
ymonmul Multiply multi-year monthly time series
ymondiv Divide multi-year monthly time series

yseasadd Add multi-year seasonal time series

yseassub Subtract multi-year seasonal time series
yseasmul Multiply multi-year seasonal time series
yseasdiv Divide multi-year seasonal time series

muldpm Multiply with days per month

divdpm Divide by days per month
muldpy Multiply with days per year
divdpy Divide by days per year

75
Arithmetic Reference manual

2.7.1. EXPR - Evaluate expressions

Synopsis

expr,instr infile outfile

exprf ,filename infile outfile
aexpr,instr infile outfile
aexprf ,filename infile outfile

Description
This module arithmetically processes every timestep of the input dataset. Each individual assignment
statement have to end with a semi-colon. Unlike regular variables, temporary variables are never
written to the output stream. To define a temporary variable simply prefix the variable name with
an underscore (e.g. _varname) when the variable is declared.
The following operators are supported:

Operator Meaning Example Result

= assignment x=y Assigns y to x
+ addition x+y Sum of x and y
- subtraction x-y Difference of x and y
* multiplication x*y Product of x and y
/ division x/y Quotient of x and y
ˆ exponentiation xˆy Exponentiates x with y
== equal to x == y 1, if x equal to y; else 0
!= not equal to x != y 1, if x not equal to y; else 0
> greater than x>y 1, if x greater than y; else 0
< less than x<y 1, if x less than y; else 0
>= greater equal x >= y 1, if x greater equal y; else 0
<= less equal x <= y 1, if x less equal y; else 0
<=> less equal greater x <=> y -1, if x less y; 1, if x greater y; else 0
&& logical AND x && y 1, if x and y not equal 0; else 0
|| logical OR x || y 1, if x or y not equal 0; else 0
! logical NOT !x 1, if x equal 0; else 0
?: ternary conditional x? y: z y, if x not equal 0, else z

The following functions are supported:

Math intrinsics:

abs(x) Absolute value of x

floor(x) Round to largest integral value not greater than x
ceil(x) Round to smallest integral value not less than x
float(x) 32-bit float value of x
int(x) Integer value of x
nint(x) Nearest integer value of x
sqr(x) Square of x
sqrt(x) Square Root of x
exp(x) Exponential of x

76
Reference manual Arithmetic

ln(x) Natural logarithm of x

log10(x) Base 10 logarithm of x
sin(x) Sine of x, where x is specified in radians
cos(x) Cosine of x, where x is specified in radians
tan(x) Tangent of x, where x is specified in radians
asin(x) Arc-sine of x, where x is specified in radians
acos(x) Arc-cosine of x, where x is specified in radians
atan(x) Arc-tangent of x, where x is specified in radians
rad(x) Convert x from degrees to radians
deg(x) Convert x from radians to degrees
Coordinates:
clon(x) Longitude coordinate of x (available only if x has geographical coordinates)
clat(x) Latitude coordinate of x (available only if x has geographical coordinates)
gridarea(x) Grid cell area of x (available only if x has geographical coordinates)
clev(x) Level coordinate of x (0, if x is a 2D surface variable)
ctimestep() Timestep number (1 to N)
cdate() Verification date as YYYYMMDD
ctime() Verification time as HHMMSS.millisecond
cdeltat() Difference between current and last timestep in seconds
cday() Day as DD
cmonth() Month as MM
cyear() Year as YYYY
csecond() Second as SS.millisecond
cminute() Minute as MM
chour() Hour as HH
Constants:
ngp(x) Number of horizontal grid points
nlev(x) Number of vertical levels
size(x) Total number of elements (ngp(x)*nlev(x))
missval(x) Returns the missing value of variable x
Statistical values over a field:
fldmin(x), fldmax(x), fldsum(x), fldmean(x), fldavg(x), fldstd(x), fldstd1(x), fldvar(x), fldvar1(x)
Zonal statistical values for regular 2D grids:
zonmin(x), zonmax(x), zonsum(x), zonmean(x), zonavg(x), zonstd(x), zonstd1(x), zonvar(x), zon-
var1(x)
Vertical statistical values:
vertmin(x), vertmax(x), vertsum(x), vertmean(x), vertavg(x), vertstd(x), vertstd1(x), vertvar(x),
vertvar1(x)
Miscellaneous:

77
Arithmetic Reference manual

sellevel(x,k) Select level k of variable x

sellevidx(x,k) Select level index k of variable x
sellevelrange(x,k1,k2) Select all levels of variable x in the range k1 to k2
sellevidxrange(x,k1,k2) Select all level indices of variable x in the range k1 to k2
remove(x) Remove variable x from output stream

Operators

expr Evaluate expressions

The processing instructions are read from the parameter.
exprf Evaluate expressions script
Contrary to expr the processing instructions are read from a file.
aexpr Evaluate expressions and append results
Same as expr, but keep input variables and append results
aexprf Evaluate expression script and append results
Same as exprf, but keep input variables and append results

Parameter
instr STRING Processing instructions (need to be ’quoted’ in most cases)
filename STRING File with processing instructions

Note
The expr commands sellevel(x,k) and sellevidx(x,k) are only available with exprf/aexprf. If the input
stream contains duplicate entries of the same variable name then the last one is used.

Example
Assume an input dataset contains at least the variables ’aprl’, ’aprc’ and ’ts’. To create a new variable
’var1’ with the sum of ’aprl’ and ’aprc’ and a variable ’var2’ which convert the temperature ’ts’ from
Kelvin to Celsius use:
cdo expr,’var1=aprl+aprc;var2=ts-273.15;’ infile outfile

The same example, but the instructions are read from a file:
cdo exprf,myexpr infile outfile

The file myexpr contains:

var1 = a p r l + a p r c ;
var2 = t s − 2 7 3 . 1 5 ;

78
Reference manual Arithmetic

2.7.2. MATH - Mathematical functions

Synopsis

< operator > infile outfile

Description
This module contains some standard mathematical functions. All trigonometric functions calculate
with radians.

Operators

abs Absolute value

o(t, x) = abs(i(t, x))
int Integer value
o(t, x) = int(i(t, x))
nint Nearest integer value
o(t, x) = nint(i(t, x))
pow Power
y
o(t, x) = i(t, x)
sqr Square
2
o(t, x) = i(t, x)
sqrt Square root
p
o(t, x) = i(t, x)
exp Exponential
o(t, x) = ei(t,x)
ln Natural logarithm
o(t, x) = ln(i(t, x))
log10 Base 10 logarithm
o(t, x) = log10 (i(t, x))
sin Sine
o(t, x) = sin(i(t, x))
cos Cosine
o(t, x) = cos(i(t, x))
tan Tangent
o(t, x) = tan(i(t, x))
asin Arc sine
o(t, x) = arcsin(i(t, x))
acos Arc cosine
o(t, x) = arccos(i(t, x))
atan Arc tangent
o(t, x) = arctan(i(t, x))
reci Reciprocal value
o(t, x) = 1/i(t, x)
not Logical NOT
o(t, x) = 1, if xequal0; else0

79
Arithmetic Reference manual

Example
To calculate the square root for all field elements use:
cdo sqrt infile outfile

80
Reference manual Arithmetic

2.7.3. ARITHC - Arithmetic with a constant

Synopsis

< operator >,c infile outfile

Description
This module performs simple arithmetic with all field elements of a dataset and a constant. The fields
in outfile inherit the meta data from infile.

Operators

addc Add a constant

o(t, x) = i(t, x) + c
subc Subtract a constant
o(t, x) = i(t, x) − c
mulc Multiply with a constant
o(t, x) = i(t, x) ∗ c
divc Divide by a constant
o(t, x) = i(t, x)/c
minc Minimum of a field and a constant
o(t, x) = min(i(t, x), c)
maxc Maximum of a field and a constant
o(t, x) = max(i(t, x), c)

Parameter
c FLOAT Constant

Example
To sum all input fields with the constant -273.15 use:
cdo addc,-273.15 infile outfile

81
Arithmetic Reference manual

2.7.4. ARITH - Arithmetic on two datasets

Synopsis

< operator > infile1 infile2 outfile

Description
This module performs simple arithmetic of two datasets. The number of fields in infile1 should be
the same as in infile2. The fields in outfile inherit the meta data from infile1. One of the input
files can contain only one timestep or one variable.

Operators

add Add two fields

o(t, x) = i1 (t, x) + i2 (t, x)
sub Subtract two fields
o(t, x) = i1 (t, x) − i2 (t, x)
mul Multiply two fields
o(t, x) = i1 (t, x) ∗ i2 (t, x)
div Divide two fields
o(t, x) = i1 (t, x)/i2 (t, x)
min Minimum of two fields
o(t, x) = min(i1 (t, x), i2 (t, x))
max Maximum of two fields
o(t, x) = max(i1 (t, x), i2 (t, x))
atan2 Arc tangent of two fields
The atan2 operator calculates the arc tangent of two fields. The result is in radians, which
is between -PI and PI (inclusive).
o(t, x) = atan2(i1 (t, x), i2 (t, x))

Example
To sum all fields of the first input file with the corresponding fields of the second input file use:
cdo add infile1 infile2 outfile

82
Reference manual Arithmetic

2.7.5. MONARITH - Monthly arithmetic

Synopsis

< operator > infile1 infile2 outfile

Description
This module performs simple arithmetic of a time series and one timestep with the same month and
year. For each field in infile1 the corresponding field of the timestep in infile2 with the same
month and year is used. The header information in infile1 have to be the same as in infile2.
Usually infile2 is generated by an operator of the module MONSTAT.

Operators

monadd Add monthly time series

Adds a time series and a monthly time series.
monsub Subtract monthly time series
Subtracts a time series and a monthly time series.
monmul Multiply monthly time series
Multiplies a time series and a monthly time series.
mondiv Divide monthly time series
Divides a time series and a monthly time series.

Example
To subtract a monthly time average from a time series use:
cdo monsub infile -monavg infile outfile

83
Arithmetic Reference manual

2.7.6. YHOURARITH - Multi-year hourly arithmetic

Synopsis

< operator > infile1 infile2 outfile

Description
This module performs simple arithmetic of a time series and one timestep with the same hour and
day of year. For each field in infile1 the corresponding field of the timestep in infile2 with the
same hour and day of year is used. The header information in infile1 have to be the same as in
infile2. Usually infile2 is generated by an operator of the module YHOURSTAT.

Operators

yhouradd Add multi-year hourly time series

Adds a time series and a multi-year hourly time series.
yhoursub Subtract multi-year hourly time series
Subtracts a time series and a multi-year hourly time series.
yhourmul Multiply multi-year hourly time series
Multiplies a time series and a multi-year hourly time series.
yhourdiv Divide multi-year hourly time series
Divides a time series and a multi-year hourly time series.

Example
To subtract a multi-year hourly time average from a time series use:
cdo yhoursub infile -yhouravg infile outfile

84
Reference manual Arithmetic

2.7.7. YDAYARITH - Multi-year daily arithmetic

Synopsis

< operator > infile1 infile2 outfile

Description
This module performs simple arithmetic of a time series and one timestep with the same day of year.
For each field in infile1 the corresponding field of the timestep in infile2 with the same day of
year is used. The header information in infile1 have to be the same as in infile2. Usually infile2
is generated by an operator of the module YDAYSTAT.

Operators

ydayadd Add multi-year daily time series

Adds a time series and a multi-year daily time series.
ydaysub Subtract multi-year daily time series
Subtracts a time series and a multi-year daily time series.
ydaymul Multiply multi-year daily time series
Multiplies a time series and a multi-year daily time series.
ydaydiv Divide multi-year daily time series
Divides a time series and a multi-year daily time series.

Example
To subtract a multi-year daily time average from a time series use:
cdo ydaysub infile -ydayavg infile outfile

85
Arithmetic Reference manual

2.7.8. YMONARITH - Multi-year monthly arithmetic

Synopsis

< operator > infile1 infile2 outfile

Description
This module performs simple arithmetic of a time series and one timestep with the same month of
year. For each field in infile1 the corresponding field of the timestep in infile2 with the same
month of year is used. The header information in infile1 have to be the same as in infile2. Usually
infile2 is generated by an operator of the module YMONSTAT.

Operators

ymonadd Add multi-year monthly time series

Adds a time series and a multi-year monthly time series.
ymonsub Subtract multi-year monthly time series
Subtracts a time series and a multi-year monthly time series.
ymonmul Multiply multi-year monthly time series
Multiplies a time series and a multi-year monthly time series.
ymondiv Divide multi-year monthly time series
Divides a time series and a multi-year monthly time series.

Example
To subtract a multi-year monthly time average from a time series use:
cdo ymonsub infile -ymonavg infile outfile

86
Reference manual Arithmetic

2.7.9. YSEASARITH - Multi-year seasonal arithmetic

Synopsis

< operator > infile1 infile2 outfile

Description
This module performs simple arithmetic of a time series and one timestep with the same season. For
each field in infile1 the corresponding field of the timestep in infile2 with the same season is used.
The header information in infile1 have to be the same as in infile2. Usually infile2 is generated
by an operator of the module YSEASSTAT.

Operators

yseasadd Add multi-year seasonal time series

Adds a time series and a multi-year seasonal time series.
yseassub Subtract multi-year seasonal time series
Subtracts a time series and a multi-year seasonal time series.
yseasmul Multiply multi-year seasonal time series
Multiplies a time series and a multi-year seasonal time series.
yseasdiv Divide multi-year seasonal time series
Divides a time series and a multi-year seasonal time series.

Example
To subtract a multi-year seasonal time average from a time series use:
cdo yseassub infile -yseasavg infile outfile

2.7.10. ARITHDAYS - Arithmetic with days

Synopsis

< operator > infile outfile

Description
This module multiplies or divides each timestep of a dataset with the corresponding days per month
or days per year. The result of these functions depends on the used calendar of the input data.

Operators

muldpm Multiply with days per month

o(t, x) = i(t, x) ∗ days_per_month
divdpm Divide by days per month
o(t, x) = i(t, x)/days_per_month
muldpy Multiply with days per year
o(t, x) = i(t, x) ∗ days_per_year
divdpy Divide by days per year
o(t, x) = i(t, x)/days_per_year

87
Statistical values Reference manual

2.8. Statistical values

This section contains modules to compute statistical values of datasets. In this program there is the dif-
ferent notion of "mean" and "average" to distinguish two different kinds of treatment of missing values.
While computing the mean, only the not missing values are considered to belong to the sample with the
side effect of a probably reduced sample size. Computing the average is just adding the sample members
and divide the result by the sample size. For example, the mean of 1, 2, miss and 3 is (1+2+3)/3 = 2,
whereas the average is (1+2+miss+3)/4 = miss/4 = miss. If there are no missing values in the sample,
the average and the mean are identical.
CDO is using the verification time to identify the time range for temporal statistics. The time bounds are
never used!

In this section the abbreviations as in the following table are used:

n
X
sum xi
i=1
n
X
mean resp. avg n−1 xi
i=1
 −1
mean resp. avg n
X n
X
weighted by  wj  w i xi
{wi , i = 1, ..., n} j=1 i=1

n
Variance X
n−1 (xi − x)2
var i=1
n
X
var1 (n − 1)−1 (xi − x)2
i=1
 −1   −1 2
n n n n
var weighted by X X X X
wj  w i  xi −  wj  wj xj 
  
{wi , i = 1, ..., n} j=1 i=1 j=1 j=1

v
u n
Standard deviation u
tn−1
X
(xi − x)2
std i=1
v
u n
X
u
std1 t(n − 1)−1 (xi − x)2
i=1
v  2
u
u n −1  −1
n n n
std weighted by u X X X X
wj  wi xi −  wj  w j xj 
u  
{wi , i = 1, ..., n}
t
j=1 i=1 j=1 j=1

Cumulative Ranked Z ∞
2
Probability Score [H(x1 ) − cdf ({x2 . . . xn })|r ] dr
−∞
crps

with cdf (X)|r being the cumulative distribution function of {xi , i = 2 . . . n} at r

and H(x) the Heavyside function jumping at x.

Here is a short overview of all operators in this section:

88
Reference manual Statistical values

timcumsum Cumulative sum over all timesteps

consecsum Consecutive Sum

consects Consecutive Timesteps

varsmin Variables minimum

varsmax Variables maximum
varsrange Variables range
varssum Variables sum
varsmean Variables mean
varsavg Variables average
varsstd Variables standard deviation
varsstd1 Variables standard deviation (n-1)
varsvar Variables variance
varsvar1 Variables variance (n-1)

ensmin Ensemble minimum

ensmax Ensemble maximum
ensrange Ensemble range
enssum Ensemble sum
ensmean Ensemble mean
ensavg Ensemble average
ensstd Ensemble standard deviation
ensstd1 Ensemble standard deviation (n-1)
ensvar Ensemble variance
ensvar1 Ensemble variance (n-1)
enspctl Ensemble percentiles

ensrkhistspace Ranked Histogram averaged over time

ensrkhisttime Ranked Histogram averaged over space
ensroc Ensemble Receiver Operating characteristics

enscrps Ensemble CRPS and decomposition

ensbrs Ensemble Brier score

fldmin Field minimum

fldmax Field maximum
fldrange Field range
fldsum Field sum
fldmean Field mean
fldavg Field average
fldstd Field standard deviation
fldstd1 Field standard deviation (n-1)
fldvar Field variance
fldvar1 Field variance (n-1)
fldpctl Field percentiles

89
Statistical values Reference manual

zonmin Zonal minimum

zonmax Zonal maximum
zonrange Zonal range
zonsum Zonal sum
zonmean Zonal mean
zonavg Zonal average
zonstd Zonal standard deviation
zonstd1 Zonal standard deviation (n-1)
zonvar Zonal variance
zonvar1 Zonal variance (n-1)
zonpctl Zonal percentiles

mermin Meridional minimum

mermax Meridional maximum
merrange Meridional range
mersum Meridional sum
mermean Meridional mean
meravg Meridional average
merstd Meridional standard deviation
merstd1 Meridional standard deviation (n-1)
mervar Meridional variance
mervar1 Meridional variance (n-1)
merpctl Meridional percentiles

gridboxmin Gridbox minimum

gridboxmax Gridbox maximum
gridboxrange Gridbox range
gridboxsum Gridbox sum
gridboxmean Gridbox mean
gridboxavg Gridbox average
gridboxstd Gridbox standard deviation
gridboxstd1 Gridbox standard deviation (n-1)
gridboxvar Gridbox variance
gridboxvar1 Gridbox variance (n-1)

vertmin Vertical minimum

vertmax Vertical maximum
vertrange Vertical range
vertsum Vertical sum
vertmean Vertical mean
vertavg Vertical average
vertstd Vertical standard deviation
vertstd1 Vertical standard deviation (n-1)
vertvar Vertical variance
vertvar1 Vertical variance (n-1)

90
Reference manual Statistical values

timselmin Time selection minimum

timselmax Time selection maximum
timselrange Time selection range
timselsum Time selection sum
timselmean Time selection mean
timselavg Time selection average
timselstd Time selection standard deviation
timselstd1 Time selection standard deviation (n-1)
timselvar Time selection variance
timselvar1 Time selection variance (n-1)

timselpctl Time range percentiles

runmin Running minimum

runmax Running maximum
runrange Running range
runsum Running sum
runmean Running mean
runavg Running average
runstd Running standard deviation
runstd1 Running standard deviation (n-1)
runvar Running variance
runvar1 Running variance (n-1)

runpctl Running percentiles

timmin Time minimum

timmax Time maximum
timrange Time range
timsum Time sum
timmean Time mean
timavg Time average
timstd Time standard deviation
timstd1 Time standard deviation (n-1)
timvar Time variance
timvar1 Time variance (n-1)

timpctl Time percentiles

hourmin Hourly minimum

hourmax Hourly maximum
hourrange Hourly range
hoursum Hourly sum
hourmean Hourly mean
houravg Hourly average
hourstd Hourly standard deviation
hourstd1 Hourly standard deviation (n-1)
hourvar Hourly variance
hourvar1 Hourly variance (n-1)

hourpctl Hourly percentiles

91
Statistical values Reference manual

daymin Daily minimum

daymax Daily maximum
dayrange Daily range
daysum Daily sum
daymean Daily mean
dayavg Daily average
daystd Daily standard deviation
daystd1 Daily standard deviation (n-1)
dayvar Daily variance
dayvar1 Daily variance (n-1)

daypctl Daily percentiles

monmin Monthly minimum

monmax Monthly maximum
monrange Monthly range
monsum Monthly sum
monmean Monthly mean
monavg Monthly average
monstd Monthly standard deviation
monstd1 Monthly standard deviation (n-1)
monvar Monthly variance
monvar1 Monthly variance (n-1)

monpctl Monthly percentiles

yearmonmean Yearly mean from monthly data

yearmin Yearly minimum

yearmax Yearly maximum
yearrange Yearly range
yearsum Yearly sum
yearmean Yearly mean
yearavg Yearly average
yearstd Yearly standard deviation
yearstd1 Yearly standard deviation (n-1)
yearvar Yearly variance
yearvar1 Yearly variance (n-1)

yearpctl Yearly percentiles

seasmin Seasonal minimum

seasmax Seasonal maximum
seasrange Seasonal range
seassum Seasonal sum
seasmean Seasonal mean
seasavg Seasonal average
seasstd Seasonal standard deviation
seasstd1 Seasonal standard deviation (n-1)
seasvar Seasonal variance
seasvar1 Seasonal variance (n-1)

seaspctl Seasonal percentiles

92
Reference manual Statistical values

yhourmin Multi-year hourly minimum

yhourmax Multi-year hourly maximum
yhourrange Multi-year hourly range
yhoursum Multi-year hourly sum
yhourmean Multi-year hourly mean
yhouravg Multi-year hourly average
yhourstd Multi-year hourly standard deviation
yhourstd1 Multi-year hourly standard deviation (n-1)
yhourvar Multi-year hourly variance
yhourvar1 Multi-year hourly variance (n-1)

ydaymin Multi-year daily minimum

ydaymax Multi-year daily maximum
ydayrange Multi-year daily range
ydaysum Multi-year daily sum
ydaymean Multi-year daily mean
ydayavg Multi-year daily average
ydaystd Multi-year daily standard deviation
ydaystd1 Multi-year daily standard deviation (n-1)
ydayvar Multi-year daily variance
ydayvar1 Multi-year daily variance (n-1)

ydaypctl Multi-year daily percentiles

ymonmin Multi-year monthly minimum

ymonmax Multi-year monthly maximum
ymonrange Multi-year monthly range
ymonsum Multi-year monthly sum
ymonmean Multi-year monthly mean
ymonavg Multi-year monthly average
ymonstd Multi-year monthly standard deviation
ymonstd1 Multi-year monthly standard deviation (n-1)
ymonvar Multi-year monthly variance
ymonvar1 Multi-year monthly variance (n-1)

ymonpctl Multi-year monthly percentiles

yseasmin Multi-year seasonal minimum

yseasmax Multi-year seasonal maximum
yseasrange Multi-year seasonal range
yseassum Multi-year seasonal sum
yseasmean Multi-year seasonal mean
yseasavg Multi-year seasonal average
yseasstd Multi-year seasonal standard deviation
yseasstd1 Multi-year seasonal standard deviation (n-1)
yseasvar Multi-year seasonal variance
yseasvar1 Multi-year seasonal variance (n-1)

yseaspctl Multi-year seasonal percentiles

93
Statistical values Reference manual

ydrunmin Multi-year daily running minimum

ydrunmax Multi-year daily running maximum
ydrunsum Multi-year daily running sum
ydrunmean Multi-year daily running mean
ydrunavg Multi-year daily running average
ydrunstd Multi-year daily running standard deviation
ydrunstd1 Multi-year daily running standard deviation (n-1)
ydrunvar Multi-year daily running variance
ydrunvar1 Multi-year daily running variance (n-1)

ydrunpctl Multi-year daily running percentiles

94
Reference manual Statistical values

2.8.1. TIMCUMSUM - Cumulative sum over all timesteps

Synopsis

timcumsum infile outfile

Description
The timcumsum operator calculates the cumulative sum over all timesteps. Missing values are treated
as numeric zero when summing.

o(t, x) = sum{i(t0 , x), 0 < t0 ≤ t}

2.8.2. CONSECSTAT - Consecute timestep periods

Synopsis

< operator > infile outfile

Description
This module computes periods over all timesteps in infile where a certain property is valid. The
property can be chosen by creating a mask from the original data, which is the expected input format
for operators of this module. Depending on the operator full information about each period or just
its length and ending date are computed.

Operators

consecsum Consecutive Sum

This operator computes periods of consecutive timesteps similar to a runsum, but
periods are finished, when the mask value is 0. That way multiple periods can be
found. Timesteps from the input are preserved. Missing values are handled like 0,
i.e. finish periods of consecutive timesteps.
consects Consecutive Timesteps
In contrast to the operator above consects only computes the length of each period
together with its last timestep. To be able to perform statistical analysis like min,
max or mean, everything else is set to missing value.

Example
For a given time series of daily temperatures, the periods of summer days can be calculated with
inplace maskting the input field:
cdo consects -gtc,20.0 infile1 outfile

95
Statistical values Reference manual

2.8.3. VARSSTAT - Statistical values over all variables

Synopsis

< operator > infile outfile

Description
This module computes statistical values over all variables for each timestep. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation is written to
outfile. All input variables need to have the same gridsize and the same number of levels.

Operators

varsmin Variables minimum

For every timestep the minimum over all variables is computed.
varsmax Variables maximum
For every timestep the maximum over all variables is computed.
varsrange Variables range
For every timestep the range over all variables is computed.
varssum Variables sum
For every timestep the sum over all variables is computed.
varsmean Variables mean
For every timestep the mean over all variables is computed.
varsavg Variables average
For every timestep the average over all variables is computed.
varsstd Variables standard deviation
For every timestep the standard deviation over all variables is computed. Normalize
by n.
varsstd1 Variables standard deviation (n-1)
For every timestep the standard deviation over all variables is computed. Normalize
by (n-1).
varsvar Variables variance
For every timestep the variance over all variables is computed. Normalize by n.
varsvar1 Variables variance (n-1)
For every timestep the variance over all variables is computed. Normalize by (n-1).

96
Reference manual Statistical values

2.8.4. ENSSTAT - Statistical values over an ensemble

Synopsis

< operator > infiles outfile

enspctl,p infiles outfile

Description
This module computes statistical values over an ensemble of input files. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance, standard deviation or a certain
percentile over all input files is written to outfile. All input files need to have the same structure
with the same variables. The date information of a timestep in outfile is the date of the first input
file.

Operators

ensmin Ensemble minimum

o(t, x) = min{i1 (t, x), i2 (t, x), · · · , in (t, x)}
ensmax Ensemble maximum
o(t, x) = max{i1 (t, x), i2 (t, x), · · · , in (t, x)}
ensrange Ensemble range
o(t, x) = range{i1 (t, x), i2 (t, x), · · · , in (t, x)}
enssum Ensemble sum
o(t, x) = sum{i1 (t, x), i2 (t, x), · · · , in (t, x)}
ensmean Ensemble mean
o(t, x) = mean{i1 (t, x), i2 (t, x), · · · , in (t, x)}
ensavg Ensemble average
o(t, x) = avg{i1 (t, x), i2 (t, x), · · · , in (t, x)}
ensstd Ensemble standard deviation
Normalize by n.
o(t, x) = std{i1 (t, x), i2 (t, x), · · · , in (t, x)}
ensstd1 Ensemble standard deviation (n-1)
Normalize by (n-1).
o(t, x) = std1{i1 (t, x), i2 (t, x), · · · , in (t, x)}
ensvar Ensemble variance
Normalize by n.
o(t, x) = var{i1 (t, x), i2 (t, x), · · · , in (t, x)}
ensvar1 Ensemble variance (n-1)
Normalize by (n-1).
o(t, x) = var1{i1 (t, x), i2 (t, x), · · · , in (t, x)}
enspctl Ensemble percentiles
o(t, x) = pth percentile{i1 (t, x), i2 (t, x), · · · , in (t, x)}

Parameter
p FLOAT Percentile number in 0, ..., 100

97
Statistical values Reference manual

Note
This operator needs to open all input files simultaneously. The maximum number of open files depends
on the operating system!

Example
To compute the ensemble mean over 6 input files use:
cdo ensmean infile1 infile2 infile3 infile4 infile5 infile6 outfile

Or shorter with filename substitution:

cdo ensmean infile[1-6] outfile

To compute the 50th percentile (median) over 6 input files use:

cdo enspctl,50 infile1 infile2 infile3 infile4 infile5 infile6 outfile

98
Reference manual Statistical values

2.8.5. ENSSTAT2 - Statistical values over an ensemble

Synopsis

< operator > obsfile ensfiles outfile

Description
This module computes statistical values over the ensemble of ensfiles using obsfile as a reference.
Depending on the operator a ranked Histogram or a roc-curve over all Ensembles ensfiles with
reference to obsfile is written to outfile. The date and grid information of a timestep in outfile
is the date of the first input file. Thus all input files are required to have the same structure in terms
of the gridsize, variable definitions and number of timesteps.
All Operators in this module use obsfile as the reference (for instance an observation) whereas
ensfiles are understood as an ensemble consisting of n (where n is the number of ensfiles) mem-
bers.
The operators ensrkhistspace and ensrkhisttime compute Ranked Histograms. Therefor the vertical
axis is utilized as the Histogram axis, which prohibits the use of files containing more than one level.
The histogram axis has nensfiles+1 bins with level 0 containing for each grid point the number of
observations being smaller as all ensembles and level nensfiles+1 indicating the number of observations
being larger than all ensembles.
ensrkhistspace computes a ranked histogram at each timestep reducing each horizontal grid to a 1x1
grid and keeping the time axis as in obsfile. Contrary ensrkhistspace computes a histogram at
each grid point keeping the horizontal grid for each variable and reducing the time-axis. The time
information is that from the last timestep in obsfile.

Operators

ensrkhistspace Ranked Histogram averaged over time

ensrkhisttime Ranked Histogram averaged over space

ensroc Ensemble Receiver Operating characteristics

Example
To compute a rank histogram over 5 input files ensfile1-ensfile5 given an observation in obsfile
use:
cdo ensrkhisttime obsfile ensfile1 ensfile2 ensfile3 ensfile4 ensfile5 outfile

Or shorter with filename substitution:

cdo ensrkhisttime obsfile ensfile[1-5] outfile

99
Statistical values Reference manual

2.8.6. ENSVAL - Ensemble validation tools

Synopsis

enscrps rfile infiles outfilebase

ensbrs,x rfile infiles outfilebase

Description
This module computes ensemble validation scores and their decomposition such as the Brier and
cumulative ranked probability score (CRPS). The first file is used as a reference it can be a clima-
tology, observation or reanalysis against which the skill of the ensembles given in infiles is measured.
Depending on the operator a number of output files is generated each containing the skill score and
its decomposition corresponding to the operator. The output is averaged over horizontal fields using
appropriate weights for each level and timestep in rfile.
All input files need to have the same structure with the same variables. The date information of a
timestep in outfile is the date of the first input file. The output files are named as <outfilebase>.<type>.<files
where <type> depends on the operator and <filesuffix> is determined from the output file type.
There are three output files for operator enscrps and four output files for operator ensbrs.
The CRPS and its decomposition into Reliability and the potential CRPS are calculated by an
appropriate averaging over the field members (note, that the CRPS does *not* average linearly). In
the three output files <type> has the following meaning: crps for the CRPS, reli for the reliability
and crpspot for the potential crps. The relation CRP S = CRP Spot + RELI
holds.
The Brier score of the Ensemble given by infiles with respect to the reference given in rfile and
the threshold x is calculated. In the four output files <type> has the following meaning: brs for the
Brier score wrt threshold x; brsreli for the Brier score reliability wrt threshold x; brsreso for the
Brier score resolution wrt threshold x; brsunct for the Brier score uncertainty wrt threshold x. In
analogy to the CRPS the following relation holds: BRS(x) = RELI(x) − RESO(x) + U N CT (x).
The implementation of the decomposition of the CRPS and Brier Score follows Hans Hersbach (2000):
Decomposition of the Continuous Ranked Probability Score for Ensemble Prediction Systems, in:
Weather and Forecasting (15) pp. 559-570.
The CRPS code decomposition has been verified against the CRAN - ensemble validation package
from R. Differences occur when grid-cell area is not uniform as the implementation in R does not
account for that.

Operators

enscrps Ensemble CRPS and decomposition

ensbrs Ensemble Brier score

Ensemble Brier Score and Decomposition

Example
To compute the field averaged Brier score at x=5 over an ensemble with 5 members ensfile1-5
w.r.t. the reference rfile and write the results to files obase.brs.<suff>, obase.brsreli<suff>,
obase.brsreso<suff>, obase.brsunct<suff> where <suff> is determined from the output file
type, use
cdo ensbrs,5 rfile ensfile1 ensfile2 ensfile3 ensfile4 ensfile5 obase

100
Reference manual Statistical values

or shorter using file name substitution:

cdo ensbrs,5 rfile ensfile[1-5] obase

101
Statistical values Reference manual

2.8.7. FLDSTAT - Statistical values over a field

Synopsis

< operator >,weights infile outfile

fldpctl,p infile outfile

Description
This module computes statistical values of the input fields. According to the chosen operator the
field minimum, maximum, range, sum, average, variance, standard deviation or a certain percentile
is written to outfile.

Operators

fldmin Field minimum

For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = min{i(t, x0 ), x1 < x0 ≤ xn }
fldmax Field maximum
For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = max{i(t, x0 ), x1 < x0 ≤ xn }
fldrange Field range
For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = range{i(t, x0 ), x1 < x0 ≤ xn }
fldsum Field sum
For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = sum{i(t, x0 ), x1 < x0 ≤ xn }
fldmean Field mean
For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = mean{i(t, x0 ), x1 < x0 ≤ xn }
weighted by area weights obtained by the input field.
fldavg Field average
For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = avg{i(t, x0 ), x1 < x0 ≤ xn }
weighted by area weights obtained by the input field.
fldstd Field standard deviation
Normalize by n. For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = std{i(t, x0 ), x1 < x0 ≤ xn }
weighted by area weights obtained by the input field.
fldstd1 Field standard deviation (n-1)
Normalize by (n-1). For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = std1{i(t, x0 ), x1 < x0 ≤ xn }
weighted by area weights obtained by the input field.
fldvar Field variance
Normalize by n. For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = var{i(t, x0 ), x1 < x0 ≤ xn }
weighted by area weights obtained by the input field.

102
Reference manual Statistical values

fldvar1 Field variance (n-1)

Normalize by (n-1). For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = var1{i(t, x0 ), x1 < x0 ≤ xn }
weighted by area weights obtained by the input field.
fldpctl Field percentiles
For every gridpoint x_1, ..., x_n of the same field it is:
o(t, 1) = pth percentile{i(t, x0 ), x1 < x0 ≤ xn }

Parameter
weights BOOL weights=FALSE disables weighting by grid cell area [default: weights=TRUE]
p FLOAT Percentile number in 0, ..., 100

Example
To compute the field mean of all input fields use:
cdo fldmean infile outfile

To compute the 90th percentile of all input fields use:

cdo fldpctl,90 infile outfile

103
Statistical values Reference manual

2.8.8. ZONSTAT - Zonal statistical values

Synopsis
< operator > infile outfile
zonpctl,p infile outfile

Description
This module computes zonal statistical values of the input fields. According to the chosen operator the
zonal minimum, maximum, range, sum, average, variance, standard deviation or a certain percentile
is written to outfile. This operator requires all variables on the same regular lon/lat grid.

Operators
zonmin Zonal minimum
For every latitude the minimum over all longitudes is computed.
zonmax Zonal maximum
For every latitude the maximum over all longitudes is computed.
zonrange Zonal range
For every latitude the range over all longitudes is computed.
zonsum Zonal sum
For every latitude the sum over all longitudes is computed.
zonmean Zonal mean
For every latitude the mean over all longitudes is computed.
zonavg Zonal average
For every latitude the average over all longitudes is computed.
zonstd Zonal standard deviation
For every latitude the standard deviation over all longitudes is computed. Normalize
by n.
zonstd1 Zonal standard deviation (n-1)
For every latitude the standard deviation over all longitudes is computed. Normalize
by (n-1).
zonvar Zonal variance
For every latitude the variance over all longitudes is computed. Normalize by n.
zonvar1 Zonal variance (n-1)
For every latitude the variance over all longitudes is computed. Normalize by (n-1).
zonpctl Zonal percentiles
For every latitude the pth percentile over all longitudes is computed.

Parameter
p FLOAT Percentile number in 0, ..., 100

Example
To compute the zonal mean of all input fields use:
cdo zonmean infile outfile

To compute the 50th meridional percentile (median) of all input fields use:
cdo zonpctl,50 infile outfile

104
Reference manual Statistical values

2.8.9. MERSTAT - Meridional statistical values

Synopsis

< operator > infile outfile

merpctl,p infile outfile

Description
This module computes meridional statistical values of the input fields. According to the chosen
operator the meridional minimum, maximum, range, sum, average, variance, standard deviation or
a certain percentile is written to outfile. This operator requires all variables on the same regular
lon/lat grid.

Operators

mermin Meridional minimum

For every longitude the minimum over all latitudes is computed.

mermax Meridional maximum

For every longitude the maximum over all latitudes is computed.

merrange Meridional range

For every longitude the range over all latitudes is computed.

mersum Meridional sum

For every longitude the sum over all latitudes is computed.

mermean Meridional mean

For every longitude the area weighted mean over all latitudes is computed.

meravg Meridional average

For every longitude the area weighted average over all latitudes is computed.

merstd Meridional standard deviation

For every longitude the standard deviation over all latitudes is computed. Normalize
by n.

merstd1 Meridional standard deviation (n-1)

For every longitude the standard deviation over all latitudes is computed. Normalize
by (n-1).

mervar Meridional variance

For every longitude the variance over all latitudes is computed. Normalize by n.

mervar1 Meridional variance (n-1)

For every longitude the variance over all latitudes is computed. Normalize by (n-1).

merpctl Meridional percentiles

For every longitude the pth percentile over all latitudes is computed.

Parameter
p FLOAT Percentile number in 0, ..., 100

105
Statistical values Reference manual

Example
To compute the meridional mean of all input fields use:
cdo mermean infile outfile

To compute the 50th meridional percentile (median) of all input fields use:
cdo merpctl,50 infile outfile

106
Reference manual Statistical values

2.8.10. GRIDBOXSTAT - Statistical values over grid boxes

Synopsis

< operator >,nx,ny infile outfile

Description
This module computes statistical values over surrounding grid boxes. According to the chosen operator
the minimum, maximum, range, sum, average, variance, or standard deviation of the neighboring grid
boxes is written to outfile. All gridbox operators only works on quadrilateral curvilinear grids.

Operators

gridboxmin Gridbox minimum

Minimum value of the selected grid boxes.
gridboxmax Gridbox maximum
Maximum value of the selected grid boxes.
gridboxrange Gridbox range
Range (max-min value) of the selected grid boxes.
gridboxsum Gridbox sum
Sum of the selected grid boxes.
gridboxmean Gridbox mean
Mean of the selected grid boxes.
gridboxavg Gridbox average
Average of the selected grid boxes.
gridboxstd Gridbox standard deviation
Standard deviation of the selected grid boxes. Normalize by n.
gridboxstd1 Gridbox standard deviation (n-1)
Standard deviation of the selected grid boxes. Normalize by (n-1).
gridboxvar Gridbox variance
Variance of the selected grid boxes. Normalize by n.
gridboxvar1 Gridbox variance (n-1)
Variance of the selected grid boxes. Normalize by (n-1).

Parameter
nx INTEGER Number of grid boxes in x direction
ny INTEGER Number of grid boxes in y direction

Example
To compute the mean over 10x10 grid boxes of the input field use:
cdo gridboxmean,10,10 infile outfile

107
Statistical values Reference manual

2.8.11. VERTSTAT - Vertical statistical values

Synopsis

< operator >,weights infile outfile

Description
This module computes statistical values over all levels of the input variables. According to chosen
operator the vertical minimum, maximum, range, sum, average, variance or standard deviation is
written to outfile.

Operators

vertmin Vertical minimum

For every gridpoint the minimum over all levels is computed.
vertmax Vertical maximum
For every gridpoint the maximum over all levels is computed.
vertrange Vertical range
For every gridpoint the range over all levels is computed.
vertsum Vertical sum
For every gridpoint the sum over all levels is computed.
vertmean Vertical mean
For every gridpoint the layer weighted mean over all levels is computed.
vertavg Vertical average
For every gridpoint the layer weighted average over all levels is computed.
vertstd Vertical standard deviation
For every gridpoint the standard deviation over all levels is computed. Normalize by
n.
vertstd1 Vertical standard deviation (n-1)
For every gridpoint the standard deviation over all levels is computed. Normalize by
(n-1).
vertvar Vertical variance
For every gridpoint the variance over all levels is computed. Normalize by n.
vertvar1 Vertical variance (n-1)
For every gridpoint the variance over all levels is computed. Normalize by (n-1).

Parameter
weights BOOL weights=FALSE disables weighting by layer thickness [default: weights=TRUE]

Example
To compute the vertical sum of all input variables use:
cdo vertsum infile outfile

108
Reference manual Statistical values

2.8.12. TIMSELSTAT - Time range statistical values

Synopsis

< operator >,nsets[,noffset[,nskip]] infile outfile

Description
This module computes statistical values for a selected number of timesteps. According to the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation of the selected
timesteps is written to outfile. The time of outfile is determined by the time in the middle of all
contributing timesteps of infile.

Operators

timselmin Time selection minimum

For every adjacent sequence t_1, ..., t_n of timesteps of the same selected time
range it is:
o(t, x) = min{i(t0 , x), t1 < t0 ≤ tn }
timselmax Time selection maximum
For every adjacent sequence t_1, ..., t_n of timesteps of the same selected time
range it is:
o(t, x) = max{i(t0 , x), t1 < t0 ≤ tn }
timselrange Time selection range
For every adjacent sequence t_1, ..., t_n of timesteps of the same selected time
range it is:
o(t, x) = range{i(t0 , x), t1 < t0 ≤ tn }
timselsum Time selection sum
For every adjacent sequence t_1, ..., t_n of timesteps of the same selected time
range it is:
o(t, x) = sum{i(t0 , x), t1 < t0 ≤ tn }
timselmean Time selection mean
For every adjacent sequence t_1, ..., t_n of timesteps of the same selected time
range it is:
o(t, x) = mean{i(t0 , x), t1 < t0 ≤ tn }
timselavg Time selection average
For every adjacent sequence t_1, ..., t_n of timesteps of the same selected time
range it is:
o(t, x) = avg{i(t0 , x), t1 < t0 ≤ tn }
timselstd Time selection standard deviation
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
selected time range it is:
o(t, x) = std{i(t0 , x), t1 < t0 ≤ tn }
timselstd1 Time selection standard deviation (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the
same selected time range it is:
o(t, x) = std1{i(t0 , x), t1 < t0 ≤ tn }
timselvar Time selection variance
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
selected time range it is:
o(t, x) = var{i(t0 , x), t1 < t0 ≤ tn }

109
Statistical values Reference manual

timselvar1 Time selection variance (n-1)

Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the
same selected time range it is:
o(t, x) = var1{i(t0 , x), t1 < t0 ≤ tn }

Parameter
nsets INTEGER Number of input timesteps for each output timestep
noffset INTEGER Number of input timesteps skipped before the first timestep range (optional)
nskip INTEGER Number of input timesteps skipped between timestep ranges (optional)

Example
Assume an input dataset has monthly means over several years. To compute seasonal means from
monthly means the first two month have to be skipped:
cdo timselmean,3,2 infile outfile

2.8.13. TIMSELPCTL - Time range percentile values

Synopsis

timselpctl,p,nsets[,noffset[,nskip]] infile1 infile2 infile3 outfile

Description
This operator computes percentile values over a selected number of timesteps in infile1. The
algorithm uses histograms with minimum and maximum bounds given in infile2 and infile3,
respectively. The default number of histogram bins is 101. The default can be overridden by setting
the environment variable CDO_PCTL_NBINS to a different value. The files infile2 and infile3 should
be the result of corresponding timselmin and timselmax operations, respectively. The time of outfile
is determined by the time in the middle of all contributing timesteps of infile1.
For every adjacent sequence t_1, ..., t_n of timesteps of the same selected time range it is:

o(t, x) = pth percentile{i(t0 , x), t1 < t0 ≤ tn }

Parameter
p FLOAT Percentile number in 0, ..., 100
nsets INTEGER Number of input timesteps for each output timestep
noffset INTEGER Number of input timesteps skipped before the first timestep range (optional)
nskip INTEGER Number of input timesteps skipped between timestep ranges (optional)

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

110
Reference manual Statistical values

2.8.14. RUNSTAT - Running statistical values

Synopsis

< operator >,nts infile outfile

Description
This module computes running statistical values over a selected number of timesteps. Depending on
the chosen operator the minimum, maximum, range, sum, average, variance or standard deviation
of a selected number of consecutive timesteps read from infile is written to outfile. The time of
outfile is determined by the time in the middle of all contributing timesteps of infile.

Operators

runmin Running minimum

o(t + (nts − 1)/2, x) = min{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runmax Running maximum
o(t + (nts − 1)/2, x) = max{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runrange Running range
o(t + (nts − 1)/2, x) = range{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runsum Running sum
o(t + (nts − 1)/2, x) = sum{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runmean Running mean
o(t + (nts − 1)/2, x) = mean{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runavg Running average
o(t + (nts − 1)/2, x) = avg{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runstd Running standard deviation
Normalize by n.
o(t + (nts − 1)/2, x) = std{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runstd1 Running standard deviation (n-1)
Normalize by (n-1).
o(t + (nts − 1)/2, x) = std1{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runvar Running variance
Normalize by n.
o(t + (nts − 1)/2, x) = var{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}
runvar1 Running variance (n-1)
Normalize by (n-1).
o(t + (nts − 1)/2, x) = var1{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}

Parameter
nts INTEGER Number of timesteps

Environment
CDO_TIMESTAT_DATE Sets the time stamp in outfile to the "first", "middle" or "last" contributing
timestep of infile.

111
Statistical values Reference manual

Example
To compute the running mean over 9 timesteps use:
cdo runmean,9 infile outfile

2.8.15. RUNPCTL - Running percentile values

Synopsis

runpctl,p,nts infile outfile

Description
This module computes running percentiles over a selected number of timesteps in infile. The time
of outfile is determined by the time in the middle of all contributing timesteps of infile.

o(t + (nts − 1)/2, x) = pth percentile{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x)}

Parameter
p FLOAT Percentile number in 0, ..., 100
nts INTEGER Number of timesteps

Example
To compute the running 50th percentile (median) over 9 timesteps use:
cdo runpctl,50,9 infile outfile

112
Reference manual Statistical values

2.8.16. TIMSTAT - Statistical values over all timesteps

Synopsis

< operator > infile outfile

Description
This module computes statistical values over all timesteps in infile. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation of all timesteps
read from infile is written to outfile. The time of outfile is determined by the time in the middle
of all contributing timesteps of infile.

Operators

timmin Time minimum

o(1, x) = min{i(t0 , x), t1 < t0 ≤ tn }
timmax Time maximum
o(1, x) = max{i(t0 , x), t1 < t0 ≤ tn }
timrange Time range
o(1, x) = range{i(t0 , x), t1 < t0 ≤ tn }
timsum Time sum
o(1, x) = sum{i(t0 , x), t1 < t0 ≤ tn }
timmean Time mean
o(1, x) = mean{i(t0 , x), t1 < t0 ≤ tn }
timavg Time average
o(1, x) = avg{i(t0 , x), t1 < t0 ≤ tn }
timstd Time standard deviation
Normalize by n.
o(1, x) = std{i(t0 , x), t1 < t0 ≤ tn }
timstd1 Time standard deviation (n-1)
Normalize by (n-1).
o(1, x) = std1{i(t0 , x), t1 < t0 ≤ tn }
timvar Time variance
Normalize by n.
o(1, x) = var{i(t0 , x), t1 < t0 ≤ tn }
timvar1 Time variance (n-1)
Normalize by (n-1).
o(1, x) = var1{i(t0 , x), t1 < t0 ≤ tn }

Example
To compute the mean over all input timesteps use:
cdo timmean infile outfile

113
Statistical values Reference manual

2.8.17. TIMPCTL - Percentile values over all timesteps

Synopsis

timpctl,p infile1 infile2 infile3 outfile

Description
This operator computes percentiles over all timesteps in infile1. The algorithm uses histograms
with minimum and maximum bounds given in infile2 and infile3, respectively. The default
number of histogram bins is 101. The default can be overridden by defining the environment variable
CDO_PCTL_NBINS. The files infile2 and infile3 should be the result of corresponding timmin and
timmax operations, respectively. The time of outfile is determined by the time in the middle of all
contributing timesteps of infile1.

o(1, x) = pth percentile{i(t0 , x), t1 < t0 ≤ tn }

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the 90th percentile over all input timesteps use:
cdo timmin infile minfile
cdo timmax infile maxfile
cdo timpctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo timpctl,90 infile -timmin infile -timmax infile outfile

114
Reference manual Statistical values

2.8.18. HOURSTAT - Hourly statistical values

Synopsis

< operator > infile outfile

Description
This module computes statistical values over timesteps of the same hour. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation of timesteps
of the same hour is written to outfile. The time of outfile is determined by the time in the middle
of all contributing timesteps of infile.

Operators

hourmin Hourly minimum

For every adjacent sequence t_1, ..., t_n of timesteps of the same hour it is:
o(t, x) = min{i(t0 , x), t1 < t0 ≤ tn }
hourmax Hourly maximum
For every adjacent sequence t_1, ..., t_n of timesteps of the same hour it is:
o(t, x) = max{i(t0 , x), t1 < t0 ≤ tn }
hourrange Hourly range
For every adjacent sequence t_1, ..., t_n of timesteps of the same hour it is:
o(t, x) = range{i(t0 , x), t1 < t0 ≤ tn }
hoursum Hourly sum
For every adjacent sequence t_1, ..., t_n of timesteps of the same hour it is:
o(t, x) = sum{i(t0 , x), t1 < t0 ≤ tn }
hourmean Hourly mean
For every adjacent sequence t_1, ..., t_n of timesteps of the same hour it is:
o(t, x) = mean{i(t0 , x), t1 < t0 ≤ tn }
houravg Hourly average
For every adjacent sequence t_1, ..., t_n of timesteps of the same hour it is:
o(t, x) = avg{i(t0 , x), t1 < t0 ≤ tn }
hourstd Hourly standard deviation
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
hour it is:
o(t, x) = std{i(t0 , x), t1 < t0 ≤ tn }
hourstd1 Hourly standard deviation (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
hour it is:
o(t, x) = std1{i(t0 , x), t1 < t0 ≤ tn }
hourvar Hourly variance
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
hour it is:
o(t, x) = var{i(t0 , x), t1 < t0 ≤ tn }
hourvar1 Hourly variance (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
hour it is:
o(t, x) = var1{i(t0 , x), t1 < t0 ≤ tn }

115
Statistical values Reference manual

Example
To compute the hourly mean of a time series use:
cdo hourmean infile outfile

2.8.19. HOURPCTL - Hourly percentile values

Synopsis

hourpctl,p infile1 infile2 infile3 outfile

Description
This operator computes percentiles over all timesteps of the same hour in infile1. The algorithm uses
histograms with minimum and maximum bounds given in infile2 and infile3, respectively. The
default number of histogram bins is 101. The default can be overridden by defining the environment
variable CDO_PCTL_NBINS. The files infile2 and infile3 should be the result of corresponding
hourmin and hourmax operations, respectively. The time of outfile is determined by the time in
the middle of all contributing timesteps of infile1.
For every adjacent sequence t_1, ..., t_n of timesteps of the same hour it is:

o(t, x) = pth percentile{i(t0 , x), t1 < t0 ≤ tn }

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the hourly 90th percentile of a time series use:
cdo hourmin infile minfile
cdo hourmax infile maxfile
cdo hourpctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo hourpctl,90 infile -hourmin infile -hourmax infile outfile

116
Reference manual Statistical values

2.8.20. DAYSTAT - Daily statistical values

Synopsis

< operator > infile outfile

Description
This module computes statistical values over timesteps of the same day. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation of timesteps
of the same day is written to outfile. The time of outfile is determined by the time in the middle
of all contributing timesteps of infile.

Operators

daymin Daily minimum

For every adjacent sequence t_1, ..., t_n of timesteps of the same day it is:
o(t, x) = min{i(t0 , x), t1 < t0 ≤ tn }
daymax Daily maximum
For every adjacent sequence t_1, ..., t_n of timesteps of the same day it is:
o(t, x) = max{i(t0 , x), t1 < t0 ≤ tn }
dayrange Daily range
For every adjacent sequence t_1, ..., t_n of timesteps of the same day it is:
o(t, x) = range{i(t0 , x), t1 < t0 ≤ tn }
daysum Daily sum
For every adjacent sequence t_1, ..., t_n of timesteps of the same day it is:
o(t, x) = sum{i(t0 , x), t1 < t0 ≤ tn }
daymean Daily mean
For every adjacent sequence t_1, ..., t_n of timesteps of the same day it is:
o(t, x) = mean{i(t0 , x), t1 < t0 ≤ tn }
dayavg Daily average
For every adjacent sequence t_1, ..., t_n of timesteps of the same day it is:
o(t, x) = avg{i(t0 , x), t1 < t0 ≤ tn }
daystd Daily standard deviation
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same day
it is:
o(t, x) = std{i(t0 , x), t1 < t0 ≤ tn }
daystd1 Daily standard deviation (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
day it is:
o(t, x) = std1{i(t0 , x), t1 < t0 ≤ tn }
dayvar Daily variance
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same day
it is:
o(t, x) = var{i(t0 , x), t1 < t0 ≤ tn }
dayvar1 Daily variance (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
day it is:
o(t, x) = var1{i(t0 , x), t1 < t0 ≤ tn }

117
Statistical values Reference manual

Example
To compute the daily mean of a time series use:
cdo daymean infile outfile

2.8.21. DAYPCTL - Daily percentile values

Synopsis

daypctl,p infile1 infile2 infile3 outfile

Description
This operator computes percentiles over all timesteps of the same day in infile1. The algorithm uses
histograms with minimum and maximum bounds given in infile2 and infile3, respectively. The
default number of histogram bins is 101. The default can be overridden by defining the environment
variable CDO_PCTL_NBINS. The files infile2 and infile3 should be the result of corresponding
daymin and daymax operations, respectively. The time of outfile is determined by the time in the
middle of all contributing timesteps of infile1.
For every adjacent sequence t_1, ..., t_n of timesteps of the same day it is:

o(t, x) = pth percentile{i(t0 , x), t1 < t0 ≤ tn }

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the daily 90th percentile of a time series use:
cdo daymin infile minfile
cdo daymax infile maxfile
cdo daypctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo daypctl,90 infile -daymin infile -daymax infile outfile

118
Reference manual Statistical values

2.8.22. MONSTAT - Monthly statistical values

Synopsis

< operator > infile outfile

Description
This module computes statistical values over timesteps of the same month. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation of timesteps
of the same month is written to outfile. The time of outfile is determined by the time in the
middle of all contributing timesteps of infile.

Operators

monmin Monthly minimum

For every adjacent sequence t_1, ..., t_n of timesteps of the same month it is:
o(t, x) = min{i(t0 , x), t1 < t0 ≤ tn }
monmax Monthly maximum
For every adjacent sequence t_1, ..., t_n of timesteps of the same month it is:
o(t, x) = max{i(t0 , x), t1 < t0 ≤ tn }
monrange Monthly range
For every adjacent sequence t_1, ..., t_n of timesteps of the same month it is:
o(t, x) = range{i(t0 , x), t1 < t0 ≤ tn }
monsum Monthly sum
For every adjacent sequence t_1, ..., t_n of timesteps of the same month it is:
o(t, x) = sum{i(t0 , x), t1 < t0 ≤ tn }
monmean Monthly mean
For every adjacent sequence t_1, ..., t_n of timesteps of the same month it is:
o(t, x) = mean{i(t0 , x), t1 < t0 ≤ tn }
monavg Monthly average
For every adjacent sequence t_1, ..., t_n of timesteps of the same month it is:
o(t, x) = avg{i(t0 , x), t1 < t0 ≤ tn }
monstd Monthly standard deviation
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
month it is:
o(t, x) = std{i(t0 , x), t1 < t0 ≤ tn }
monstd1 Monthly standard deviation (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
month it is:
o(t, x) = std1{i(t0 , x), t1 < t0 ≤ tn }
monvar Monthly variance
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
month it is:
o(t, x) = var{i(t0 , x), t1 < t0 ≤ tn }
monvar1 Monthly variance (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
month it is:
o(t, x) = var1{i(t0 , x), t1 < t0 ≤ tn }

119
Statistical values Reference manual

Example
To compute the monthly mean of a time series use:
cdo monmean infile outfile

2.8.23. MONPCTL - Monthly percentile values

Synopsis

monpctl,p infile1 infile2 infile3 outfile

Description
This operator computes percentiles over all timesteps of the same month in infile1. The algorithm
uses histograms with minimum and maximum bounds given in infile2 and infile3, respectively.
The default number of histogram bins is 101. The default can be overridden by defining the environ-
ment variable CDO_PCTL_NBINS. The files infile2 and infile3 should be the result of corresponding
monmin and monmax operations, respectively. The time of outfile is determined by the time in the
middle of all contributing timesteps of infile1.
For every adjacent sequence t_1, ..., t_n of timesteps of the same month it is:

o(t, x) = pth percentile{i(t0 , x), t1 < t0 ≤ tn }

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the monthly 90th percentile of a time series use:
cdo monmin infile minfile
cdo monmax infile maxfile
cdo monpctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo monpctl,90 infile -monmin infile -monmax infile outfile

120
Reference manual Statistical values

2.8.24. YEARMONSTAT - Yearly mean from monthly data

Synopsis

yearmonmean infile outfile

Description
This operator computes the yearly mean of a monthly time series. Each month is weighted with the
number of days per month. The time of outfile is determined by the time in the middle of all
contributing timesteps of infile.
For every adjacent sequence t_1, ..., t_n of timesteps of the same year it is:
o(t, x) = mean{i(t0 , x), t1 < t0 ≤ tn }

Environment
CDO_TIMESTAT_DATE Sets the date information in outfile to the "first", "middle" or "last" contribut-
ing timestep of infile.

Example
To compute the yearly mean of a monthly time series use:
cdo yearmonmean infile outfile

121
Statistical values Reference manual

2.8.25. YEARSTAT - Yearly statistical values

Synopsis
< operator > infile outfile

Description
This module computes statistical values over timesteps of the same year. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation of timesteps
of the same year is written to outfile. The time of outfile is determined by the time in the middle
of all contributing timesteps of infile.

Operators
yearmin Yearly minimum
For every adjacent sequence t_1, ..., t_n of timesteps of the same year it is:
o(t, x) = min{i(t0 , x), t1 < t0 ≤ tn }
yearmax Yearly maximum
For every adjacent sequence t_1, ..., t_n of timesteps of the same year it is:
o(t, x) = max{i(t0 , x), t1 < t0 ≤ tn }
yearrange Yearly range
For every adjacent sequence t_1, ..., t_n of timesteps of the same year it is:
o(t, x) = range{i(t0 , x), t1 < t0 ≤ tn }
yearsum Yearly sum
For every adjacent sequence t_1, ..., t_n of timesteps of the same year it is:
o(t, x) = sum{i(t0 , x), t1 < t0 ≤ tn }
yearmean Yearly mean
For every adjacent sequence t_1, ..., t_n of timesteps of the same year it is:
o(t, x) = mean{i(t0 , x), t1 < t0 ≤ tn }
yearavg Yearly average
For every adjacent sequence t_1, ..., t_n of timesteps of the same year it is:
o(t, x) = avg{i(t0 , x), t1 < t0 ≤ tn }
yearstd Yearly standard deviation
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
year it is:
o(t, x) = std{i(t0 , x), t1 < t0 ≤ tn }
yearstd1 Yearly standard deviation (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
year it is:
o(t, x) = std1{i(t0 , x), t1 < t0 ≤ tn }
yearvar Yearly variance
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
year it is:
o(t, x) = var{i(t0 , x), t1 < t0 ≤ tn }
yearvar1 Yearly variance (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
year it is:
o(t, x) = var1{i(t0 , x), t1 < t0 ≤ tn }

Note
The operators yearmean and yearavg compute only arithmetical means!

122
Reference manual Statistical values

Example
To compute the yearly mean of a time series use:
cdo yearmean infile outfile

To compute the yearly mean from the correct weighted monthly mean use:
cdo yearmonmean infile outfile

2.8.26. YEARPCTL - Yearly percentile values

Synopsis

yearpctl,p infile1 infile2 infile3 outfile

Description
This operator computes percentiles over all timesteps of the same year in infile1. The algorithm uses
histograms with minimum and maximum bounds given in infile2 and infile3, respectively. The
default number of histogram bins is 101. The default can be overridden by defining the environment
variable CDO_PCTL_NBINS. The files infile2 and infile3 should be the result of corresponding
yearmin and yearmax operations, respectively. The time of outfile is determined by the time in the
middle of all contributing timesteps of infile1.
For every adjacent sequence t_1, ..., t_n of timesteps of the same year it is:

o(t, x) = pth percentile{i(t0 , x), t1 < t0 ≤ tn }

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the yearly 90th percentile of a time series use:
cdo yearmin infile minfile
cdo yearmax infile maxfile
cdo yearpctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo yearpctl,90 infile -yearmin infile -yearmax infile outfile

123
Statistical values Reference manual

2.8.27. SEASSTAT - Seasonal statistical values

Synopsis

< operator > infile outfile

Description
This module computes statistical values over timesteps of the same season. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation of timesteps of
the same season is written to outfile. The time of outfile is determined by the time in the middle
of all contributing timesteps of infile. Be careful about the first and the last output timestep, they
may be incorrect values if the seasons have incomplete timesteps.

Operators

seasmin Seasonal minimum

For every adjacent sequence t_1, ..., t_n of timesteps of the same season it is:
o(t, x) = min{i(t0 , x), t1 < t0 ≤ tn }
seasmax Seasonal maximum
For every adjacent sequence t_1, ..., t_n of timesteps of the same season it is:
o(t, x) = max{i(t0 , x), t1 < t0 ≤ tn }
seasrange Seasonal range
For every adjacent sequence t_1, ..., t_n of timesteps of the same season it is:
o(t, x) = range{i(t0 , x), t1 < t0 ≤ tn }
seassum Seasonal sum
For every adjacent sequence t_1, ..., t_n of timesteps of the same season it is:
o(t, x) = sum{i(t0 , x), t1 < t0 ≤ tn }
seasmean Seasonal mean
For every adjacent sequence t_1, ..., t_n of timesteps of the same season it is:
o(t, x) = mean{i(t0 , x), t1 < t0 ≤ tn }
seasavg Seasonal average
For every adjacent sequence t_1, ..., t_n of timesteps of the same season it is:
o(t, x) = avg{i(t0 , x), t1 < t0 ≤ tn }
seasstd Seasonal standard deviation
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
season it is:
o(t, x) = std{i(t0 , x), t1 < t0 ≤ tn }
seasstd1 Seasonal standard deviation (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
season it is:
o(t, x) = std1{i(t0 , x), t1 < t0 ≤ tn }
seasvar Seasonal variance
Normalize by n. For every adjacent sequence t_1, ..., t_n of timesteps of the same
season it is:
o(t, x) = var{i(t0 , x), t1 < t0 ≤ tn }
seasvar1 Seasonal variance (n-1)
Normalize by (n-1). For every adjacent sequence t_1, ..., t_n of timesteps of the same
season it is:
o(t, x) = var1{i(t0 , x), t1 < t0 ≤ tn }

124
Reference manual Statistical values

Example
To compute the seasonal mean of a time series use:
cdo seasmean infile outfile

2.8.28. SEASPCTL - Seasonal percentile values

Synopsis

seaspctl,p infile1 infile2 infile3 outfile

Description
This operator computes percentiles over all timesteps in infile1 of the same season. The algorithm
uses histograms with minimum and maximum bounds given in infile2 and infile3, respectively.
The default number of histogram bins is 101. The default can be overridden by defining the environ-
ment variable CDO_PCTL_NBINS. The files infile2 and infile3 should be the result of corresponding
seasmin and seasmax operations, respectively. The time of outfile is determined by the time in
the middle of all contributing timesteps of infile1. Be careful about the first and the last output
timestep, they may be incorrect values if the seasons have incomplete timesteps.
For every adjacent sequence t_1, ..., t_n of timesteps of the same season it is:

o(t, x) = pth percentile{i(t0 , x), t1 < t0 ≤ tn }

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the seasonal 90th percentile of a time series use:
cdo seasmin infile minfile
cdo seasmax infile maxfile
cdo seaspctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo seaspctl,90 infile -seasmin infile -seasmax infile outfile

125
Statistical values Reference manual

2.8.29. YHOURSTAT - Multi-year hourly statistical values

Synopsis

< operator > infile outfile

Description
This module computes statistical values of each hour and day of year. Depending on the chosen
operator the minimum, maximum, range, sum, average, variance or standard deviation of each hour
and day of year in infile is written to outfile. The date information in an output field is the date
of the last contributing input field.

Operators

yhourmin Multi-year hourly minimum

o(0001, x) = min{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = min{i(t, x), day(i(t)) = 8784}
yhourmax Multi-year hourly maximum
o(0001, x) = max{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = max{i(t, x), day(i(t)) = 8784}
yhourrange Multi-year hourly range
o(0001, x) = range{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = range{i(t, x), day(i(t)) = 8784}
yhoursum Multi-year hourly sum
o(0001, x) = sum{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = sum{i(t, x), day(i(t)) = 8784}
yhourmean Multi-year hourly mean
o(0001, x) = mean{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = mean{i(t, x), day(i(t)) = 8784}
yhouravg Multi-year hourly average
o(0001, x) = avg{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = avg{i(t, x), day(i(t)) = 8784}
yhourstd Multi-year hourly standard deviation
Normalize by n.
o(0001, x) = std{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = std{i(t, x), day(i(t)) = 8784}
yhourstd1 Multi-year hourly standard deviation (n-1)
Normalize by (n-1).

126
Reference manual Statistical values

o(0001, x) = std1{i(t, x), day(i(t)) = 0001}

..
.
o(8784, x) = std1{i(t, x), day(i(t)) = 8784}
yhourvar Multi-year hourly variance
Normalize by n.
o(0001, x) = var{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = var{i(t, x), day(i(t)) = 8784}
yhourvar1 Multi-year hourly variance (n-1)
Normalize by (n-1).
o(0001, x) = var1{i(t, x), day(i(t)) = 0001}
..
.
o(8784, x) = var1{i(t, x), day(i(t)) = 8784}

127
Statistical values Reference manual

2.8.30. YDAYSTAT - Multi-year daily statistical values

Synopsis

< operator > infile outfile

Description
This module computes statistical values of each day of year. Depending on the chosen operator
the minimum, maximum, range, sum, average, variance or standard deviation of each day of year
in infile is written to outfile. The date information in an output field is the date of the last
contributing input field.

Operators

ydaymin Multi-year daily minimum

o(001, x) = min{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = min{i(t, x), day(i(t)) = 366}
ydaymax Multi-year daily maximum
o(001, x) = max{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = max{i(t, x), day(i(t)) = 366}
ydayrange Multi-year daily range
o(001, x) = range{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = range{i(t, x), day(i(t)) = 366}
ydaysum Multi-year daily sum
o(001, x) = sum{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = sum{i(t, x), day(i(t)) = 366}
ydaymean Multi-year daily mean
o(001, x) = mean{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = mean{i(t, x), day(i(t)) = 366}
ydayavg Multi-year daily average
o(001, x) = avg{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = avg{i(t, x), day(i(t)) = 366}
ydaystd Multi-year daily standard deviation
Normalize by n.
o(001, x) = std{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = std{i(t, x), day(i(t)) = 366}
ydaystd1 Multi-year daily standard deviation (n-1)
Normalize by (n-1).

128
Reference manual Statistical values

o(001, x) = std1{i(t, x), day(i(t)) = 001}

..
.
o(366, x) = std1{i(t, x), day(i(t)) = 366}
ydayvar Multi-year daily variance
Normalize by n.
o(001, x) = var{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = var{i(t, x), day(i(t)) = 366}
ydayvar1 Multi-year daily variance (n-1)
Normalize by (n-1).
o(001, x) = var1{i(t, x), day(i(t)) = 001}
..
.
o(366, x) = var1{i(t, x), day(i(t)) = 366}

Example
To compute the daily mean over all input years use:
cdo ydaymean infile outfile

129
Statistical values Reference manual

2.8.31. YDAYPCTL - Multi-year daily percentile values

Synopsis

ydaypctl,p infile1 infile2 infile3 outfile

Description
This operator writes a certain percentile of each day of year in infile1 to outfile. The algorithm
uses histograms with minimum and maximum bounds given in infile2 and infile3, respectively.
The default number of histogram bins is 101. The default can be overridden by setting the environment
variable CDO_PCTL_NBINS to a different value. The files infile2 and infile3 should be the result
of corresponding ydaymin and ydaymax operations, respectively. The date information in an output
field is the date of the last contributing input field.

o(001, x) = pth percentile{i(t, x), day(i(t)) = 001}

..
.
o(366, x) = pth percentile{i(t, x), day(i(t)) = 366}

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the daily 90th percentile over all input years use:
cdo ydaymin infile minfile
cdo ydaymax infile maxfile
cdo ydaypctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo ydaypctl,90 infile -ydaymin infile -ydaymax infile outfile

130
Reference manual Statistical values

2.8.32. YMONSTAT - Multi-year monthly statistical values

Synopsis

< operator > infile outfile

Description
This module computes statistical values of each month of year. Depending on the chosen operator
the minimum, maximum, range, sum, average, variance or standard deviation of each month of year
in infile is written to outfile. The date information in an output field is the date of the last
contributing input field.

Operators

ymonmin Multi-year monthly minimum

o(01, x) = min{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = min{i(t, x), month(i(t)) = 12}
ymonmax Multi-year monthly maximum
o(01, x) = max{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = max{i(t, x), month(i(t)) = 12}
ymonrange Multi-year monthly range
o(01, x) = range{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = range{i(t, x), month(i(t)) = 12}
ymonsum Multi-year monthly sum
o(01, x) = sum{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = sum{i(t, x), month(i(t)) = 12}
ymonmean Multi-year monthly mean
o(01, x) = mean{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = mean{i(t, x), month(i(t)) = 12}
ymonavg Multi-year monthly average
o(01, x) = avg{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = avg{i(t, x), month(i(t)) = 12}
ymonstd Multi-year monthly standard deviation
Normalize by n.
o(01, x) = std{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = std{i(t, x), month(i(t)) = 12}
ymonstd1 Multi-year monthly standard deviation (n-1)
Normalize by (n-1).

131
Statistical values Reference manual

o(01, x) = std1{i(t, x), month(i(t)) = 01}

..
.
o(12, x) = std1{i(t, x), month(i(t)) = 12}
ymonvar Multi-year monthly variance
Normalize by n.
o(01, x) = var{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = var{i(t, x), month(i(t)) = 12}
ymonvar1 Multi-year monthly variance (n-1)
Normalize by (n-1).
o(01, x) = var1{i(t, x), month(i(t)) = 01}
..
.
o(12, x) = var1{i(t, x), month(i(t)) = 12}

Example
To compute the monthly mean over all input years use:
cdo ymonmean infile outfile

132
Reference manual Statistical values

2.8.33. YMONPCTL - Multi-year monthly percentile values

Synopsis

ymonpctl,p infile1 infile2 infile3 outfile

Description
This operator writes a certain percentile of each month of year in infile1 to outfile. The algorithm
uses histograms with minimum and maximum bounds given in infile2 and infile3, respectively.
The default number of histogram bins is 101. The default can be overridden by setting the environment
variable CDO_PCTL_NBINS to a different value. The files infile2 and infile3 should be the result of
corresponding ymonmin and ymonmax operations, respectively. The date information in an output
field is the date of the last contributing input field.

o(01, x) = pth percentile{i(t, x), month(i(t)) = 01}

..
.
o(12, x) = pth percentile{i(t, x), month(i(t)) = 12}

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the monthly 90th percentile over all input years use:
cdo ymonmin infile minfile
cdo ymonmax infile maxfile
cdo ymonpctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo ymonpctl,90 infile -ymonmin infile -ymonmax infile outfile

133
Statistical values Reference manual

2.8.34. YSEASSTAT - Multi-year seasonal statistical values

Synopsis

< operator > infile outfile

Description
This module computes statistical values of each season. Depending on the chosen operator the
minimum, maximum, range, sum, average, variance or standard deviation of each season in infile
is written to outfile. The date information in an output field is the date of the last contributing
input field.

Operators

yseasmin Multi-year seasonal minimum

o(1, x) = min{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = min{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = min{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = min{i(t, x), month(i(t)) = 09, 10, 11}
yseasmax Multi-year seasonal maximum
o(1, x) = max{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = max{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = max{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = max{i(t, x), month(i(t)) = 09, 10, 11}
yseasrange Multi-year seasonal range
o(1, x) = range{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = range{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = range{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = range{i(t, x), month(i(t)) = 09, 10, 11}
yseassum Multi-year seasonal sum
o(1, x) = sum{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = sum{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = sum{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = sum{i(t, x), month(i(t)) = 09, 10, 11}
yseasmean Multi-year seasonal mean
o(1, x) = mean{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = mean{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = mean{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = mean{i(t, x), month(i(t)) = 09, 10, 11}
yseasavg Multi-year seasonal average
o(1, x) = avg{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = avg{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = avg{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = avg{i(t, x), month(i(t)) = 09, 10, 11}
yseasstd Multi-year seasonal standard deviation
o(1, x) = std{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = std{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = std{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = std{i(t, x), month(i(t)) = 09, 10, 11}

134
Reference manual Statistical values

yseasstd1 Multi-year seasonal standard deviation (n-1)

o(1, x) = std1{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = std1{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = std1{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = std1{i(t, x), month(i(t)) = 09, 10, 11}
yseasvar Multi-year seasonal variance
o(1, x) = var{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = var{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = var{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = var{i(t, x), month(i(t)) = 09, 10, 11}
yseasvar1 Multi-year seasonal variance (n-1)
o(1, x) = var1{i(t, x), month(i(t)) = 12, 01, 02}
o(2, x) = var1{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = var1{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = var1{i(t, x), month(i(t)) = 09, 10, 11}

Example
To compute the seasonal mean over all input years use:
cdo yseasmean infile outfile

135
Statistical values Reference manual

2.8.35. YSEASPCTL - Multi-year seasonal percentile values

Synopsis

yseaspctl,p infile1 infile2 infile3 outfile

Description
This operator writes a certain percentile of each season in infile1 to outfile. The algorithm uses
histograms with minimum and maximum bounds given in infile2 and infile3, respectively. The
default number of histogram bins is 101. The default can be overridden by setting the environment
variable CDO_PCTL_NBINS to a different value. The files infile2 and infile3 should be the result of
corresponding yseasmin and yseasmax operations, respectively. The date information in an output
field is the date of the last contributing input field.

o(1, x) = pth percentile{i(t, x), month(i(t)) = 12, 01, 02}

o(2, x) = pth percentile{i(t, x), month(i(t)) = 03, 04, 05}
o(3, x) = pth percentile{i(t, x), month(i(t)) = 06, 07, 08}
o(4, x) = pth percentile{i(t, x), month(i(t)) = 09, 10, 11}

Parameter
p FLOAT Percentile number in 0, ..., 100

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
To compute the seasonal 90th percentile over all input years use:
cdo yseasmin infile minfile
cdo yseasmax infile maxfile
cdo yseaspctl,90 infile minfile maxfile outfile

Or shorter using operator piping:

cdo yseaspctl,90 infile -yseasmin infile -yseasmax infile outfile

136
Reference manual Statistical values

2.8.36. YDRUNSTAT - Multi-year daily running statistical values

Synopsis

< operator >,nts infile outfile

Description
This module writes running statistical values for each day of year in infile to outfile. Depending
on the chosen operator, the minimum, maximum, sum, average, variance or standard deviation of all
timesteps in running windows of which the medium timestep corresponds to a certain day of year is
computed. The date information in an output field is the date of the timestep in the middle of the last
contributing running window. Note that the operator have to be applied to a continuous time series
of daily measurements in order to yield physically meaningful results. Also note that the output time
series begins (nts-1)/2 timesteps after the first timestep of the input time series and ends (nts-1)/2
timesteps before the last one. For input data which are complete but not continuous, such as time
series of daily measurements for the same month or season within different years, the operator yields
physically meaningful results only if the input time series does include the (nts-1)/2 days before and
after each period of interest.

Operators

ydrunmin Multi-year daily running minimum

o(001, x) = min{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = min{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}
ydrunmax Multi-year daily running maximum
o(001, x) = max{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = max{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}
ydrunsum Multi-year daily running sum
o(001, x) = sum{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = sum{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}
ydrunmean Multi-year daily running mean
o(001, x) = mean{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = mean{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}
ydrunavg Multi-year daily running average
o(001, x) = avg{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = avg{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}
ydrunstd Multi-year daily running standard deviation
Normalize by n.

o(001, x) = std{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = std{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}

137
Statistical values Reference manual

ydrunstd1 Multi-year daily running standard deviation (n-1)

Normalize by (n-1).
o(001, x) = std1{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = std1{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}
ydrunvar Multi-year daily running variance
Normalize by n.
o(001, x) = var{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = var{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}
ydrunvar1 Multi-year daily running variance (n-1)
Normalize by (n-1).
o(001, x) = var1{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = var1{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}

Parameter
nts INTEGER Number of timesteps

Example
Assume the input data provide a continuous time series of daily measurements. To compute the
running multi-year daily mean over all input timesteps for a running window of five days use:
cdo ydrunmean,5 infile outfile

Note that except for the standard deviation the results of the operators in this module are equivalent
to a composition of corresponding operators from the YDAYSTAT and RUNSTAT modules. For
instance, the above command yields the same result as:
cdo ydaymean -runmean,5 infile outfile

138
Reference manual Statistical values

2.8.37. YDRUNPCTL - Multi-year daily running percentile values

Synopsis

ydrunpctl,p,nts infile1 infile2 infile3 outfile

Description
This operator writes running percentile values for each day of year in infile1 to outfile. A certain
percentile is computed for all timesteps in running windows of which the medium timestep corresponds
to a certain day of year. The algorithm uses histograms with minimum and maximum bounds given
in infile2 and infile3, respectively. The default number of histogram bins is 101. The default
can be overridden by setting the environment variable CDO_PCTL_NBINS to a different value. The files
infile2 and infile3 should be the result of corresponding ydrunmin and ydrunmax operations,
respectively. The date information in an output field is the date of the timestep in the middle of
the last contributing running window. Note that the operator have to be applied to a continuous
time series of daily measurements in order to yield physically meaningful results. Also note that the
output time series begins (nts-1)/2 timesteps after the first timestep of the input time series and ends
(nts-1)/2 timesteps before the last. For input data which are complete but not continuous, such as
time series of daily measurements for the same month or season within different years, the operator
only yields physically meaningful results if the input time series does include the (nts-1)/2 days before
and after each period of interest.

o(001, x) = pth percentile{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 001}
..
.
o(366, x) = pth percentile{i(t, x), i(t + 1, x), ..., i(t + nts − 1, x); day[(i(t + (nts − 1)/2)] = 366}

Parameter
p FLOAT Percentile number in 0, ..., 100
nts INTEGER Number of timesteps

Environment
CDO_PCTL_NBINS Sets the number of histogram bins. The default number is 101.

Example
Assume the input data provide a continuous time series of daily measurements. To compute the
running multi-year daily 90th percentile over all input timesteps for a running window of five days
use:
cdo ydrunmin,5 infile minfile
cdo ydrunmax,5 infile maxfile
cdo ydrunpctl,90,5 infile minfile maxfile outfile

Or shorter using operator piping:

cdo ydrunpctl,90,5 infile -ydrunmin infile -ydrunmax infile outfile

139
Correlation and co. Reference manual

2.9. Correlation and co.

This sections contains modules for correlation and co. in grid space and over time.
In this section the abbreviations as in the following table are used:

n
Covariance X
n−1 (xi − x)(yi − y)
covar i=1
 −1   −1   −1 
n n n n n n
covar weighted by X X X X X X
wj  wi xi −  wj  wj xj  yi −  wj  wj yj 
   
{wi , i = 1, ..., n} j=1 i=1 j=1 j=1 j=1 j=1

Here is a short overview of all operators in this section:

fldcor Correlation in grid space

timcor Correlation over time

fldcovar Covariance in grid space

timcovar Covariance over time

140
Reference manual Correlation and co.

2.9.1. FLDCOR - Correlation in grid space

Synopsis

fldcor infile1 infile2 outfile

Description
The correlation coefficient is a quantity that gives the quality of a least squares fitting to the original
data. This operator correlates all gridpoints of two fields for each timestep. With
S(t) = {x, i1 (t, x) 6= missval ∧ i2 (t, x) 6= missval}
it is
P P
i1 (t, x)i2 (t, x)w(x) − i1 (t, x) i2 (t, x) w(x)
x∈S(t) x∈S(t)
o(t, 1) = v ! !
u
u P 2 P P 2 P
t i1 (t, x)2 w(x) − i1 (t, x) w(x) i2 (t, x)2 w(x) − i2 (t, x) w(x)
x∈S(t) x∈S(t) x∈S(t) x∈S(t)

where w(x) are the area weights obtained by the input streams. For every timestep t only those field
elements x belong to the sample, which have i1 (t, x) 6= missval and i2 (t, x) =
6 missval.

2.9.2. TIMCOR - Correlation over time

Synopsis

timcor infile1 infile2 outfile

Description
The correlation coefficient is a quantity that gives the quality of a least squares fitting to the original
data. This operator correlates each gridpoint of two fields over all timesteps. With
S(x) = {t, i1 (t, x) 6= missval ∧ i2 (t, x) 6= missval}
it is
P
i1 (t, x)i2 (t, x) − n i1 (t, x) i2 (t, x)
t∈S(x)
o(1, x) = v ! !
u
u P 2 P 2
t i1 (t, x)2 − n i1 (t, x) i2 (t, x)2 − n i2 (t, x)
t∈S(x) t∈S(x)

For every gridpoint x only those timesteps t belong to the sample, which have i1 (t, x) 6= missval and
i2 (t, x) 6= missval.

141
Correlation and co. Reference manual

2.9.3. FLDCOVAR - Covariance in grid space

Synopsis

fldcovar infile1 infile2 outfile

Description
This operator calculates the covariance of two fields over all gridpoints for each timestep. With
S(t) = {x, i1 (t, x) 6= missval ∧ i2 (t, x) 6= missval}
it is
−1  P  P 
 w(x) i1 (t, x) w(x) i2 (t, x)
X X x∈S(t) x∈S(t)
o(t, 1) =  w(x) w(x) i1 (t, x) −  i2 (t, x) −
 P  P 
w(x) w(x)

x∈S(t) x∈S(t)
x∈S(t) x∈S(t)

where w(x) are the area weights obtained by the input streams. For every timestep t only those field
elements x belong to the sample, which have i1 (t, x) 6= missval and i2 (t, x) =
6 missval.

2.9.4. TIMCOVAR - Covariance over time

Synopsis

timcovar infile1 infile2 outfile

Description
This operator calculates the covariance of two fields at each gridpoint over all timesteps. With
S(x) = {t, i1 (t, x) 6= missval ∧ i2 (t, x) 6= missval}
it is
X   
o(1, x) = n−1 i1 (t, x) − i1 (t, x) i2 (t, x) − i2 (t, x)
t∈S(x)

For every gridpoint x only those timesteps t belong to the sample, which have i1 (t, x) 6= missval and
i2 (t, x) 6= missval.

142
Reference manual Regression

2.10. Regression
This sections contains modules for linear regression of time series.
Here is a short overview of all operators in this section:

regres Regression

detrend Detrend

trend Trend

subtrend Subtract trend

143
Regression Reference manual

2.10.1. REGRES - Regression

Synopsis

regres infile outfile

Description
The values of the input file infile are assumed to be distributed as N (a + bt, σ 2 ) with unknown a,
b and σ 2 . This operator estimates the parameter b. For every field element x only those timesteps t
belong to the sample S(x), which have i(t, x) 6= miss. It is
! !
0
P 1
P 1
P 0
i(t, x) − #S(x) i(t , x) t − #S(x) t
t∈S(x) t0 ∈S(x) t0 ∈S(x)
o(1, x) = !2
1
t0
P P
t− #S(x)
t∈S(x) t0 ∈S(x)

2.10.2. DETREND - Detrend time series

Synopsis

detrend infile outfile

Description
Every time series in infile is linearly detrended. For every field element x only those timesteps t
belong to the sample S(x), which have i(t, x) 6= miss. With
 
1 X 1 X
a(x) = i(t, x) − b(x)  t
#S(x) #S(x)
t∈S(x) t∈S(x)

and
! !
1
i(t0 , x) 1
t0
P P P
i(t, x) − #S(x) t− #S(x)
t∈S(x) t0 ∈S(x) t0 ∈S(x)
b(x) = !2
1
t0
P P
t− #S(x)
t∈S(x) t0 ∈S(x)

it is
o(t, x) = i(t, x) − (a(x) + b(x)t)

Note
This operator has to keep the fields of all timesteps concurrently in the memory. If not enough
memory is available use the operators trend and subtrend.

Example
To detrend the data in infile and to store the detrended data in outfile use:
cdo detrend infile outfile

144
Reference manual Regression

2.10.3. TREND - Trend of time series

Synopsis

trend infile outfile1 outfile2

Description
The values of the input file infile are assumed to be distributed as N (a + bt, σ 2 ) with unknown
a, b and σ 2 . This operator estimates the parameter a and b. For every field element x only those
timesteps t belong to the sample S(x), which have i(t, x) 6= miss. It is
 
1 X 1 X
o1 (1, x) = i(t, x) − b(x)  t
#S(x) #S(x)
t∈S(x) t∈S(x)

and
! !
1
i(t0 , x) 1
t0
P P P
i(t, x) − #S(x) t− #S(x)
t∈S(x) t0 ∈S(x) t0 ∈S(x)
o2 (1, x) = !2
1
t0
P P
t− #S(x)
t∈S(x) t0 ∈S(x)

Thus the estimation for a is stored in outfile1 and that for b is stored in outfile2. To subtract the
trend from the data see operator subtrend.

2.10.4. SUBTREND - Subtract a trend

Synopsis

subtrend infile1 infile2 infile3 outfile

Description
This operator is for subtracting a trend computed by the operator trend. It is
o(t, x) = i1 (t, x) − (i2 (1, x) + i3 (1, x) · t)
where t is the timesteps.

Example
The typical call for detrending the data in infile and storing the detrended data in outfile is:
cdo trend infile afile bfile
cdo subtrend infile afile bfile outfile

The result is identical to a call of the operator detrend:

cdo detrend infile outfile

145
EOFs Reference manual

2.11. EOFs
This section contains modules to compute Empirical Orthogonal Functions and - once they are computed
- their principal coefficients.
An introduction to the theory of principal component analysis as applied here can be found in:
Principal Component Analysis [Peisendorfer]
Details about calculation in the time- and spatial spaces are found in:
Statistical Analysis in Climate Research [vonStorch]
EOFs are defined as the eigen values of the scatter matrix (covariance matrix) of the data. For the sake of
simplicity, samples are regarded as time series of anomalies

(z(t)) , t ∈ {1, . . . , n}

of (column-) vectors z(t) with p entries (where p is the gridsize). Thus, using the fact, that zj (t) are
anomalies, i.e.
n
X
hzj i = n−1 zj (i) = 0 ∀ 1 ≤ j ≤ p
i=1

the scatter matrix S can be written as

n h
X √ i h√ iT
S= Wz(t) Wz(t)
t=1

where W is the diagonal matrix containing the area weight of cell p0 in z at W(x, x).
The matrix S has a set of orthonormal eigenvectors ej , j = 1, . . . p, which are called empirical orthogonal
functions (EOFs) of the sample z. (Please note, that ej is the eigenvector of S and not the weighted
eigen-vector which would be Wej .) Let the corresponding eigenvalues be denoted λj . The vectors ej are
spatial patterns which explain a certain amount of variance of the time series z(t) that is related linearly
to λj . Thus, the spatial pattern defined by the first eigenvector (the one with the largest eigenvalue ) is the
pattern which explains a maximum possible amount of variance of the sample z(t). The orthonormality of
eigenvectors reads as
p h p 
6 k
0 if j =
X p i hp i X
W(x, x)ej (x) W(x, x)ek (x) = W(x, x)ej (x)ek (x) =
x=1 x=1
1 if j = k

If all EOFs ej with λj 6= 0 are calculated, the data can be reconstructed from
p
X
z(t, x) = W(x, x)aj (t)ej (x)
j=1

where aj are called the principal components or principal coefficients or EOF coefficients of z. These
coefficients - as readily seen from above - are calculated as the projection of an EOF ej onto a time step
of the data sample z(t0 ) as
p h
X i hp i h√ iT h√ i
p
aj (t0 ) = W(x, x)ej (x) W(x, x)z(t0 , x) = Wz(t0 ) Wej .
x=1

Here is a short overview of all operators in this section:

eof Calculate EOFs in spatial or time space

eoftime Calculate EOFs in time space
eofspatial Calculate EOFs in spatial space
eof3d Calculate 3-Dimensional EOFs in time space

eofcoeff Calculate principal coefficients of EOFs

146
Reference manual EOFs

2.11.1. EOFS - Empirical Orthogonal Functions

Synopsis
< operator >,neof infile outfile1 outfile2

Description
This module calculates empirical orthogonal functions of the data in infile as the eigen values of
the scatter matrix (covariance matrix) S of the data sample z(t). A more detailed description can be
found above.
Please note, that the input data are assumed to be anomalies.
If operator eof is chosen, the EOFs are computed in either time or spatial space, whichever is the
fastest. If the user already knows, which computation is faster, the module can be forced to perform
a computation in time- or gridspace by using the operators eoftime or eofspatial, respectively. This
can enhance performance, especially for very long time series, where the number of timesteps is larger
than the number of grid-points. Data in infile are assumed to be anomalies. If they are not, the
behavior of this module is not well defined. After execution outfile1 will contain all eigen-values
and outfile2 the eigenvectors e_j. All EOFs and eigen-values are computed. However, only the
first neof EOFs are written to outfile2. Nonetheless, outfile1 contains all eigen-values.
Missing values are not fully supported. Support is only checked for non-changing masks of missing
values in time. Although there still will be results, they are not trustworthy, and a warning will occur.
In the latter case we suggest to replace missing values by 0 in infile.

Operators
eof Calculate EOFs in spatial or time space

eoftime Calculate EOFs in time space

eofspatial Calculate EOFs in spatial space

eof3d Calculate 3-Dimensional EOFs in time space

Parameter
neof INTEGER Number of eigen functions

Environment
CDO_SVD_MODE Is used to choose the algorithm for eigenvalue calculation. Options are ’jacobi’
for a one-sided parallel jacobi-algorithm (only executed in parallel if -P flag
is set) and ’danielson_lanczos’ for a non-parallel d/l algorithm. The default
setting is ’jacobi’.
CDO_WEIGHT_MODE It is used to set the weight mode. The default is ’off’. Set it to ’on’ for a
weighted version.
MAX_JACOBI_ITER Is the maximum integer number of annihilation sweeps that is executed if the
jacobi-algorithm is used to compute the eigen values. The default value is 12.
FNORM_PRECISION Is the Frobenius norm of the matrix consisting of an annihilation pair of eigen-
vectors that is used to determine if the eigenvectors have reached a sufficient
level of convergence. If all annihilation-pairs of vectors have a norm below this
value, the computation is considered to have converged properly. Otherwise, a
warning will occur. The default value 1e-12.

147
EOFs Reference manual

Example
To calculate the first 40 EOFs of a data-set containing anomalies use:
cdo eof,40 infile outfile1 outfile2

If the dataset does not containt anomalies, process them first, and use:
cdo sub infile1 -timmean infile1 anom_file
cdo eof,40 anom_file outfile1 outfile2

148
Reference manual EOFs

2.11.2. EOFCOEFF - Principal coefficients of EOFs

Synopsis

eofcoeff infile1 infile2 obase

Description
This module calculates the time series of the principal coefficients for given EOF (empirical orthogonal
functions) and data. Time steps in infile1 are assumed to be the EOFs, time steps in infile2 are
assumed to be the time series. Note, that this operator calculates a non weighted dot product of the
fields in infile1 and infile2. For consistency set the environment variable CDO_WEIGHT_MODE=off
when using eof or eof3d. Given a set of EOFs e_j and a time series of data z(t) with p entries for
each timestep from which e_j have been calculated, this operator calculates the time series of the
projections of data onto each EOF
p
X
oj (t) = z(t, x)ej (x)
x=1

There will be a seperate file o_j for the principal coefficients of each EOF.
As the EOFs e_j are uncorrelated, so are their principal coefficients, i.e.
n  n
X 0 if j 6= k X
oj (t)ok (t) = with oj (t) = 0∀j ∈ {1, . . . , p}.
t=1
λj if j = k t=1

There will be a separate file containing a time series of principal coefficients with time information from
infile2 for each EOF in infile1. Output files will be numbered as <obase><neof><suffix>
where neof+1 is the number of the EOF (timestep) in infile1 and suffix is the filename extension
derived from the file format.

Environment
CDO_FILE_SUFFIX Set the default file suffix. This suffix will be added to the output file names
instead of the filename extension derived from the file format. Set this variable
to NULL to disable the adding of a file suffix.

Example
To calculate principal coefficients of the first 40 EOFs of anom_file, and write them to files beginning
with obase, use:
export CDO_WEIGHT_MODE=off
cdo eof,40 anom_file eval_file eof_file
cdo eofcoeff eof_file anom_file obase

The principal coefficients of the first EOF will be in the file obase000000.nc (and so forth for higher
EOFs, nth EOF will be in obase<n-1>).
If the dataset infile does not containt anomalies, process them first, and use:
export CDO_WEIGHT_MODE=off
cdo sub infile -timmean infile anom_file
cdo eof,40 anom_file eval_file eof_file
cdo eofcoeff eof_file anom_file obase

149
Interpolation Reference manual

2.12. Interpolation
This section contains modules to interpolate datasets. There are several operators to interpolate horizontal
fields to a new grid. Some of those operators can handle only 2D fields on a regular rectangular grid. Vertical
interpolation of 3D variables is possible from hybrid model levels to height or pressure levels. Interpolation
in time is possible between time steps and years.
Here is a short overview of all operators in this section:

remapbil Bilinear interpolation

genbil Generate bilinear interpolation weights

remapbic Bicubic interpolation

genbic Generate bicubic interpolation weights

remapnn Nearest neighbor remapping

gennn Generate nearest neighbor remap weights

remapdis Distance-weighted average remapping

gendis Generate distance-weighted average remap weights

remapcon First order conservative remapping

gencon Generate 1st order conservative remap weights

remapcon2 Second order conservative remapping

gencon2 Generate 2nd order conservative remap weights

remaplaf Largest area fraction remapping

genlaf Generate largest area fraction remap weights

remap Grid remapping

remapeta Remap vertical hybrid level

ml2pl Model to pressure level interpolation

ml2hl Model to height level interpolation

ap2pl Air pressure to pressure level interpolation

ap2hl Air pressure to height level interpolation

intlevel Linear level interpolation

intlevel3d Linear level interpolation onto a 3d vertical coordinate

intlevelx3d like intlevel3d but with extrapolation

inttime Interpolation between timesteps

intntime Interpolation between timesteps

intyear Interpolation between two years

150
Reference manual Interpolation

2.12.1. REMAPBIL - Bilinear interpolation

Synopsis
< operator >,grid infile outfile

Description
This module contains operators for a bilinear remapping of fields between grids in spherical coordi-
nates. The interpolation is based on an adapted SCRIP library version. For a detailed description
of the interpolation method see [SCRIP]. This interpolation method only works on quadrilateral
curvilinear source grids. Below is a schematic illustration of the bilinear remapping:

The figure on the left side shows the input data on a regular lon/lat source grid and on the right side
the remapped result on an unstructured triangular target grid. The figure in the middle shows the
input data with the target grid. Grid cells with missing value are grey colored.

Operators
remapbil Bilinear interpolation
Performs a bilinear interpolation on all input fields.
genbil Generate bilinear interpolation weights
Generates bilinear interpolation weights for the first input field and writes the result
to a file. The format of this file is NetCDF following the SCRIP convention. Use the
operator remap to apply this remapping weights to a data file with the same source
grid.

Parameter
grid STRING Target grid description file or name

Environment
REMAP_EXTRAPOLATE This variable is used to switch the extrapolation feature ’on’ or ’off’. By
default the extrapolation is enabled for circular grids.

Example
Say infile contains fields on a quadrilateral curvilinear grid. To remap all fields bilinear to a Gaussian
N32 grid, type:
cdo remapbil,n32 infile outfile

151
Interpolation Reference manual

2.12.2. REMAPBIC - Bicubic interpolation

Synopsis
< operator >,grid infile outfile

Description
This module contains operators for a bicubic remapping of fields between grids in spherical coordi-
nates. The interpolation is based on an adapted SCRIP library version. For a detailed description
of the interpolation method see [SCRIP]. This interpolation method only works on quadrilateral
curvilinear source grids. Below is a schematic illustration of the bicubic remapping:

The figure on the left side shows the input data on a regular lon/lat source grid and on the right side
the remapped result on an unstructured triangular target grid. The figure in the middle shows the
input data with the target grid. Grid cells with missing value are grey colored.

Operators
remapbic Bicubic interpolation
Performs a bicubic interpolation on all input fields.
genbic Generate bicubic interpolation weights
Generates bicubic interpolation weights for the first input field and writes the result
to a file. The format of this file is NetCDF following the SCRIP convention. Use the
operator remap to apply this remapping weights to a data file with the same source
grid.

Parameter
grid STRING Target grid description file or name

Environment
REMAP_EXTRAPOLATE This variable is used to switch the extrapolation feature ’on’ or ’off’. By
default the extrapolation is enabled for circular grids.

Example
Say infile contains fields on a quadrilateral curvilinear grid. To remap all fields bicubic to a Gaussian
N32 grid, type:
cdo remapbic,n32 infile outfile

152
Reference manual Interpolation

2.12.3. REMAPNN - Nearest neighbor remapping

Synopsis

< operator >,grid infile outfile

Description
This module contains operators for a nearest neighbor remapping of fields between grids in spherical
coordinates. Below is a schematic illustration of the nearest neighbor remapping:

The figure on the left side shows the input data on a regular lon/lat source grid and on the right side
the remapped result on an unstructured triangular target grid. The figure in the middle shows the
input data with the target grid. Grid cells with missing value are grey colored.

Operators

remapnn Nearest neighbor remapping

Performs a nearest neighbor remapping on all input fields.
gennn Generate nearest neighbor remap weights
Generates nearest neighbor remapping weights for the first input field and writes the
result to a file. The format of this file is NetCDF following the SCRIP convention.
Use the operator remap to apply this remapping weights to a data file with the same
source grid.

Parameter
grid STRING Target grid description file or name

Environment
REMAP_EXTRAPOLATE This variable is used to switch the extrapolation feature ’on’ or ’off’. By
default the extrapolation is enabled for this remapping method.
CDO_GRIDSEARCH_RADIUS Grid search radius in degree, default 180 degree.

153
Interpolation Reference manual

2.12.4. REMAPDIS - Distance-weighted average remapping

Synopsis

remapdis,grid[,neighbors] infile outfile

gendis,grid infile outfile

Description
This module contains operators for a distance-weighted average remapping of the four nearest neighbor
values of fields between grids in spherical coordinates. The interpolation is based on an adapted
SCRIP library version. For a detailed description of the interpolation method see [SCRIP]. Below is
a schematic illustration of the distance-weighted average remapping:

The figure on the left side shows the input data on a regular lon/lat source grid and on the right side
the remapped result on an unstructured triangular target grid. The figure in the middle shows the
input data with the target grid. Grid cells with missing value are grey colored.

Operators

remapdis Distance-weighted average remapping

Performs a distance-weighted average remapping of the nearest neighbors value on all
input fields. The default number of nearest neighbors is 4.
gendis Generate distance-weighted average remap weights
Generates distance-weighted average remapping weights of the four nearest neighbor
values for the first input field and writes the result to a file. The format of this file
is NetCDF following the SCRIP convention. Use the operator remap to apply this
remapping weights to a data file with the same source grid.

Parameter
grid STRING Target grid description file or name
neighbors INTEGER Number of nearest neighbors

Environment
REMAP_EXTRAPOLATE This variable is used to switch the extrapolation feature ’on’ or ’off’. By
default the extrapolation is enabled for this remapping method.
CDO_GRIDSEARCH_RADIUS Grid search radius in degree, default 180 degree.

154
Reference manual Interpolation

2.12.5. REMAPCON - First order conservative remapping

Synopsis
< operator >,grid infile outfile

Description
This module contains operators for a first order conservative remapping of fields between grids in
spherical coordinates. The operators in this module uses code from the YAC software package to
compute the conservative remapping weights. For a detailed description of the interpolation method
see [YAC]. The interpolation method is completely general and can be used for any grid on a sphere.
The search algorithm for the conservative remapping requires that no grid cell occurs more than once.
Below is a schematic illustration of the 1st order conservative remapping:

The figure on the left side shows the input data on a regular lon/lat source grid and on the right side
the remapped result on an unstructured triangular target grid. The figure in the middle shows the
input data with the target grid. Grid cells with missing value are grey colored.

Operators
remapcon First order conservative remapping
Performs a first order conservative remapping on all input fields.
gencon Generate 1st order conservative remap weights
Generates first order conservative remapping weights for the first input field and writes
the result to a file. The format of this file is NetCDF following the SCRIP convention.
Use the operator remap to apply this remapping weights to a data file with the same
source grid.

Parameter
grid STRING Target grid description file or name

Environment
CDO_REMAP_NORM This variable is used to choose the normalization of the conservative interpola-
tion. By default CDO_REMAP_NORM is set to ’fracarea’. ’fracarea’ uses the sum of
the non-masked source cell intersected areas to normalize each target cell field
value. This results in a reasonable flux value but the flux is not locally con-
served. The option ’destarea’ uses the total target cell area to normalize each
target cell field value. Local flux conservation is ensured, but unreasonable flux
values may result.

155
Interpolation Reference manual

REMAP_AREA_MIN This variable is used to set the minimum destination area fraction. The default
of this variable is 0.0.

Example
Say infile contains fields on a quadrilateral curvilinear grid. To remap all fields conservative to a
Gaussian N32 grid, type:
cdo remapcon,n32 infile outfile

156
Reference manual Interpolation

2.12.6. REMAPCON2 - Second order conservative remapping

Synopsis
< operator >,grid infile outfile

Description
This module contains operators for a second order conservative remapping of fields between grids in
spherical coordinates. The interpolation is based on an adapted SCRIP library version. For a detailed
description of the interpolation method see [SCRIP]. The interpolation method is completely general
and can be used for any grid on a sphere. The search algorithm for the conservative remapping
requires that no grid cell occurs more than once. Below is a schematic illustration of the 2nd order
conservative remapping:

The figure on the left side shows the input data on a regular lon/lat source grid and on the right side
the remapped result on an unstructured triangular target grid. The figure in the middle shows the
input data with the target grid. Grid cells with missing value are grey colored.

Operators
remapcon2 Second order conservative remapping
Performs a second order conservative remapping on all input fields.
gencon2 Generate 2nd order conservative remap weights
Generates second order conservative remapping weights for the first input field and
writes the result to a file. The format of this file is NetCDF following the SCRIP
convention. Use the operator remap to apply this remapping weights to a data file
with the same source grid.

Parameter
grid STRING Target grid description file or name

Environment
CDO_REMAP_NORM This variable is used to choose the normalization of the conservative interpola-
tion. By default CDO_REMAP_NORM is set to ’fracarea’. ’fracarea’ uses the sum of
the non-masked source cell intersected areas to normalize each target cell field
value. This results in a reasonable flux value but the flux is not locally con-
served. The option ’destarea’ uses the total target cell area to normalize each
target cell field value. Local flux conservation is ensured, but unreasonable flux
values may result.

157
Interpolation Reference manual

REMAP_AREA_MIN This variable is used to set the minimum destination area fraction. The default
of this variable is 0.0.

Note
The SCRIP conservative remapping method doesn’t work correctly for some grid combinations.

Example
Say infile contains fields on a quadrilateral curvilinear grid. To remap all fields conservative (2nd
order) to a Gaussian N32 grid, type:
cdo remapcon2,n32 infile outfile

158
Reference manual Interpolation

2.12.7. REMAPLAF - Largest area fraction remapping

Synopsis

< operator >,grid infile outfile

Description
This module contains operators for a largest area fraction remapping of fields between grids in spherical
coordinates. The operators in this module uses code from the YAC software package to compute
the largest area fraction. For a detailed description of the interpolation method see [YAC]. The
interpolation method is completely general and can be used for any grid on a sphere. The search
algorithm for this remapping method requires that no grid cell occurs more than once. Below is a
schematic illustration of the largest area fraction conservative remapping:

The figure on the left side shows the input data on a regular lon/lat source grid and on the right side
the remapped result on an unstructured triangular target grid. The figure in the middle shows the
input data with the target grid. Grid cells with missing value are grey colored.

Operators

remaplaf Largest area fraction remapping

Performs a largest area fraction remapping on all input fields.
genlaf Generate largest area fraction remap weights
Generates largest area fraction remapping weights for the first input field and writes
the result to a file. The format of this file is NetCDF following the SCRIP convention.
Use the operator remap to apply this remapping weights to a data file with the same
source grid.

Parameter
grid STRING Target grid description file or name

Environment
REMAP_AREA_MIN This variable is used to set the minimum destination area fraction. The default
of this variable is 0.0.

159
Interpolation Reference manual

2.12.8. REMAP - Grid remapping

Synopsis

remap,grid,weights infile outfile

Description
Interpolation between different horizontal grids can be a very time-consuming process. Especially if
the data are on an unstructured and/or a large grid. In this case the interpolation process can be split
into two parts. Firstly the generation of the interpolation weights, which is the most time-consuming
part. These interpolation weights can be reused for every remapping process with the operator remap.
This operator remaps all input fields to a new horizontal grid. The remap type and the interpolation
weights of one input grid are read from a NetCDF file. More weights are computed if the input
fields are on different grids. The NetCDF file with the weights should follow the [SCRIP] convention.
Normally these weights come from a previous call to one of the genXXX operators (e.g. genbil) or
were created by the original SCRIP package.

Parameter
grid STRING Target grid description file or name
weights STRING Interpolation weights (SCRIP NetCDF file)

Environment
CDO_REMAP_NORM This variable is used to choose the normalization of the conservative
interpolation. By default CDO_REMAP_NORM is set to ’fracarea’. ’fracarea’
uses the sum of the non-masked source cell intersected areas to normalize
each target cell field value. This results in a reasonable flux value but
the flux is not locally conserved. The option ’destarea’ uses the total
target cell area to normalize each target cell field value. Local flux
conservation is ensured, but unreasonable flux values may result.
REMAP_EXTRAPOLATE This variable is used to switch the extrapolation feature ’on’ or ’off’.
By default the extrapolation is enabled for remapdis, remapnn and for
circular grids.
REMAP_AREA_MIN This variable is used to set the minimum destination area fraction. The
default of this variable is 0.0.
CDO_GRIDSEARCH_RADIUS Grid search radius in degree, default 180 degree.

Example
Say infile contains fields on a quadrilateral curvilinear grid. To remap all fields bilinear to a Gaussian
N32 grid use:
cdo genbil,n32 infile remapweights.nc
cdo remap,n32,remapweights.nc infile outfile

The result will be the same as:

cdo remapbil,n32 infile outfile

160
Reference manual Interpolation

2.12.9. REMAPETA - Remap vertical hybrid level

Synopsis

remapeta,vct[,oro] infile outfile

Description
This operator interpolates between different vertical hybrid levels. This include the preparation of
consistent data for the free atmosphere. The procedure for the vertical interpolation is based on the
HIRLAM scheme and was adapted from [INTERA]. The vertical interpolation is based on the vertical
integration of the hydrostatic equation with few adjustments. The basic tasks are the following one:
• at first integration of hydrostatic equation
• extrapolation of surface pressure
• Planetary Boundary-Layer (PBL) proutfile interpolation
• interpolation in free atmosphere
• merging of both proutfiles
• final surface pressure correction
The vertical interpolation corrects the surface pressure. This is simply a cut-off or an addition of
air mass. This mass correction should not influence the geostrophic velocity field in the middle
troposhere. Therefore the total mass above a given reference level is conserved. As reference level the
geopotential height of the 400 hPa level is used. Near the surface the correction can affect the vertical
structure of the PBL. Therefore the interpolation is done using the potential temperature. But in
the free atmosphere above a certain n (n=0.8 defining the top of the PBL) the interpolation is done
linearly. After the interpolation both proutfiles are merged. With the resulting temperature/pressure
correction the hydrostatic equation is integrated again and adjusted to the reference level finding the
final surface pressure correction. A more detailed description of the interpolation can be found in
[INTERA]. This operator requires all variables on the same horizontal grid.

Parameter
vct STRING File name of an ASCII dataset with the vertical coordinate table
oro STRING File name with the orography (surf. geopotential) of the target dataset (optional)

Environment
REMAPETA_PTOP Sets the minimum pressure level for condensation. Above this level the humidity
is set to the constant 1.E-6. The default value is 0 Pa.

Note
The code numbers or the variable names of the required parameter have to follow the [ECHAM]
convention.
Use the sinfo command to test if your vertical coordinate system is recognized as hybrid system.
In case remapeta complains about not finding any data on hybrid model levels you may wish to use
the setzaxis command to generate a zaxis description which conforms to the ECHAM convention. See
section "1.4 Z-axis description" for an example how to define a hybrid Z-axis.

161
Interpolation Reference manual

Example
To remap between different hybrid model level data use:
cdo remapeta,vct infile outfile

Here is an example vct file with 19 hybrid model level:

0 0.00000000000000000 0.00000000000000000
1 2000.00000000000000000 0.00000000000000000
2 4000.00000000000000000 0.00000000000000000
3 6046.10937500000000000 0.00033899326808751
4 8267.92968750000000000 0.00335718691349030
5 10609.51171875000000000 0.01307003945112228
6 12851.10156250000000000 0.03407714888453484
7 14698.50000000000000000 0.07064980268478394
8 15861.12890625000000000 0.12591671943664551
9 16116.23828125000000000 0.20119541883468628
10 15356.92187500000000000 0.29551959037780762
11 13621.46093750000000000 0.40540921688079834
12 11101.55859375000000000 0.52493220567703247
13 8127.14453125000000000 0.64610791206359863
14 5125.14062500000000000 0.75969839096069336
15 2549.96899414062500000 0.85643762350082397
16 783.19506835937500000 0.92874687910079956
17 0.00000000000000000 0.97298520803451538
18 0.00000000000000000 0.99228149652481079
19 0.00000000000000000 1.00000000000000000

162
Reference manual Interpolation

2.12.10. VERTINTML - Vertical interpolation

Synopsis

ml2pl,plevels infile outfile

ml2hl,hlevels infile outfile

Description
Interpolate 3D variables on hybrid sigma pressure level to pressure or height levels. The input file
should contain the log. surface pressure or the surface pressure. To extrapolate the temperature, the
surface geopotential is also needed. The pressure, temperature, and surface geopotential are identified
by their GRIB1 code number or NetCDF CF standard name. Supported parameter tables are: WMO
standard table number 2 and ECMWF local table number 128. Use the alias ml2plx/ml2hlx or the
environment variable EXTRAPOLATE to extrapolate missing values. This operator requires all variables
on the same horizontal grid.

Operators

ml2pl Model to pressure level interpolation

Interpolates 3D variables on hybrid sigma pressure level to pressure level.
ml2hl Model to height level interpolation
Interpolates 3D variables on hybrid sigma pressure level to height level. The procedure is
the same as for the operator ml2pl except for the pressure levels being calculated from the
heights by: plevel = 101325 ∗ exp(hlevel/ − 7000)

Parameter
plevels FLOAT Pressure levels in pascal
hlevels FLOAT Height levels in meter (max level: 65535 m)

Environment
EXTRAPOLATE If set to 1 extrapolate missing values.

Example
To interpolate hybrid model level data to pressure levels of 925, 850, 500 and 200 hPa use:

cdo ml2pl,92500,85000,50000,20000 infile outfile

2.12.11. VERTINTAP - Vertical interpolation

Synopsis

ap2pl,plevels infile outfile

ap2hl,hlevels infile outfile

163
Interpolation Reference manual

Description
Interpolate 3D variables on hybrid sigma height coordinates to pressure or height levels. The input
file must contain the 3D air pressure. The air pressure is identified by the NetCDF CF standard name
air_pressure. Use the alias ap2plx/ap2hlx or the environment variable EXTRAPOLATE to extrapolate
missing values. This operator requires all variables on the same horizontal grid.

Operators

ap2pl Air pressure to pressure level interpolation

Interpolates 3D variables on hybrid sigma height coordinates to pressure level.
ap2hl Air pressure to height level interpolation
Interpolates 3D variables on hybrid sigma height coordinates to height level. The procedure
is the same as for the operator ap2pl except for the pressure levels being calculated from
the heights by: plevel = 101325 ∗ exp(hlevel/ − 7000)

Parameter
plevels FLOAT Pressure levels in pascal
hlevels FLOAT Height levels in meter (max level: 65535 m)

Environment
EXTRAPOLATE If set to 1 extrapolate missing values.

Note
This is a specific implementation for NetCDF files from the ICON model, it may not work with data
from other sources.

Example
To interpolate 3D variables on hybrid sigma height level to pressure levels of 925, 850, 500 and 200
hPa use:
cdo ap2pl,92500,85000,50000,20000 infile outfile

2.12.12. INTLEVEL - Linear level interpolation

Synopsis

intlevel,levels infile outfile

Description
This operator performs a linear vertical interpolation of non hybrid 3D variables.

Parameter
levels FLOAT Target levels

164
Reference manual Interpolation

Example
To interpolate 3D variables on height levels to a new set of height levels use:
cdo intlevel,10,50,100,500,1000 infile outfile

165
Interpolation Reference manual

2.12.13. INTLEVEL3D - Linear level interpolation from/to 3d vertical coordinates

Synopsis

< operator >,icoordinate infile1 infile2 outfile

Description
This operator performs a linear vertical interpolation of 3D variables fields with given 3D vertical
coordinates.

Operators

intlevel3d Linear level interpolation onto a 3d vertical coordinate

intlevelx3d like intlevel3d but with extrapolation

Parameter
icoordinate STRING filename for vertical source coordinates variable
infile2 STRING target vertical coordinate field (intlevel3d only)

Example
To interpolate 3D variables from one set of 3d height levels into another one where
• icoordinate contains a single 3d variable, which represents the input 3d vertical coordinate
• infile1 contains the source data, which the vertical coordinate from icoordinate belongs to
• infile2 only contains the target 3d height levels

cdo intlevel3d,icoordinate infile1 infile2 outfile

166
Reference manual Interpolation

2.12.14. INTTIME - Time interpolation

Synopsis

inttime,date,time[,inc] infile outfile

intntime,n infile outfile

Description
This module performs linear interpolation between timesteps.

Operators

inttime Interpolation between timesteps

This operator creates a new dataset by linear interpolation between timesteps. The
user has to define the start date/time with an optional increment.
intntime Interpolation between timesteps
This operator performs linear interpolation between timesteps. The user has to define
the number of timesteps from one timestep to the next.

Parameter
date STRING Start date (format YYYY-MM-DD)
time STRING Start time (format hh:mm:ss)
inc STRING Optional increment (seconds, minutes, hours, days, months, years) [default:
0hour]
n INTEGER Number of timesteps from one timestep to the next

Example
Assumed a 6 hourly dataset starts at 1987-01-01 12:00:00. To interpolate this time series to a one
hourly dataset use:
cdo inttime,1987-01-01,12:00:00,1hour infile outfile

167
Interpolation Reference manual

2.12.15. INTYEAR - Year interpolation

Synopsis

intyear,years infile1 infile2 obase

Description
This operator performs linear interpolation between two years, timestep by timestep. The input
files need to have the same structure with the same variables. The output files will be named
<obase><yyyy><suffix> where yyyy will be the year and suffix is the filename extension derived
from the file format.

Parameter
years INTEGER Comma separated list of years

Environment
CDO_FILE_SUFFIX Set the default file suffix. This suffix will be added to the output file names
instead of the filename extension derived from the file format. Set this variable
to NULL to disable the adding of a file suffix.

Note
This operator needs to open all output files simultaneously. The maximum number of open files
depends on the operating system!

Example
Assume there are two monthly mean datasets over a year. The first dataset has 12 timesteps for the
year 1985 and the second one for the year 1990. To interpolate the years between 1985 and 1990
month by month use:
cdo intyear,1986,1987,1988,1989 infile1 infile2 year

Example result of ’dir year*’ for NetCDF datasets:

y e a r 1 9 8 6 . nc y e a r 1 9 8 7 . nc y e a r 1 9 8 8 . nc y e a r 1 9 8 9 . nc

168
Reference manual Transformation

2.13. Transformation
This section contains modules to perform spectral transformations.
Here is a short overview of all operators in this section:

sp2gp Spectral to gridpoint

gp2sp Gridpoint to spectral

sp2sp Spectral to spectral

dv2ps D and V to velocity potential and stream function

dv2uv Divergence and vorticity to U and V wind

uv2dv U and V wind to divergence and vorticity

fourier Fourier transformation

169
Transformation Reference manual

2.13.1. SPECTRAL - Spectral transformation

Synopsis

< operator >[,gridtype] infile outfile

Description
This module transforms fields on a global regular Gaussian grid to spectral coefficients and vice
versa. The transformation is achieved by applying Fast Fourier Transformation (FFT) first and
direct Legendre Transformation afterwards in gp2sp. In sp2gp the inverse Legendre Transformation
and inverse FFT are used. Missing values are not supported.
The relationship between the spectral resolution, governed by the truncation number T, and the grid
resolution depends on the number of grid points at which the shortest wavelength field is represented.
For a grid with 2N points between the poles (so 4N grid points in total around the globe) the
relationship is:
linear grid: the shortest wavelength is represented by 2 grid points → 4N ' 2(TL + 1)
quadratic grid: the shortest wavelength is represented by 3 grid points → 4N ' 3(TQ + 1)
The quadratic grid is used by ECHAM and ERA15. ERA40 is using a linear Gaussian grid reflected
by the TL notation.
The following table shows the calculation of the number of latitudes and the triangular truncation
for the different grid types:

Gridtype Number of latitudes: nlat Triangular truncation: ntr

linear NINT((ntr*2 + 1)/2) (nlat*2 - 1) / 2
quadratic NINT((ntr*3 + 1)/2) (nlat*2 - 1) / 3

Operators

sp2gp Spectral to gridpoint

Convert all spectral fields to a global regular Gaussian grid.
gp2sp Gridpoint to spectral
Convert all Gaussian gridpoint fields to spectral fields.

Parameter
gridtype STRING Type of the grid: quadratic, linear (default: quadratic)

Example
To transform spectral coefficients from T106 to N80 Gaussian grid use:
cdo sp2gp infile outfile

To transform spectral coefficients from TL159 to N80 Gaussian grid use:

cdo sp2gp,linear infile outfile

170
Reference manual Transformation

2.13.2. SPECCONV - Spectral conversion

Synopsis

sp2sp,trunc infile outfile

Description
Changed the triangular truncation of all spectral fields. This operator performs downward conversion
by cutting the resolution. Upward conversions are achieved by filling in zeros.

Parameter
trunc INTEGER New spectral resolution

2.13.3. WIND2 - D and V to velocity potential and stream function

Synopsis

dv2ps infile outfile

Description
Calculate spherical harmonic coefficients of velocity potential and stream function from spherical
harmonic coefficients of relative divergence and vorticity. The divergence and vorticity need to have
the names sd and svo or code numbers 155 and 138.

171
Transformation Reference manual

2.13.4. WIND - Wind transformation

Synopsis

< operator >[,gridtype] infile outfile

Description
This module converts relative divergence and vorticity to U and V wind and vice versa. Divergence
and vorticity are spherical harmonic coefficients in spectral space and U and V are on a global regular
Gaussian grid. The Gaussian latitudes need to be ordered from north to south. Missing values are
not supported.
The relationship between the spectral resolution, governed by the truncation number T, and the grid
resolution depends on the number of grid points at which the shortest wavelength field is represented.
For a grid with 2N points between the poles (so 4N grid points in total around the globe) the
relationship is:
linear grid: the shortest wavelength is represented by 2 grid points → 4N ' 2(TL + 1)
quadratic grid: the shortest wavelength is represented by 3 grid points → 4N ' 3(TQ + 1)
The quadratic grid is used by ECHAM and ERA15. ERA40 is using a linear Gaussian grid reflected
by the TL notation.
The following table shows the calculation of the number of latitudes and the triangular truncation
for the different grid types:

Gridtype Number of latitudes: nlat Triangular truncation: ntr

linear NINT((ntr*2 + 1)/2) (nlat*2 - 1) / 2
quadratic NINT((ntr*3 + 1)/2) (nlat*2 - 1) / 3

Operators

dv2uv Divergence and vorticity to U and V wind

Calculate U and V wind on a Gaussian grid from spherical harmonic coefficients of relative
divergence and vorticity. The divergence and vorticity need to have the names sd and svo
or code numbers 155 and 138.
uv2dv U and V wind to divergence and vorticity
Calculate spherical harmonic coefficients of relative divergence and vorticity from U and
V wind. The U and V wind need to be on a Gaussian grid and need to have the names u
and v or the code numbers 131 and 132.

Parameter
gridtype STRING Type of the grid: quadratic, linear (default: quadratic)

Example
Assume a dataset has at least spherical harmonic coefficients of divergence and vorticity. To transform
the spectral divergence and vorticity to U and V wind on a Gaussian grid use:
cdo dv2uv infile outfile

172
Reference manual Transformation

2.13.5. FOURIER - Fourier transformation

Synopsis

fourier,epsilon infile outfile

Description
The fourier operator performs the fourier transformation or the inverse fourier transformation of
all input fields. If the number of grid cells is a power of 2 then the algorithm of the Fast Fourier
Transformation (FFT) is used.
It is
n−1
1 X
o(t, x) = √ i(t, x)e2πij
n j=0

where a user given epsilon = −1 leads to the forward transformation and a user given epsilon = 1
leads to the backward transformation. If the input stream infile consists only of complex fields,
then the fields of outfile, computed by
cdo -f ext fourier,1 -fourier,-1 infile outfile

are the same than that of infile. For real input files see function retocomplex.

Parameter
epsilon INTEGER -1: forward transformation; 1: backward transformation

Note
Currently only the EXTRA format supports fields with complex numbers.

173
Import/Export Reference manual

2.14. Import/Export
This section contains modules to import and export data files which can not read or write directly with
CDO.
Here is a short overview of all operators in this section:

import_binary Import binary data sets

import_cmsaf Import CM-SAF HDF5 files

import_amsr Import AMSR binary files

input ASCII input

inputsrv SERVICE ASCII input
inputext EXTRA ASCII input

output ASCII output

outputf Formatted output
outputint Integer output
outputsrv SERVICE ASCII output
outputext EXTRA ASCII output

outputtab Table output

gmtxyz GMT xyz format

gmtcells GMT multiple segment format

174
Reference manual Import/Export

2.14.1. IMPORTBINARY - Import binary data sets

Synopsis

import_binary infile outfile

Description
This operator imports gridded binary data sets via a GrADS data descriptor file. The GrADS data
descriptor file contains a complete description of the binary data as well as instructions on where to
find the data and how to read it. The descriptor file is an ASCII file that can be created easily with
a text editor. The general contents of a gridded data descriptor file are as follows:
• Filename for the binary data
• Missing or undefined data value
• Mapping between grid coordinates and world coordinates
• Description of variables in the binary data set
A detailed description of the components of a GrADS data descriptor file can be found in [GrADS].
Here is a list of the supported components: BYTESWAPPED, CHSUB, DSET, ENDVARS, FILE-
HEADER, HEADERBYTES, OPTIONS, TDEF, TITLE, TRAILERBYTES, UNDEF, VARS, XDEF,
XYHEADER, YDEF, ZDEF

Note
Only 32-bit IEEE floats are supported for standard binary files!

Example
To convert a binary data file to NetCDF use:
cdo -f nc import_binary infile.ctl outfile.nc

Here is an example of a GrADS data descriptor file:

DSET ^ i n f i l e . b i n
OPTIONS s e q u e n t i a l
UNDEF −9e+33
XDEF 360 LINEAR −179.5 1
YDEF 180 LINEAR −89.5 1
ZDEF 1 LINEAR 1 1
TDEF 1 LINEAR 0 0 : 0 0 Z15jun1989 12 hr
VARS 1
param 1 99 d e s c r i p t i o n o f t h e v a r i a b l e
ENDVARS

The binary data file infile.bin contains one parameter on a global 1 degree lon/lat grid written with
FORTRAN record length headers (sequential).

175
Import/Export Reference manual

2.14.2. IMPORTCMSAF - Import CM-SAF HDF5 files

Synopsis

import_cmsaf infile outfile

Description
This operator imports gridded CM-SAF (Satellite Application Facility on Climate Monitoring) HDF5
files. CM-SAF exploits data from polar-orbiting and geostationary satellites in order to provide
climate monitoring products of the following parameters:

Cloud parameters: cloud fraction (CFC), cloud type (CTY), cloud phase (CPH), cloud top height,
pressure and temperature (CTH,CTP,CTT), cloud optical thickness (COT), cloud water
path (CWP).
Surface radiation components: Surface albedo (SAL); surface incoming (SIS) and net (SNS) shortwave
radiation; surface downward (SDL) and outgoing (SOL) longwave radiation, surface net
longwave radiation (SNL) and surface radiation budget (SRB).
Top-of-atmosphere radiation components: Incoming (TIS) and reflected (TRS) solar radiative flux
at top-of-atmosphere. Emitted thermal radiative flux at top-of-atmosphere (TET).
Water vapour: Vertically integrated water vapour (HTW), layered vertically integrated water vapour
and layer mean temperature and relative humidity for 5 layers (HLW), temperature and
mixing ratio at 6 pressure levels.

Daily and monthly mean products can be ordered via the CM-SAF web page (www.cmsaf.eu). Prod-
ucts with higher spatial and temporal resolution, i.e. instantaneous swath-based products, are avail-
able on request ([email protected]). All products are distributed free-of-charge. More informa-
tion on the data is available on the CM-SAF homepage (www.cmsaf.eu).
Daily and monthly mean products are provided in equal-area projections. CDO reads the projection
parameters from the metadata in the HDF5-headers in order to allow spatial operations like remap-
ping. For spatial operations with instantaneous products on original satellite projection, additional
files with arrays of latitudes and longitudes are needed. These can be obtained from CM-SAF together
with the data.

Note
To use this operator, it is necessary to build CDO with HDF5 support (version 1.6 or higher). The
PROJ.4 library (version 4.6 or higher) is needed for full support of the remapping functionality.

Example
A typical sequence of commands with this operator could look like this:
cdo -f nc remapbil,r360x180 -import_cmsaf cmsaf_product.hdf output.nc

(bilinear remapping to a predefined global grid with 1 deg resolution and conversion to NetCDF).

If you work with CM-SAF data on original satellite project, an additional file with information on
geolocation is required, to perform such spatial operations:
cdo -f nc remapbil,r720x360 -setgrid,cmsaf_latlon.h5 -import_cmsaf cmsaf.hdf out.nc

Some CM-SAF data are stored as scaled integer values. For some operations, it could be desirable
(or necessary) to increase the accuracy of the converted products:

176
Reference manual Import/Export

cdo -b f32 -f nc fldmean -sellonlatbox,0,10,0,10 -remapbil,r720x360 \

-import_cmsaf cmsaf_product.hdf output.nc

2.14.3. IMPORTAMSR - Import AMSR binary files

Synopsis

import_amsr infile outfile

Description
This operator imports gridded binary AMSR (Advanced Microwave Scanning Radiometer) data. The
binary data files are available from the AMSR ftp site (ftp://ftp.ssmi.com/amsre). Each file consists
of twelve (daily) or five (averaged) 0.25 x 0.25 degree grid (1440,720) byte maps. For daily files,
six daytime maps in the following order, Time (UTC), Sea Surface Temperature (SST), 10 meter
Surface Wind Speed (WSPD), Atmospheric Water Vapor (VAPOR), Cloud Liquid Water (CLOUD),
and Rain Rate (RAIN), are followed by six nighttime maps in the same order. Time-Averaged files
contain just the geophysical layers in the same order [SST, WSPD, VAPOR, CLOUD, RAIN]. More
information to the data is available on the AMSR homepage http://www.remss.com/amsr.

Example
To convert monthly binary AMSR files to NetCDF use:
cdo -f nc amsre_yyyymmv5 amsre_yyyymmv5.nc

177
Import/Export Reference manual

2.14.4. INPUT - Formatted input

Synopsis

input,grid[,zaxis] outfile
inputsrv outfile
inputext outfile

Description
This module reads time series of one 2D variable from standard input. All input fields need to have
the same horizontal grid. The format of the input depends on the chosen operator.

Operators

input ASCII input

Reads fields with ASCII numbers from standard input and stores them in outfile.
The numbers read are exactly that ones which are written out by the output operator.
inputsrv SERVICE ASCII input
Reads fields with ASCII numbers from standard input and stores them in outfile.
Each field should have a header of 8 integers (SERVICE likely). The numbers that are
read are exactly that ones which are written out by the outputsrv operator.
inputext EXTRA ASCII input
Read fields with ASCII numbers from standard input and stores them in outfile.
Each field should have header of 4 integers (EXTRA likely). The numbers read are
exactly that ones which are written out by the outputext operator.

Parameter
grid STRING Grid description file or name
zaxis STRING Z-axis description file

Example
Assume an ASCII dataset contains a field on a global regular grid with 32 longitudes and 16 latitudes
(512 elements). To create a GRIB1 dataset from the ASCII dataset use:
cdo -f grb input,r32x16 outfile.grb < my_ascii_data

178
Reference manual Import/Export

2.14.5. OUTPUT - Formatted output

Synopsis
output infiles
outputf ,format[,nelem] infiles
outputint infiles
outputsrv infiles
outputext infiles

Description
This module prints all values of all input datasets to standard output. All input fields need to have
the same horizontal grid. All input files need to have the same structure with the same variables.
The format of the output depends on the chosen operator.

Operators
output ASCII output
Prints all values to standard output. Each row has 6 elements with the C-style format
"%13.6g".
outputf Formatted output
Prints all values to standard output. The format and number of elements for each row
have to be specified by the parameters format and nelem. The default for nelem is 1.
outputint Integer output
Prints all values rounded to the nearest integer to standard output.
outputsrv SERVICE ASCII output
Prints all values to standard output. Each field with a header of 8 integers (SERVICE
likely).
outputext EXTRA ASCII output
Prints all values to standard output. Each field with a header of 4 integers (EXTRA
likely).

Parameter
format STRING C-style format for one element (e.g. %13.6g)
nelem INTEGER Number of elements for each row (default: nelem = 1)

Example
To print all field elements of a dataset formatted with "%8.4g" and 8 values per line use:
cdo outputf,%8.4g,8 infile

Example result of a dataset with one field on 64 grid points:

261.7 262 257.8 252.5 248.8 247.7 246.3 246.1
250.6 252.6 253.9 254.8 252 246.6 249.7 257.9
273.4 266.2 259.8 261.6 257.2 253.4 251 263.7
267.5 267.4 272.2 266.7 259.6 255.2 272.9 277.1
275.3 275.5 276.4 278.4 282 269.6 278.7 279.5
282.3 284.5 280.3 280.3 280 281.5 284.7 283.6
292.9 290.5 293.9 292.6 292.7 292.8 294.1 293.6
293.8 292.6 291.2 292.6 293.2 292.8 291 291.2

179
Import/Export Reference manual

2.14.6. OUTPUTTAB - Table output

Synopsis

outputtab,params infiles outfile

Description
This operator prints a table of all input datasets to standard output. infiles is an arbitrary number
of input files. All input files need to have the same structure with the same variables on different
timesteps. All input fields need to have the same horizontal grid.
The contents of the table depends on the chosen paramters. The format of each table parameter is
keyname[:len]. len is the optional length of a table entry. Here is a list of all valid keynames:

Keyname Type Description

value FLOAT Value of the variable [len:8]
name STRING Name of the variable [len:8]
param STRING Parameter ID (GRIB1: code[.tabnum]; GRIB2: num[.cat[.dis]]) [len:11]
code INTEGER Code number [len:4]
lon FLOAT Longitude coordinate [len:6]
lat FLOAT Latitude coordinate [len:6]
lev FLOAT Vertical level [len:6]
xind INTEGER Grid x index [len:4]
yind INTEGER Grid y index [len:4]
timestep INTEGER Timestep number [len:6]
date STRING Date (format YYYY-MM-DD) [len:10]
time STRING Time (format hh:mm:ss) [len:8]
year INTEGER Year [len:5]
month INTEGER Month [len:2]
day INTEGER Day [len:2]
nohead INTEGER Disable output of header line

Parameter
params STRING Comma separated list of keynames, one for each column of the table

Example
To print a table with name, date, lon, lat and value information use:
cdo outputtab,name,date,lon,lat,value infile

Here is an example output of a time series with the yearly mean temperatur at lon=10/lat=53.5:
# name date lon lat value
tsurf 1991−12−31 10 53.5 8.83903
tsurf 1992−12−31 10 53.5 8.17439
tsurf 1993−12−31 10 53.5 7.90489
tsurf 1994−12−31 10 53.5 10.0216
tsurf 1995−12−31 10 53.5 9.07798

180
Reference manual Import/Export

2.14.7. OUTPUTGMT - GMT output

Synopsis

< operator > infile

Description
This module prints the first field of the input dataset to standard output. The output can be used to
generate 2D Lon/Lat plots with [GMT]. The format of the output depends on the chosen operator.

Operators

gmtxyz GMT xyz format

The operator exports the first field to the GMT xyz ASCII format. The output can be
used to create contour plots with the GMT module pscontour.
gmtcells GMT multiple segment format
The operator exports the first field to the GMT multiple segment ASCII format. The
output can be used to create shaded gridfill plots with the GMT module psxy.

Example
1) GMT shaded contour plot of a global temperature field with a resolution of 4 degree. The contour
interval is 3 with a rainbow color table.

cdo gmtxyz temp > data.gmt

makecpt -T213/318/3 -Crainbow > gmt.cpt
pscontour -K -JQ0/10i -Rd -I -Cgmt.cpt data.gmt > gmtplot.ps
pscoast -O -J -R -Dc -W -B40g20 >> gmtplot.ps

−160˚ −120˚ −80˚ −40˚ 0˚ 40˚ 80˚ 120˚ 160˚

80˚ 80˚

40˚ 40˚

0˚ 0˚

−40˚ −40˚

−80˚ −80˚

−160˚ −120˚ −80˚ −40˚ 0˚ 40˚ 80˚ 120˚ 160˚

2) GMT shaded gridfill plot of a global temperature field with a resolution of 4 degree. The contour
interval is 3 with a rainbow color table.

181
Import/Export Reference manual

cdo gmtcells temp > data.gmt

makecpt -T213/318/3 -Crainbow > gmt.cpt
psxy -K -JQ0/10i -Rd -L -Cgmt.cpt -m data.gmt > gmtplot.ps
pscoast -O -J -R -Dc -W -B40g20 >> gmtplot.ps

−160˚ −120˚ −80˚ −40˚ 0˚ 40˚ 80˚ 120˚ 160˚

80˚ 80˚

40˚ 40˚

0˚ 0˚

−40˚ −40˚

−80˚ −80˚

−160˚ −120˚ −80˚ −40˚ 0˚ 40˚ 80˚ 120˚ 160˚

182
Reference manual Miscellaneous

2.15. Miscellaneous
This section contains miscellaneous modules which do not fit to the other sections before.
Here is a short overview of all operators in this section:

gradsdes GrADS data descriptor file

after ECHAM standard post processor

bandpass Bandpass filtering

lowpass Lowpass filtering
highpass Highpass filtering

gridarea Grid cell area

gridweights Grid cell weights

smooth Smooth grid points

smooth9 9 point smoothing

setvals Set list of old values to new values

setrtoc Set range to constant
setrtoc2 Set range to constant others to constant2

timsort Sort over the time

const Create a constant field

random Create a field with random numbers
topo Create a field with topography
for Create a time series
stdatm Create values for pressure and temperature for hydrostatic atmosphere

uvDestag Destaggering of u/v wind components

rotuvNorth Rotate u/v wind to North pole.
projuvLatLon Cylindrical Equidistant projection

rotuvb Backward rotation

mastrfu Mass stream function

sealevelpressure Sea level pressure

adisit Potential temperature to in-situ temperature

adipot In-situ temperature to potential temperature

rhopot Calculates potential density

histcount Histogram count

histsum Histogram sum
histmean Histogram mean
histfreq Histogram frequency

sethalo Set the left and right bounds of a field

wct Windchill temperature

fdns Frost days where no snow index per time period

183
Miscellaneous Reference manual

strwin Strong wind days index per time period

strbre Strong breeze days index per time period

strgal Strong gale days index per time period

hurr Hurricane days index per time period

cmorlite CMOR lite

184
Reference manual Miscellaneous

2.15.1. GRADSDES - GrADS data descriptor file

Synopsis

gradsdes[,mapversion] infile

Description
Creates a [GrADS] data descriptor file. Supported file formats are GRIB1, NetCDF, SERVICE,
EXTRA and IEG. For GRIB1 files the GrADS map file is also generated. For SERVICE and EXTRA
files the grid have to be specified with the CDO option ’-g <grid>’. This module takes infile in
order to create filenames for the descriptor (infile.ctl) and the map (infile.gmp) file.

Parameter
mapversion INTEGER Format version of the GrADS map file for GRIB1 datasets. Use 1 for
a machine specific version 1 GrADS map file, 2 for a machine independent version 2
GrADS map file and 4 to support GRIB files >2GB. A version 2 map file can be used
only with GrADS version 1.8 or newer. A version 4 map file can be used only with
GrADS version 2.0 or newer. The default is 4 for files >2GB, otherwise 2.

Example
To create a GrADS data descriptor file from a GRIB1 dataset use:
cdo gradsdes infile.grb

This will create a descriptor file with the name infile.ctl and the map file infile.gmp.
Assumed the input GRIB1 dataset has 3 variables over 12 timesteps on a Gaussian N16 grid. The
contents of the resulting GrADS data description file is approximately:
DSET ^ i n f i l e . grb
DTYPE GRIB
INDEX ^ i n f i l e . gmp
XDEF 64 LINEAR 0 . 0 0 0 0 0 0 5 . 6 2 5 0 0 0
YDEF 32 LEVELS −85.761 −80.269 −74.745 −69.213 −63.679 −58.143
−52.607 −47.070 −41.532 −35.995 −30.458 −24.920
−19.382 −13.844 −8.307 −2.769 2.769 8.307
13.844 19.382 24.920 30.458 35.995 41.532
47.070 52.607 58.143 63.679 69.213 74.745
80.269 85.761
ZDEF 4 LEVELS 925 850 500 200
TDEF 12 LINEAR 1 2 : 0 0 Z1jan1987 1mo
TITLE i n f i l e . grb T21 g r i d
OPTIONS y r e v
UNDEF −9e+33
VARS 3
geosp 0 129 ,1 ,0 s u r f a c e g e o p o t e n t i a l ( o ro g ra ph y ) [m^2/ s ^ 2 ]
t 4 130 ,99 ,0 temperature [K]
tslm1 0 139 ,1 ,0 s u r f a c e temperature of land [K]
ENDVARS

185
Miscellaneous Reference manual

2.15.2. AFTERBURNER - ECHAM standard post processor

Synopsis

after[,vct] infiles outfile

Description
The "afterburner" is the standard post processor for [ECHAM] data which provides the following
operations:

• Extract specified variables and levels

• Compute derived variables

• Transform spectral data to Gaussian grid representation

• Vertical interpolation to pressure levels

• Compute temporal means

This operator reads selection parameters as namelist from stdin. Use the UNIX redirection "<namelistfile"
to read the namelist from file.

The input files can’t be combined with other CDO operators because of an optimized reader for this
operator.

Namelist
Namelist parameter and there defaults:
TYPE=0 , CODE=−1, LEVEL=−1, INTERVAL=0 , MEAN=0, EXTRAPOLATE=0

TYPE controls the transformation and vertical interpolation. Transforming spectral data to Gaussian
grid representation and vertical interpolation to pressure levels are performed in a chain of steps. The
TYPE parameter may be used to stop the chain at a certain step. Valid values are:

TYPE = 0 : Hybrid level spectral coefficients

TYPE = 10 : Hybrid level fourier coefficients
TYPE = 11 : Hybrid level z o n a l mean s e c t i o n s
TYPE = 20 : Hybrid level gauss g r i d s
TYPE = 30 : Pressure level gauss g r i d s
TYPE = 40 : Pressure level fourier coefficients
TYPE = 41 : Pressure level z o n a l mean s e c t i o n s
TYPE = 50 : Pressure level spectral coefficients
TYPE = 60 : Pressure level fourier coefficients
TYPE = 61 : Pressure level z o n a l mean s e c t i o n s
TYPE = 70 : Pressure level gauss g r i d s

Vorticity, divergence, streamfunction and velocity potential need special treatment in the vertical
transformation. They are not available as types 30, 40 and 41. If you select one of these combinations,
type is automatically switched to the equivalent types 70, 60 and 61. The type of all other variables
will be switched too, because the type is a global parameter.

CODE selects the variables by the ECHAM GRIB1 code number (1-255). The default value -1 processes
all detected codes. Derived variables computed by the afterburner:

186
Reference manual Miscellaneous

Code Name Longname Units Level Needed Codes

34 low_cld low cloud single 223 on modellevel
35 mid_cld mid cloud single 223 on modellevel
36 hih_cld high cloud single 223 on modellevel
131 u u-velocity m/s atm (ml+pl) 138, 155
132 v v-velocity m/s atm (ml+pl) 138, 155
135 omega vertical velocity Pa/s atm (ml+pl) 138, 152, 155
148 stream streamfunction mˆ 2/s atm (ml+pl) 131, 132
149 velopot velocity potential mˆ 2/s atm (ml+pl) 131, 132
151 slp mean sea level pressure Pa surface 129, 130, 152
156 geopoth geopotential height m atm (ml+pl) 129, 130, 133, 152
157 rhumidity relative humidity atm (ml+pl) 130, 133, 152
189 sclfs surface solar cloud forcing surface 176-185
190 tclfs surface thermal cloud forcing surface 177-186
191 sclf0 top solar cloud forcing surface 178-187
192 tclf0 top thermal cloud forcing surface 179-188
259 windspeed windspeed m/s atm (ml+pl) sqrt(u*u+v*v)
260 precip total precipitation surface 142+143

LEVEL selects the hybrid or pressure levels. The allowed values depends on the parameter TYPE. The
default value -1 processes all detected levels.
INTERVAL selects the processing interval. The default value 0 process data on monthly intervals.
INTERVAL=1 sets the interval to daily.
MEAN=1 compute and write monthly or daily mean fields. The default value 0 writes out all timesteps.
EXTRAPOLATE=0 switch of the extrapolation of missing values during the interpolation from model to
pressure level (only available with MEAN=0 and TYPE=30). The default value 1 extrapolate missing
values.
Possible combinations of TYPE, CODE and MEAN:

TYPE CODE MEAN

0/10/11 130 temperature 0
0/10/11 131 u-velocity 0
0/10/11 132 v-velocity 0
0/10/11 133 specific humidity 0
0/10/11 138 vorticity 0
0/10/11 148 streamfunction 0
0/10/11 149 velocity potential 0
0/10/11 152 LnPs 0
0/10/11 155 divergence 0
>11 all codes 0/1

Parameter
vct STRING File with VCT in ASCII format

Example
To interpolate ECHAM hybrid model level data to pressure levels of 925, 850, 500 and 200 hPa, use:
cdo after infile outfile << EON
TYPE=30 LEVEL=92500,85000,50000,20000
EON

187
Miscellaneous Reference manual

2.15.3. FILTER - Time series filtering

Synopsis

bandpass,fmin,fmax infile outfile

lowpass,fmax infile outfile
highpass,fmin infile outfile

Description
This module takes the time series for each gridpoint in infile and (fast fourier) transforms it into the
frequency domain. According to the particular operator and its parameters certain frequencies are
filtered (set to zero) in the frequency domain and the spectrum is (inverse fast fourier) transformed
back into the time domain. To determine the frequency the time-axis of infile is used. (Data should
have a constant time increment since this assumption applies for transformation. However, the time
increment has to be different from zero.) All frequencies given as parameter are interpreted per year.
This is done by the assumption of a 365-day calendar. Consequently if you want to perform multiyear-
filtering accurately you have to delete the 29th of February. If your infile has a 360 year calendar
the frequency parameters fmin respectively fmax should be multiplied with a factor of 360/365 in
order to obtain accurate results. For the set up of a frequency filter the frequency parameters have to
be adjusted to a frequency in the data. Here fmin is rounded down and fmax is always rounded up.
Consequently it is possible to use bandpass with fmin=fmax without getting a zero-field for outfile.
Hints for efficient usage:
• to get reliable results the time-series has to be detrended (cdo detrend)
• the lowest frequency greater zero that can be contained in infile is 1/(N*dT),
• the greatest frequency is 1/(2dT) (Nyquist frequency),
with N the number of timesteps and dT the time increment of infile in years.
Missing value support for operators in this module is not implemented, yet!

Operators

bandpass Bandpass filtering

Bandpass filtering (pass for frequencies between fmin and fmax). Suppresses all vari-
ability outside the frequency range specified by [fmin,fmax].
lowpass Lowpass filtering
Lowpass filtering (pass for frequencies lower than fmax). Suppresses all variability
with frequencies greater than fmax.
highpass Highpass filtering
Highpass filtering (pass for frequencies greater than fmin). Suppresses all variabilty
with frequencies lower than fmin.

Parameter
fmin FLOAT Minimum frequency per year that passes the filter.
fmax FLOAT Maximum frequency per year that passes the filter.

Note
For better performace of these operators use the CDO configure option –with-fftw3.

188
Reference manual Miscellaneous

Example
Now assume your data are still hourly for a time period of 5 years but with a 365/366-day- calendar
and you want to suppress the variability on timescales greater or equal to one year (we suggest here
to use a number x bigger than one (e.g. x=1.5) since there will be dominant frequencies around the
peak (if there is one) as well due to the issue that the time series is not of infinite length). Therefor
you can use the following:
cdo highpass,x -del29feb infile outfile

Accordingly you might use the following to suppress variability on timescales shorter than one year:
cdo lowpass,1 -del29feb infile outfile

Finally you might be interested in 2-year variability. If you want to suppress the seasonal cycle as
well as say the longer cycles in climate system you might use
cdo bandpass,x,y -del29feb infile outfile

with x<=0.5 and y >=0.5.

2.15.4. GRIDCELL - Grid cell quantities

Synopsis

< operator > infile outfile

Description
This module reads the grid cell area of the first grid from the input stream. If the grid cell area is
missing it will be computed from the grid description. Depending on the chosen operator the grid
cell area or weights are written to the output stream.

Operators

gridarea Grid cell area

Writes the grid cell area to the output stream. If the grid cell area have to be
computed it is scaled with the earth radius to square meters.
gridweights Grid cell weights
Writes the grid cell area weights to the output stream.

Environment
PLANET_RADIUS This variable is used to scale the computed grid cell areas to square meters. By
default PLANET_RADIUS is set to an earth radius of 6371000 meter.

189
Miscellaneous Reference manual

2.15.5. SMOOTH - Smooth grid points

Synopsis

smooth[,options] infile outfile

smooth9 infile outfile

Description
Smooth all grid points of a horizontal grid. Options is a comma separated list of "key=value" pairs
with optional parameters.

Operators

smooth Smooth grid points

Performs a N point smoothing on all input fields. The number of points used depend
on the search radius (radius) and the maximum number of points (maxpoints). Per
default all points within the search radius of 1degree are used. The weights for the
points depend on the form of the curve and the distance. The implemented form of the
curve is linear with constant default weights of 0.25 at distance 0 (weight0) and at the
search radius (weightR).
smooth9 9 point smoothing
Performs a 9 point smoothing on all fields with a quadrilateral curvilinear grid. The
result at each grid point is a weighted average of the grid point plus the 8 surrounding
points. The center point receives a weight of 1.0, the points at each side and above and
below receive a weight of 0.5, and corner points receive a weight of 0.3. All 9 points are
multiplied by their weights and summed, then divided by the total weight to obtain the
smoothed value. Any missing data points are not included in the sum; points beyond
the grid boundary are considered to be missing. Thus the final result may be the result
of an averaging with less than 9 points.

Parameter
nsmooth INTEGER Number of times to smooth, default nsmooth=1
radius STRING Search radius, default radius=1deg (units: deg, rad, km, m)
maxpoints INTEGER Maximum number of points, default maxpoints=<gridsize>
form STRING Form of the curve, default form=linear
weight0 FLOAT Weight at distance 0, default weight0=0.25
weightR FLOAT Weight at the search radius, default weightR=0.25

190
Reference manual Miscellaneous

2.15.6. REPLACEVALUES - Replace variable values

Synopsis

setvals,oldval,newval[,...] infile outfile

setrtoc,rmin,rmax,c infile outfile
setrtoc2,rmin,rmax,c,c2 infile outfile

Description
This module replaces old variable values with new values, depending on the operator.

Operators

setvals Set list of old values to new values

Supply a list of n pairs of old and new values.
setrtoc Set rangeto constant
c if i(t, x) ≥ rmin ∧ i(t, x) ≤ rmax
o(t, x) =
i(t, x) if i(t, x) < rmin ∨ i(t, x) > rmax
setrtoc2 Set rangeto constant others to constant2
c if i(t, x) ≥ rmin ∧ i(t, x) ≤ rmax
o(t, x) =
c2 if i(t, x) < rmin ∨ i(t, x) > rmax

Parameter
oldval,newval,... FLOAT Pairs of old and new values
rmin FLOAT Lower bound
rmax FLOAT Upper bound
c FLOAT New value - inside range
c2 FLOAT New value - outside range

191
Miscellaneous Reference manual

2.15.7. TIMSORT - Timsort

Synopsis

timsort infile outfile

Description
Sorts the elements in ascending order over all timesteps for every field position. After sorting it is:

o(t1 , x) <= o(t2 , x) ∀(t1 < t2 ), x

Example
To sort all field elements of a dataset over all timesteps use:
cdo timsort infile outfile

2.15.8. VARGEN - Generate a field

Synopsis

const,const,grid outfile
random,grid[,seed] outfile
topo[,grid] outfile
for,start,end[,inc] outfile
stdatm,levels outfile

Description
Generates a dataset with one or more fields

Operators

const Create a constant field

Creates a constant field. All field elements of the grid have the same value.
random Create a field with random numbers
Creates a field with rectangularly distrubuted random numbers in the interval [0,1].
topo Create a field with topography
Creates a field with topography data, per default on a global half degree grid.
for Create a time series
Creates a time series with field size 1 and field elements beginning with a start value in
time step 1 which is increased from one time step to the next.
stdatm Create values for pressure and temperature for hydrostatic atmosphere
Creates pressure and temperature values for the given list of vertical levels. The formu-
lars are:
  
z
g H exp( H )T0 +∆T
P (z) = P0 exp − R T0 log T0 +∆T

z


T (z) = T0 + ∆T exp − H

192
Reference manual Miscellaneous

with the following constants

T0 = 213K : offset to get a surface temperature of 288K

∆T = 75K : Temperature lapse rate for 10Km
P0 = 1013.25hPa : surface pressure
H= 10000.0m : scale height
g= 9.80665 m
s2 : earth gravity
R= J
287.05 kgK : gas constant for air

This is the solution for the hydrostatic equations and is only valid for the troposphere
(constant positive lapse rate). The temperature increase in the stratosphere and other
effects of the upper atmosphere are not taken into account.

Parameter
const FLOAT Constant
seed INTEGER The seed for a new sequence of pseudo-random numbers [default: 1]
grid STRING Target grid description file or name
start FLOAT Start value of the loop
end FLOAT End value of the loop
inc FLOAT Increment of the loop [default: 1]
levels FLOAT Target levels in metre above surface

Example
To create a standard atmosphere dataset on a given horizontal grid:
cdo enlarge,gridfile -stdatm,10000,8000,5000,3000,2000,1000,500,200,0 outfile

193
Miscellaneous Reference manual

2.15.9. WINDTRANS - Wind Transformation

Synopsis

uvDestag,u,v[,-/+0.5[,-/+0.5]] infile outfile

rotuvNorth,u,v infile outfile
projuvLatLon,u,v infile outfile

Description
This module contains special operators for datsets with wind components on a rotated lon/lat grid,
e.g. data from the regional model HIRLAM or REMO.

Operators

uvDestag Destaggering of u/v wind components

This is a special operator for destaggering of wind components. If the file contains
a grid with temperature (name=’t’ or code=11) then grid_temp will be used for
destaggered wind.
rotuvNorth Rotate u/v wind to North pole.
This is an operator for transformation of wind-vectors from grid-relative to north-
pole relative for the whole file. (FAST implementation with JACOBIANS)
projuvLatLon Cylindrical Equidistant projection
Thus is an operator for transformation of wind-vectors from the globe-spherical
coordinate system into a flat Cylindrical Equidistant (lat-lon) projection. (FAST
JACOBIAN implementation)

Parameter
u,v STRING Pair of u,v wind components (use variable names or code numbers)
-/+0.5,-/+0.5 STRING Destaggered grid offsets are optional (default -0.5,-0.5)

Example
Typical operator sequence on HIRLAM NWP model output (LAMH_D11 files):
cdo uvDestag,33,34 inputfile inputfile_destag
cdo rotuvNorth,33,34 inputfile_destag inputfile_rotuvN

194
Reference manual Miscellaneous

2.15.10. ROTUVB - Rotation

Synopsis

rotuvb,u,v,... infile outfile

Description
This is a special operator for datsets with wind components on a rotated grid, e.g. data from the
regional model REMO. It performs a backward transformation of velocity components U and V from
a rotated spherical system to a geographical system.

Parameter
u,v,... STRING Pairs of zonal and meridional velocity components (use variable names or
code numbers)

Example
To transform the u and v velocity of a dataset from a rotated spherical system to a geographical
system use:
cdo rotuvb,u,v infile outfile

2.15.11. MASTRFU - Mass stream function

Synopsis

mastrfu infile outfile

Description
This is a special operator for the post processing of the atmospheric general circulation model
[ECHAM]. It computes the mass stream function (code=272). The input dataset have to be a
zonal mean of v-velocity [m/s] (code=132) on pressure levels.

Example
To compute the mass stream function from a zonal mean v-velocity dataset use:
cdo mastrfu infile outfile

2.15.12. DERIVEPAR - Sea level pressure

Synopsis

sealevelpressure infile outfile

Description
This operator computes the sea level pressure (air_pressure_at_sea_level). Required input fields are
surface_air_pressure, surface_geopotential and air_temperature on hybrid sigma pressure levels.

195
Miscellaneous Reference manual

2.15.13. ADISIT - Potential temperature to in-situ temperature and vice versa

Synopsis

adisit[,pressure] infile outfile

adipot infile outfile

Description

Operators

adisit Potential temperature to in-situ temperature

This is a special operator for the post processing of the ocean and sea ice model output. It
converts potential temperature adiabatically to in-situ temperature to(t, s, p). Required
input fields are sea water potential temperature (name=tho; code=2) and sea water salinity
(name=sao; code=5). Pressure is calculated from the level information or can be specified
by the optional parameter. Output fields are sea water temperature (name=to; code=20)
and sea water salinity (name=s; code=5).
adipot In-situ temperature to potential temperature
This is a special operator for the post processing of the ocean and sea ice model outpu. It
converts in-situ temperature to potential temperature tho(to, s, p). Required input fields
are sea water in-situ temperature (name=t; code=2) and sea water salinity (name=sao,s;
code=5). Pressure is calculated from the level information or can be specified by the
optional parameter. Output fields are sea water temperature (name=tho; code=2) and
sea water salinity (name=s; code=5).

Parameter
pressure FLOAT Pressure in bar (constant value assigned to all levels)

2.15.14. RHOPOT - Calculates potential density

Synopsis

rhopot[,pressure] infile outfile

Description
This is a special operator for the post processing of the ocean and sea ice model [MPIOM]. It calculates
the sea water potential density (name=rhopoto; code=18). Required input fields are sea water in-situ
temperature (name=to; code=20) and sea water salinity (name=sao; code=5). Pressure is calculated
from the level information or can be specified by the optional parameter.

Parameter
pressure FLOAT Pressure in bar (constant value assigned to all levels)

Example
To compute the sea water potential density from the potential temperature use this operator in
combination with adisit:
cdo rhopot -adisit infile outfile

196
Reference manual Miscellaneous

2.15.15. HISTOGRAM - Histogram

Synopsis

< operator >,bounds infile outfile

Description
This module creates bins for a histogram of the input data. The bins have to be adjacent and have
non-overlapping intervals. The user has to define the bounds of the bins. The first value is the lower
bound and the second value the upper bound of the first bin. The bounds of the second bin are
defined by the second and third value, aso. Only 2-dimensional input fields are allowed. The output
file contains one vertical level for each of the bins requested.

Operators

histcount Histogram count

Number of elements in the bin range.
histsum Histogram sum
Sum of elements in the bin range.
histmean Histogram mean
Mean of elements in the bin range.
histfreq Histogram frequency
Relative frequency of elements in the bin range.

Parameter
bounds FLOAT Comma separated list of the bin bounds (-inf and inf valid)

2.15.16. SETHALO - Set the left and right bounds of a field

Synopsis

sethalo,lhalo,rhalo infile outfile

Description
This operator sets the left and right bounds of the rectangularly understood fields. Positive numbers
of the parameter lhalo enlarges the left bound by the given number of columns from the right bound.
The parameter rhalo does the similar for the right bound. Negative numbers of the parameter
lhalo/rhalo can be used to remove the given number of columns of the left and right bounds.

Parameter
lhalo INTEGER Left halo
rhalo INTEGER Right halo

197
Miscellaneous Reference manual

2.15.17. WCT - Windchill temperature

Synopsis

wct infile1 infile2 outfile

Description
Let infile1 and infile2 be time series of temperature and wind speed records, then a corresponding
time series of resulting windchill temperatures is written to outfile. The wind chill temperature cal-
culation is only valid for a temperature of T <= 33 ℃ and a wind speed of v >= 1.39 m/s. Whenever
these conditions are not satisfied, a missing value is written to outfile. Note that temperature and
wind speed records have to be given in units of ℃ and m/s, respectively.

2.15.18. FDNS - Frost days where no snow index per time period

Synopsis

fdns infile1 infile2 outfile

Description
Let infile1 be a time series of the daily minimum temperature TN and infile2 be a corresponding
series of daily surface snow amounts. Then the number of days where TN < 0 ℃ and the surface snow
amount is less than 1 cm is counted. The temperature TN have to be given in units of Kelvin. The
date information of a timestep in outfile is the date of the last contributing timestep in infile.

2.15.19. STRWIN - Strong wind days index per time period

Synopsis

strwin[,v] infile outfile

Description
Let infile be a time series of the daily maximum horizontal wind speed VX, then the number of
days where VX > v is counted. The horizontal wind speed v is an optional parameter with default v
= 10.5 m/s. A further output variable is the maximum number of consecutive days with maximum
wind speed greater than or equal to v. Note that both VX and v have to be given in units of m/s.
Also note that the horizontal wind speed is defined as the square root of the sum of squares of the
zonal and meridional wind speeds. The date information of a timestep in outfile is the date of the
last contributing timestep in infile.

Parameter
v FLOAT Horizontal wind speed threshold (m/s, default v = 10.5 m/s)

198
Reference manual Miscellaneous

2.15.20. STRBRE - Strong breeze days index per time period

Synopsis

strbre infile outfile

Description
Let infile be a time series of the daily maximum horizontal wind speed VX, then the number of
days where VX is greater than or equal to 10.5 m/s is counted. A further output variable is the
maximum number of consecutive days with maximum wind speed greater than or equal to 10.5 m/s.
Note that VX is defined as the square root of the sum of squares of the zonal and meridional wind
speeds and have to be given in units of m/s. The date information of a timestep in outfile is the
date of the last contributing timestep in infile.

2.15.21. STRGAL - Strong gale days index per time period

Synopsis

strgal infile outfile

Description
Let infile be a time series of the daily maximum horizontal wind speed VX, then the number of
days where VX is greater than or equal to 20.5 m/s is counted. A further output variable is the
maximum number of consecutive days with maximum wind speed greater than or equal to 20.5 m/s.
Note that VX is defined as the square root of the sum of square of the zonal and meridional wind
speeds and have to be given in units of m/s. The date information of a timestep in outfile is the
date of the last contributing timestep in infile.

2.15.22. HURR - Hurricane days index per time period

Synopsis

hurr infile outfile

Description
Let infile be a time series of the daily maximum horizontal wind speed VX, then the number of
days where VX is greater than or equal to 32.5 m/s is counted. A further output variable is the
maximum number of consecutive days with maximum wind speed greater than or equal to 32.5 m/s.
Note that VX is defined as the square root of the sum of squares of the zonal and meridional wind
speeds and have to be given in units of m/s. The date information of a timestep in outfile is the
date of the last contributing timestep in infile.

199
Miscellaneous Reference manual

2.15.23. CMORLITE - CMOR lite

Synopsis

cmorlite,table[,convert] infile outfile

Description
The [CMOR] (Climate Model Output Rewriter) library comprises a set of functions, that can be
used to produce CF-compliant NetCDF files that fulfill the requirements of many of the climate
community’s standard model experiments. These experiments are collectively referred to as MIP’s.
Much of the metadata written to the output files is defined in MIP-specific tables, typically made
available from each MIP’s web site.
The CDO operator cmorlite process the header and variable section of such MIP tables and writes
the result with the internal IO library [CDI]. In addition to the CMOR 2 and 3 table format, the
CDO parameter table format is also supported. The following parameter table entries are available:

Entry Type Description

name WORD Name of the variable
out_name WORD New name of the variable
type WORD Data type (real or double)
standard_name WORD As defined in the CF standard name table
long_name STRING Describing the variable
units STRING Specifying the units for the variable
comment STRING Information concerning the variable
cell_methods STRING Information concerning calculation of means or climatologies
cell_measures STRING Indicates the names of the variables containing cell areas and volumes
missing_value FLOAT Specifying how missing data will be identified
valid_min FLOAT Minimum valid value
valid_max FLOAT Maximum valid value
ok_min_mean_abs FLOAT Minimum absolute mean
ok_max_mean_abs FLOAT Maximum absolute mean
factor FLOAT Scale factor
delete INTEGER Set to 1 to delete variable
convert INTEGER Set to 1 to convert the unit if necessary

Most of the above entries are stored as variables attributes, some of them are handled differently. The
variable name is used as a search key for the parameter table. valid_min, valid_max, ok_min_mean_abs
and ok_max_mean_abs are used to check the range of the data.

Parameter
table STRING Name of the CMOR table as specified from PCMDI
convert STRING Converts the units if necessary

Example
Here is an example of a parameter table for one variable:
prompt> cat mypartab
&parameter
name = t

200
Reference manual Miscellaneous

out_name = ta
standard_name = air_temperature
units = "K"
missing_value = 1e+20
valid_min = 157.1
valid_max = 336.3
/

To apply this parameter table to a dataset use:

cdo -f nc cmorlite,mypartab,convert infile outfile

This command renames the variable t to ta. The standard name of this variable is set to air_temperature
and the unit is set to [K] (converts the unit if necessary). The missing value will be set to 1e+20. In
addition it will be checked whether the values of the variable are in the range of 157.1 to 336.3. The
result will be stored in NetCDF.

201
3. Contributors

3.1. History
CDO was originally developed by Uwe Schulzweida at the Max Planck Institute for Meteorology (MPI-M).
The first public release is available since 2003. The MPI-M, together with the DKRZ, has a long history
in the development of tools for processing climate data. CDO was inspired by some of these tools, such as
the PINGO package and the GRIB-Modules.
PINGO1 was developed by Jürgen Waszkewitz, Peter Lenzen, and Nathan Gillet in 1995 at the DKRZ,
Hamburg (Germany). CDO has a similar user interface and uses some of the PINGO routines.
The GRIB-Modules was developed by Heiko Borgert and Wolfgang Welke in 1991 at the MPI-M. CDO is
using a similar module structure and also some of the routines.

3.2. External sources

CDO has incorporated code from several sources:

afterburner is a postprocessing program for ECHAM data and ECMWF analysis data, originally developed
by Edilbert Kirk, Michael Ponater and Arno Hellbach. The afterburner code was modified for the
CDO operators after, sp2gp, gp2sp.
SCRIP is a software package used to generate interpolation weights for remapping fields from one grid to
another in spherical geometry [SCRIP]. It was developed at the Los Alamos National Laboratory by
Philip W. Jones. The SCRIP library was converted from Fortran to ANSI C and is used as the base
for the remapping operators in CDO.
YAC (Yet Another Coupler) was jointly developed by DKRZ and MPI-M by Moritz Hanke and Rene
Redler [YAC]. CDO is using the clipping and cell search routines for the conservative remapping
with remapcon.
libkdtree a C99 implementation of the kd-tree algorithm developed by Jörg Dietrich.

CDO uses tools from the GNU project, including automake, and libtool.

3.3. Contributors
The primary contributors to the CDO development have been:

Uwe Schulweida : Concept, design and implementation of CDO, project coordination, and releases.
Luis Kornblueh : He supports CDO from the beginning. His main contributions are GRIB performance
and compression, GME and unstructured grid support. Luis also helps with design and planning.
Ralf Müller : He is working on CDO since 2009. His main contributions are the implementation of the
User Portal, the ruby and python interface for all CDO operators, the building process and the
Windows support. The CDO User Portal was funded by the European Commission infracstructure
project IS-ENES. Ralf also helps a lot with the user support. Implemented operators: showunit,
stdatm, intlevel3d, consecsum, consects, ngrids, ngridpoints, reducegrid
1 Procedural INterface for GRIB formatted Objects

202
Contributors Contributors

Cedrick Ansorge : He worked on the CDO software package as a student assistant at MPI-M from 2007-
2011. Implemented operators: eof, eof3d, enscrps, ensbrs, maskregion, bandpass, lowpass, highpass,
smooth9
Oliver Heidmann : He worked on the CDO software package as a student assistant at MPI-M from 2015-
2018.
Karin Meier-Fleischer : She is working in the CDO user support since 2017.
Fabian Wachsmann : He is working on CDO for the CMIP6 project and is responsible for the operator
cmor.
Ralf Quast : He worked on CDO on behalf of the Service Gruppe Anpassung (SGA), DKRZ in 2006. Im-
plemented all ECA Indices of Daily Temperature and Precipitation Extremes, all percentile operators,
module YDRUNSTAT and wct.
Kameswarrao Modali : Implemented operators: contour, shaded, grfill, vector, graph.
Michal Koutek : Implemented operators: selmulti delmulti, changemulti, samplegrid, uvDestag, rotu-
vNorth, projuvLatLon.
Etienne Tourigny : Implemented operators: setclonlatbox, setcindexbox, setvals, splitsel, histfreq, setrtoc,
setrtoc2.
Karl-Hermann Wieners : Implemented operators: aexpr, aexprf, selzaxisname.

Many users have contributed to CDO by sending bug reports, patches and suggestions over time. Very
helpful is also the active participation in the user forum of some users. Here is an incomplete list:
Jaison-Thomas Ambadan, Harald Anlauf, Andy Aschwanden, Stefan Bauer, Simon Blessing,
Renate Brokopf, Michael Boettinger, Tim Brücher, Reinhard Budich, Martin Claus,
Traute Crüger, Irene Fischer-Bruns, Chris Fletscher, Helmut Frank, Kristina Fröhlich,
Oliver Fuhrer, Monika Esch, Pier Giuseppe Fogli, Beate Gayer, Veronika Gayler,
Marco Giorgetta, David Gobbett, Holger Goettel, Helmut Haak, Stefan Hagemann,
Angelika Heil, Barbara Hennemuth, Daniel Hernandez, Nathanael Huebbe, Thomas Jahns,
Frank Kaspar, Daniel Klocke, Edi Kirk, Yvonne Küstermann, Stefanie Legutke,
Leonidas Linardakis, Stephan Lorenz, Frank Lunkeit, Uwe Mikolajewicz, Laura Niederdrenk,
Dirk Notz, Hans-Jürgen Panitz, Ronny Petrik, Swantje Preuschmann, Florian Prill,
Asela Rajapakse, Daniel Reinert, Hannes Reuter, Mathis Rosenhauer, Reiner Schnur,
Martin Schultz, Dennis Shea, Kevin Sieck, Martin Stendel, Bjorn Stevens,
Martina Stockhaus, Claas Teichmann, Jörg Trentmann, Álvaro M. Valdebenito,
Geert Jan van Oldenborgh, Jin-Song von Storch, David Wang, Joerg Wegner, Heiner Widmann,
Claudia Wunram, Klaus Wyser

Please let me know if your name was omitted!

203
Bibliography
[CDI]
Climate Data Interface, from the Max Planck Institute for Meteorologie
[CM-SAF]
Satellite Application Facility on Climate Monitoring, from the German Weather Service (Deutscher
Wetterdienst, DWD)
[CMOR]
Climate Model Output Rewriter, from the Program For Climate Model Diagnosis and Intercomparison
(PCMDI)
[ecCodes]
API for GRIB decoding/encoding, from the European Centre for Medium-Range Weather Forecasts
(ECMWF)
[ECHAM]
The atmospheric general circulation model ECHAM5, from the Max Planck Institute for Meteorologie
[GMT]
The Generic Mapping Tool, from the School of Ocean and Earth Science and Technology (SOEST)
[GrADS]
Grid Analysis and Display System, from the Center for Ocean-Land-Atmosphere Studies (COLA)
[GRIB]
GRIB version 1, from the World Meteorological Organisation (WMO)
[HDF5]
HDF version 5, from the HDF Group
[INTERA]
INTERA Software Package, from the Max Planck Institute for Meteorologie
[Magics]
Magics Software Package, from the European Centre for Medium-Range Weather Forecasts (ECMWF)
[MPIOM]
Ocean and sea ice model, from the Max Planck Institute for Meteorologie
[NetCDF]
NetCDF Software Package, from the UNIDATA Program Center of the University Corporation for
Atmospheric Research
[PINGO]
The PINGO package, from the Model & Data group at the Max Planck Institute for Meteorologie
[REMO]
Regional Model, from the Max Planck Institute for Meteorologie
[Peisendorfer]
Rudolph W. Peisendorfer: Principal Component Analysis, Elsevier (1988)
[PROJ.4]
Cartographic Projections Library, originally written by Gerald Evenden then of the USGS.
[SCRIP]
SCRIP Software Package, from the Los Alamos National Laboratory

204
Bibliography Bibliography

[szip]
Szip compression software, developed at University of New Mexico.
[vonStorch]
Hans von Storch, Walter Zwiers: Statistical Analysis in Climate Research, Cambridge University Press
(1999)
[YAC]
YAC - Yet Another Coupler Software Package, from DKRZ and MPI for Meteorologie

205
A. Environment Variables
The following table describes the environment variables that affect CDO.

Variable name Default Description

CDO_FILE_SUFFIX None Default file suffix. This suffix will be added to the output file
name instead of the filename extension derived from the file
format. NULL will disable the adding of a file suffix.
CDO_HISTORY_INFO 1 Append NetCDF global attribute histroy
CDO_PCTL_NBINS 101 Number of histogram bins.
CDO_RESET_HISTORY 0 Set to 1 to reset the NetCDF history global attribute.
CDO_REMAP_NORM fracarea Choose the normalization for the conservative interpolation
CDO_GRIDSEARCH_RADIUS 180 Grid search radius in degree. Used by the operators
setmisstonn, remapdis and remapnn.
CDO_TIMESTAT_DATE None Set target timestamp of a temporal statistic operator to the "first",
"middle", "midhigh" or "last" contributing source timestep.
CDO_USE_FFTW 1 Set to 0 to switch off usage of FFTW. Used in the Filter module.
CDO_VERSION_INFO 1 Set to 0 to disable NetCDF global attribute CDO

206
B. Parallelized operators
Some of the CDO operators are parallelized with OpenMP. To use CDO with multiple OpenMP threads,
you have to set the number of threads with the option ’-P’. Here is an example to distribute the bilinear
interpolation on 8 OpenMP threads:
cdo -P 8 remapbil,targetgrid infile outfile

The following CDO operators are parallelized with OpenMP:

Module Operator Description

Detrend detrend Detrend
Ensstat ensmin Ensemble minimum
Ensstat ensmax Ensemble maximum
Ensstat enssum Ensemble sum
Ensstat ensmean Ensemble mean
Ensstat ensavg Ensemble average
Ensstat ensvar Ensemble variance
Ensstat ensstd Ensemble standard deviation
Ensstat enspctl Ensemble percentiles
Filter bandpass Bandpass filtering
Filter lowpass Lowpass filtering
Filter highpass Highpass filtering
Fourier fourier Fourier transformation
Genweights genbil Generate bilinear interpolation weights
Genweights genbic Generate bicubic interpolation weights
Genweights gendis Generate distance-weighted average remap weights
Genweights gennn Generate nearest neighbor remap weights
Genweights gencon Generate 1st order conservative remap weights
Genweights gencon2 Generate 2nd order conservative remap weights
Genweights genlaf Generate largest area fraction remap weights
Gridboxstat gridboxmin Gridbox minimum
Gridboxstat gridboxmax Gridbox maximum
Gridboxstat gridboxsum Gridbox sum
Gridboxstat gridboxmean Gridbox mean
Gridboxstat gridboxavg Gridbox average
Gridboxstat gridboxvar Gridbox variance
Gridboxstat gridboxstd Gridbox standard deviation
Remapeta remapeta Remap vertical hybrid level
Remap remapbil Bilinear interpolation
Remap remapbic Bicubic interpolation
Remap remapdis Distance-weighted average remapping
Remap remapnn Nearest neighbor remapping
Remap remapcon First order conservative remapping
Remap remapcon2 Second order conservative remapping
Remap remaplaf Largest area fraction remapping

207
C. Standard name table
The following CF standard names are supported by CDO.

CF standard name Units GRIB 1 code variable name

surface_geopotential m2 s-2 129 geosp
air_temperature K 130 ta
specific_humidity 1 133 hus
surface_air_pressure Pa 134 aps
air_pressure_at_sea_level Pa 151 psl
geopotential_height m 156 zg

208
D. Grid description examples

D.1. Example of a curvilinear grid description

Here is an example for the CDO description of a curvilinear grid. xvals/yvals describe the positions of the
6x5 quadrilateral grid cells. The first 4 values of xbounds/ybounds are the corners of the first grid cell.
gridtype = curvilinear
gridsize = 30
xsize = 6
ysize = 5
xvals = −21 −11 0 11 21 30 −25 −13 0 13
25 36 −31 −16 0 16 31 43 −38 −21
0 21 38 52 −51 −30 0 30 51 64
xbounds = −23 −14 −17 −28 −14 −5 −6 −17 −5 5 6 −6
5 14 17 6 14 23 28 17 23 32 38 28
−28 −17 −21 −34 −17 −6 −7 −21 −6 6 7 −7
6 17 21 7 17 28 34 21 28 38 44 34
−34 −21 −27 −41 −21 −7 −9 −27 −7 7 9 −9
7 21 27 9 21 34 41 27 34 44 52 41
−41 −27 −35 −51 −27 −9 −13 −35 −9 9 13 −13
9 27 35 13 27 41 51 35 41 52 63 51
−51 −35 −51 −67 −35 −13 −21 −51 −13 13 21 −21
13 35 51 21 35 51 67 51 51 63 77 67
yvals = 29 32 32 32 29 26 39 42 42 42
39 35 48 51 52 51 48 43 57 61
62 61 57 51 65 70 72 70 65 58
ybounds = 23 26 36 32 26 27 37 36 27 27 37 37
27 26 36 37 26 23 32 36 23 19 28 32
32 36 45 41 36 37 47 45 37 37 47 47
37 36 45 47 36 32 41 45 32 28 36 41
41 45 55 50 45 47 57 55 47 47 57 57
47 45 55 57 45 41 50 55 41 36 44 50
50 55 64 58 55 57 67 64 57 57 67 67
57 55 64 67 55 50 58 64 50 44 51 58
58 64 72 64 64 67 77 72 67 67 77 77
67 64 72 77 64 58 64 72 58 51 56 64

Figure D.1.: Orthographic and Robinson projection of the curvilinear grid, the first grid cell is colored red

209
Example description for an unstructured grid Grid description examples

D.2. Example description for an unstructured grid

Here is an example of the CDO description for an unstructured grid. xvals/yvals describe the positions
of 30 independent hexagonal grid cells. The first 6 values of xbounds/ybounds are the corners of the first
grid cell. The grid cell corners have to rotate counterclockwise. The first grid cell is colored red.
gridtype = unstructured
gridsize = 30
nvertex = 6
xvals = −36 36 0 −18 18 108 72 54 90 180 144 126 162 −108 −144
−162 −126 −72 −90 −54 0 72 36 144 108 −144 180 −72 −108 −36
xbounds = 339 0 0 288 288 309 21 51 72 72 0 0
0 16 21 0 339 344 340 0 −0 344 324 324
20 36 36 16 0 0 93 123 144 144 72 72
72 88 93 72 51 56 52 72 72 56 36 36
92 108 108 88 72 72 165 195 216 216 144 144
144 160 165 144 123 128 124 144 144 128 108 108
164 180 180 160 144 144 237 267 288 288 216 216
216 232 237 216 195 200 196 216 216 200 180 180
236 252 252 232 216 216 288 304 309 288 267 272
268 288 288 272 252 252 308 324 324 304 288 288
345 324 324 36 36 15 36 36 108 108 87 57
20 15 36 57 52 36 108 108 180 180 159 129
92 87 108 129 124 108 180 180 252 252 231 201
164 159 180 201 196 180 252 252 324 324 303 273
236 231 252 273 268 252 308 303 324 345 340 324
yvals = 58 58 32 0 0 58 32 0 0 58 32 0 0 58 32
0 0 32 0 0 −58 −58 −32 −58 −32 −58 −32 −58 −32 −32
ybounds = 41 53 71 71 53 41 41 41 53 71 71 53
11 19 41 53 41 19 −19 −7 11 19 7 −11
−19 −11 7 19 11 −7 41 41 53 71 71 53
11 19 41 53 41 19 −19 −7 11 19 7 −11
−19 −11 7 19 11 −7 41 41 53 71 71 53
11 19 41 53 41 19 −19 −7 11 19 7 −11
−19 −11 7 19 11 −7 41 41 53 71 71 53
11 19 41 53 41 19 −19 −7 11 19 7 −11
−19 −11 7 19 11 −7 11 19 41 53 41 19
−19 −7 11 19 7 −11 −19 −11 7 19 11 −7
−41 −53 −71 −71 −53 −41 −53 −71 −71 −53 −41 −41
−19 −41 −53 −41 −19 −11 −53 −71 −71 −53 −41 −41
−19 −41 −53 −41 −19 −11 −53 −71 −71 −53 −41 −41
−19 −41 −53 −41 −19 −11 −53 −71 −71 −53 −41 −41
−19 −41 −53 −41 −19 −11 −19 −41 −53 −41 −19 −11

Figure D.2.: Orthographic and Robinson projection of the unstructured grid

210
Operator index

A delcode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
abs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 delete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
acos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 delgridcell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
add . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 delmulti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
addc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 delname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
adipot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 delparam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
adisit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 detrend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
aexpr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 diff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
aexprf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 diffn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
after . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186 distgrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
ap2hl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 div . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
ap2pl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 divc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
asin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 divdpm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
atan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 divdpy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
atan2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 duplicate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
dv2ps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
B dv2uv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
bandpass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
E
C enlarge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
cat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 ensavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
changemulti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 ensbrs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
chcode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 enscrps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
chlevel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 ensmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
chlevelc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 ensmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
chlevelv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 ensmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
chname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 enspctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
chparam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 ensrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
chunit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 ensrkhistspace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
cmorlite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 ensrkhisttime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
codetab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 ensroc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
collgrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 ensstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
consecsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 ensstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
consects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 enssum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
const . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 ensvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 ensvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
cos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 eof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
eof3d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
D eofcoeff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
dayavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 eofspatial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
daymax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 eoftime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
daymean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 eq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
daymin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 eqc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
daypctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 exp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
dayrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 expr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
daystd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 exprf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
daystd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
daysum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 F
dayvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 fdns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
dayvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 fldavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

211
Operator index Operator index

fldcor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 hourstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

fldcovar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 hourstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
fldmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 hoursum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
fldmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 hourvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
fldmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 hourvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
fldpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 hurr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
fldrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
fldstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 I
fldstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 ifnotthen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
fldsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 ifnotthenc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
fldvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 ifthen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
fldvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 ifthenc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
for . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 ifthenelse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
fourier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 import_amsr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
import_binary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
G import_cmsaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
ge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 info . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
gec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 infon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
genbic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
genbil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 inputext . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
gencon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 inputsrv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
gencon2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 int . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
gendis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 intlevel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
genlaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 intlevel3d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
genlevelbounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 intlevelx3d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
gennn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 intntime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
gmtcells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 inttime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
gmtxyz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 intyear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
gp2sp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 invertlat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
gradsdes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 invertlev . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
gridarea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
gridboxavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 L
gridboxmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 le . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
gridboxmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 lec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
gridboxmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 ln . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
gridboxrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 log10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
gridboxstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 lowpass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
gridboxstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 lt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
gridboxsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 ltc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
gridboxvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
gridboxvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 M
griddes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
gridweights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 maskindexbox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
gt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 masklonlatbox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
gtc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 maskregion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
mastrfu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
H max . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
highpass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 maxc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
histcount . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 meravg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
histfreq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 merge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
histmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 mergegrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
histsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 mergetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
houravg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 mermax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
hourmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 mermean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
hourmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 mermin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
hourmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 merpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
hourpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 merrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
hourrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 merstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

212
Operator index Operator index

merstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 reducegrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

mersum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 regres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
mervar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 remap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
mervar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 remapbic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
min . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 remapbil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
minc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 remapcon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
ml2hl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 remapcon2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
ml2pl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 remapdis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
monadd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 remapeta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
monavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 remaplaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
mondiv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 remapnn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
monmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 replace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
monmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 rhopot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
monmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 rotuvb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
monmul . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 rotuvNorth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
monpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 runavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
monrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 runmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
monstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 runmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
monstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 runmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
monsub . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 runpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
monsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 runrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
monvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 runstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
monvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 runstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
mul . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 runsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
mulc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 runvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
muldpm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 runvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
muldpy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
S
N samplegrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
ndate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 sealevelpressure . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
ne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 seasavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
nec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 seasmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
ngridpoints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 seasmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
ngrids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 seasmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
nint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 seaspctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
nlevel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 seasrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
nmon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 seasstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
not . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 seasstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
npar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 seassum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
ntime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 seasvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
nyear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 seasvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
selcode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
O seldate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 selday . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
outputext . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 select . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
outputf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 selgrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
outputint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 selgridcell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
outputsrv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 selhour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
outputtab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 selindexbox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
sellevel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
P sellevidx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
partab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 sellonlatbox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
pow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 selltype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
projuvLatLon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 selmonth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
selmulti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
R selname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
random . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 selparam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
reci . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 selseason . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

213
Operator index Operator index

selsmon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 showmon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
selstdname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 showname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
seltabnum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 showstdname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
seltime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 showtime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
seltimestep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 showtimestamp . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
selyear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 showyear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
selzaxis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 sin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
selzaxisname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 sinfo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
setattribute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 sinfon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
setcalendar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 smooth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
setcindexbox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 smooth9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
setclonlatbox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 sp2gp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
setcode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 sp2sp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
setcodetab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 splitcode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
setctomiss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 splitday . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
setdate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 splitgrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
setday . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 splithour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
setgrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 splitlevel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
setgridarea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 splitmon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
setgridmask . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 splitname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
setgridtype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 splitparam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
sethalo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 splitseas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
setlevel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 splitsel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
setltype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 splittabnum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
setmisstoc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 splityear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
setmisstodis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 splityearmon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
setmisstonn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 splitzaxis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
setmissval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 sqr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
setmon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 sqrt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
setname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 stdatm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
setparam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 strbre . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
setpartabn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 strgal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
setpartabp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 strwin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
setreftime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 sub . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
setrtoc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 subc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
setrtoc2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 subtrend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
setrtomiss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
settaxis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 T
settbounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 tan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
settime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 tee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
settunits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 timavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
setunit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 timcor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
setvals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 timcovar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
setvrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 timcumsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
setyear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 timmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
setzaxis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 timmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
shifttime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 timmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
shiftx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 timpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
shifty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 timrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
showattribute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 timselavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
showatts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 timselmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
showattsglob . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 timselmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
showattsvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 timselmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
showcode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 timselpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
showdate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 timselrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
showformat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 timselstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
showlevel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 timselstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
showltype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 timselsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

214
Operator index Operator index

timselvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 ydrunmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

timselvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 ydrunmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
timsort . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 ydrunmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
timstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 ydrunpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
timstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 ydrunstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
timsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 ydrunstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
timvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 ydrunsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
timvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 ydrunvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
topo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 ydrunvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
trend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 yearavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
yearmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
U yearmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
uv2dv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 yearmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
uvDestag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 yearmonmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
yearpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
V yearrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
varsavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yearstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
varsmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yearstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
varsmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yearsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
varsmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yearvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
varsrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yearvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
varsstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yhouradd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
varsstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yhouravg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
varssum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yhourdiv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
varsvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yhourmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
varsvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 yhourmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 yhourmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vertavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 yhourmul . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
vertmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 yhourrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vertmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 yhourstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vertmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 yhourstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vertrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 yhoursub . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
vertstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 yhoursum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vertstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 yhourvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vertsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 yhourvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vertvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 ymonadd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
vertvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 ymonavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
ymondiv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
W ymonmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
wct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198 ymonmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
ymonmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Y ymonmul . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
ydayadd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 ymonpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
ydayavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 ymonrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
ydaydiv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 ymonstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
ydaymax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 ymonstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
ydaymean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 ymonsub . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
ydaymin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 ymonsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
ydaymul . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 ymonvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
ydaypctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 ymonvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
ydayrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 yseasadd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
ydaystd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 yseasavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
ydaystd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 yseasdiv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
ydaysub . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 yseasmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
ydaysum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 yseasmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
ydayvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 yseasmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
ydayvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 yseasmul . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
ydrunavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 yseaspctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

215
Operator index Operator index

yseasrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
yseasstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
yseasstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
yseassub . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
yseassum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
yseasvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
yseasvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

Z
zaxisdes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
zonavg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonmax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonmin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonpctl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonstd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonstd1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
zonvar1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

216