Ohr No.1 - PMP Manual - Wmo No.332 - 1973

WMO
0 HP_ _i
33 ~
WORLD METEOROLOGICAL ORGANIZATION
Operational Hydrology
Report No. I
MANUAL FOR ESTIMATION OF

PROBABLE MAXIMUM PRECIPITATION
I WMO - No. 332 I

Secretariat of the World Meteorological Organization • Geneva • Switzerland
1973
..- I { ,\ \ ·. ~., l ..1 l; ( , ,(,. (
\1'. ;' -
© 1973, World Meteorological Organization
NOTE
The designations employed and the presentation of the material in this publication do not
imply the expression of any opinion whatsoever on the part of the Secretariat of the World
Meteorological Organization concerning the legal status of any conntry or territory or of its
authorities, or concerning the delimitation of its frontiers.
C 0 N T E N T S
Page
FOREWORD •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• XI
SUMMARY (Engli s h, French, Russian, Spanish) XIII
Chapter l - INTRODUCTION •••••••• .......................................... l
1.1 Definitions of probable maximum precipitation (PMP) l
1.1.l Conceptual definition ......................................... l
1.1.2 Operational definition ........................................ l
1.1.3 Maximum possible precipitation ................................ l
1.1.4 Probable maximum stor m (PMS) 2
1.2 Lower and upper limits of PMP ................................ . 2
1.2. l Accuracy of PMP estimates ..................................... 2
1.2.2 Confidence bands 3
1.3 The manual •...•.••••••.•••.•.•..•••.••.•••..•••.•.••....•..•••. 3
l. 3 . 1 Purpose 3
l. 3. 2 Scope 3
l. 3. 3 Use of examples from actual studies 4
References ............................................................... 5
Chapter 2 - ESTIMATES FOR NON- OROGRAPHIC REGIONS 7
2.1 Introduction 7
2.1. l Convergence model ••••.•••••••••...•••••••• e ••• • •••••••• o •••••• 7
2.1. 2 Observed storm rainfall as an indicator of convergence and

vertical motion 7
2.2 Estimation of atmospheric moisture 7
2.2.l Assumption of a saturated pseudo- adiabatic atmosphere 7

II CONTENTS
Page
2.2.2 Surface dew poi nts as a moisture index ........................ 8
?..2.3 Persisting 12-hour dew points 8
2 . 2.4 Representative persisting 12-hour 1 000 mb storm dew points • • •• 10
2. 2.5 Maximum persisting 12- hour 1 000 mb dew points 11
2 . 2.6 Precipitable water 13
2.3 Moisture maximization 14
2.3.1 Seas on al limitations 14
2.3.2 Depth of precipitable water 14
2.3.3 Applicability of persisting 12-hour dew points for all storm

durations .............•..•.................................... 14
2 . 3. 4 Maximization of storm in place 15
2. 4 Wind maximization 15
2.4.1 Introduction 15
2.4.2 Use in non-orographic region s 16
2.4 . 3 Winds repre sentati ve of moisture inflow in stor ms 16
2.4.3 . 1 Wind direction 17
2 . 4. 3. 2 Wind speed 17
2.4.4 Wind maximization ratio 17
2. 5 Storm transposition ........................................... 18
2. 5.1 Definitions ... ................................................ 18
2.5.2 Steps in transposition •·•·••••••••••••••••• • •••••••••••••••••• 18
2.5.2.1 The storm ••••••••••••••• 0 ••••••••••• 0 ••••••••••••• • ••••••••••• 18
2.5.2.2 Region of influence of sto r m type 18
2.5.2. 3 Topog r aphic controls I I I I I I 9 I I I I I I I I I I I I I I I I I I 9 I 9 I I I I I I I I I I I I 19

CONTENTS II I
Page
2.5 . 2. 4 Adjustments 20
2.6 Traosposition adjustments 20
2.6.l Moisture adjustment for relocation 20
2.6 . l.l Reference dew point for moisture adjustment 20
2.6.2 Elevation adjustments ....................................... 20
2.6.2.l General storms 22.
2.6.2.2 Local thunderstorms 22
2.6.3 Barrier adjustment •••••••••••••••••••••••••••••••••••••••••• 22
2.6.4 Example of storm transposition and maximization 22
2.6.4.l Hypothetical situation •••••••••••••••••••••••••••••••••••••• 22
2.6.4.2 Computation of adjustment factor••••••••••••••••••• ••••••••• 23
2.7 Sequential and spatial maximization 24
2.7.l Definition 24
2.7.2 Sequential maximization 24
2.7.3 Spatial maximization •••••••••••••••••••••••••••••••••••••••• 25
2.7.4 Combined sequential and spatial maximization 28
2.8 Envelopment •......••••••••••.••••••••••••••.•••.....•.•..... 28
2.8.l Introduction ................................................ 28
2.8.2 Envelopment••••••••••••• • ••••••••••••••••••••••••••••••••••• 29
2.8.3 Undercutting ••••.•••.••••••.•.•••••.•.•••••.•••.......•..... 32
2.9 Summary outline of procedure for estimating PMP ........... .. 32
2.9.l Introduction 32
2.9.2 Procedural steps•••••••••••••••••••••••••••••••••••••••••••• 32
2.10 Seasonal variation of PMP •••••••••••••••••••••••••••······~~ 35

IV CONTENTS
Page
2.10.2 Observed storms 35
2.10.3 Maximum persisting 12-hour dew points ••••••••••••••••••••• 35
2.10.4 Moisture inflow•••• • •••••••••••••••••••••••••••••••••••••• 36
2.10.5 Daily station precipitation .............................. . 36
2.10.6 Weekly precipitation data ..................•.....•.....•.• 36
2.11 Areal distribution of PMP ••••••••••••••••••••••••••••••••• 36
2.11.2 Observed storm pattern•••••••••••••••••••••••••••••••••••• 37
2.11.3 Idealized storm pattern ••••••••••••·•·•·•·•••••••••••••••• 37
2.11.3.1 Areal distribution 37
2.11.3.2 Example ••.•••••••••••••••••••••••••••••••••.••..•.•.•.•..• 39
2.12 Time distribution of PMP •••••••••••••••••••••••••••••••••• 43

2.12.1 Order of presentation••••••••••••••••••••••••••••••••• •••• 43
2. 12.2 Chronological order based on observed storm•••• ••• •••••••• 43
2.13 Cautionary remarks •••••••••••••••••••••••••••••••••••••••• 43
2.13.1 Importance of adequate storm sample ••••••••••••••••••••••• 43
2.13.2 Comparison with record rainfalls •••••••••••••••••••••••••· 45
2.13.3 Consistency of estimates 0 ••••••••••••••••••••••••••••• fl ••• 45
2. 13.4 Seasonal variation 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 e 0 0 0 e 0 e 0 0 0 0 0 0 0 0 0 e 0 0 0 0 I 45
2.13.5 Areal distribution .................................. 46
References ••••••••••••••••••••••• 0 •••••••••••••••••••••••••••••••••••• 46
Chapter 3 - ESTIMATES FOR OROGRAPHIC REGIONS .......................... 49
3.1 Precipitation in mountainous regions 49

CONTENTS v
Page
3.1. l Or ograp hic influences ........................................ 49
3. 1. 2 Meteorological influences .................................... 49
3.1. 3 Mean annual and seasonal precipitation ....................... 50
3. 1.4 Storm transposition ••••••••••••••••••••••••• 0 •••••••••••••••• 50
3.1.5 Probable maximum precipitation 50
3.1.5. l Or ographic separation method 50
3. 1.5.2 Modification of non-orographic PMP for orography ••••••••••••• 50
3.1.5. 3 Examples of procedures 51
3. 2 Orographic sepa r ation method 51
3. 2. l Introduction ................................................. 51
3. 2. 2 Orographic model 51
3.2 . 2. l Single layer model 52
3.2.2 . 2 Multiple layer model 54
3 . 2.2.3 Precipitation trajectories 55
3.2. 3 Test of orographic model on observed storm 55
3.2.3.l Ground profile .............................................. . 57
3.2.3.2 Inflow data •..............•.. • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 59
3.2.3.3 Air streamlines •••••••••••••••••• •••••••••••• ••••••• • ••••••••• 59
3.2.3.4 Freezing level ........................................ 59
3.2.3 . 5 Precipitation trajectories •••••••• •••••••••••• •••••••••••••••• 59
3.2.3.6 Precipitation computation ................ • • .. • • • • • • • • • • • · · • • • • 60
3.2.3.7 Comparison of results •••••• ••••• •••••••••••••••••• •••••••••• •• 63
Sources of error .. .. •............•............ . • ·. • · • · · • • • · • • · 65

3.2.3.8
Orographic separation method for estimating PMP •••••••••••••• • 65
3.3
VI CONTENTS
Page
3.3. l Or og rap hic PMP 66
3.3. l. l Maximum winds 66
3.3 .l. 2 Maximum moisture 67
3.3.2 Generalized estimates of orographic PMP 67
3.3 . 3 Variations in orographic PMP •••••••••••••• •• •••••••••••••••••• 68
3.3.3.l Seasonal variation ••• • ••••••••••••••••••••••••••••••••••••••• 0 68
3. 3.3.2 Durational variation 70
3 . 3. 3 . 3 Areal variation 70
3.3.4 Convergence PMP for combination with orographic PMP ••••••••••• 71
3.3.4.l Moisture (dew point) envelopes 71
3.3.4. 2 Envelopes of P/M ratios 71
3.3.4 . 3 Reduction of convergence PMP for elevation 72
3.3.4.4 Reduction for upwind barriers 74
3.3.4.5 Reduction of point, or 25 km2, convergence PMP for basin size •• 75
3.3.4 . 6 Construction of convergence PMP index map 76
3.3.4.7 Adjustmen t of index map values for other durations,

basin sizes and months •..•..•........................ • ........ 78
3.3.5 Combination of orographic and convergence PMP ••• •••••••••••••• 78
3.4 Modification of non-orographic PMP for orography 80
3. 4.l Introduction 80
3. 4.2 Tennessee river basin above Chattanooga, Tennessee 81
3.4.2.l Topographic effects 82
3.4.2.2 Derivation of PMP ..•.•..•.......•••..•.•••••••.••...••...•.... 83
3.4.2.3 Seasonal variation 84

CONTENTS VII
Page
3.4.2 . 4 Depth- duration relations ...................................... 85
3.4.2.5 Geographic distribution of PMP ................................ 85
3. 4.2 . 6 Time distribution of PMP •••••••••••••••••••••••••••••••••••••• 88
3.5 Cautionary remarks on estimating PMP in orographic regions 91
3.5. l Basic data deficien cies ....................................... 91
3.5.2 Orographic separation method .................................. 91
References ................................................................ 93
Chapter 4 - STATISTICAL ESTIMATES ••••••••••••••••••••••••••••••••••••••••• 95
4. 1 Use of statistical procedure .................. ................ 95
4. 2 Development of procedure ...................... ................ 95
4.2.l Basis ........................... .............................. 95
4.2.2 Adjustment of X and S fo r maximum observed event 96

n n
4.2.3 Adjustment of X and S for sample size ••••••••••••••••••.•••• 98

n n
4.2 . 4 Adjustment for fixed observational time intervals •••.•••.•.••• 99
4. 2.5 Area-reduction curves 101
4. 2. 6 Depth- duration relationships •••••••••••.••.••••.•.••••••••• •• . 102
4.3 Application of procedure •••.•.•••..••.••••••••••••.••••••••.•• 103
4.4 Generalized estimates 106
4.5 Cautionary remarks ............................. .. ...... . ... .. . 106
References ......•......................................................... 107
Chapter 5 - GENERALIZED ESTIMATES ••••••.••••••••••••••••••••••••••.•...... 109
5.1 Generalized charts ..........••..............•.......•......... 109
5 . 1.l Bas e n1ap s ••••••••••••••••• •• •••••••••••••••••••••••••••••••••• 109
5 . 1.l.l Scale ... .. ...... .................. . .... . ........... . ......... . 109

VIII CONTENTS
Page
5 . 1.l.2 Grid system 109
5.1.2 Durational and areal consistency llO
5.1.2.l Depth-duration smoothing llO
5.1.2.2 Depth-area smoothing llO
5 . 1.2. 3 Combined depth- a r ea-duration smoothing llO
5 ol a3 Regional smoothing •••• ••••• 0 • ••••••••••••••••••••••••••••••••• 111
5 . 1.3 . l Supplementary aids •••••• •• ••••••••••••••••••• •• ••••••••••••• 0 • 111
5.1.3.2 Maintenance of consistency between maps 113
5 . 1.4 General remarks •••••••••••••••••••• •••••• •••••••••••••• 0 •••••• 113
5. 2 Estimates fo r non-orographic regions 115
5.2.l Basic procedure • •••••••••••••••••••••••••••••••••••••••••••••• 115
5 . 2. 2 Moistu re maximi zation 115
5 . 2.3 Storm transposition 115
5. 2.4 Summary of procedural steps 117
5. 3 Estimates fo r orographic regions 117
5 . 3. l Introduction 117
5 . 3. 2 Hawaiian I slands PMP .. .••...••.•..•...•.•• ..•... •. ...•........ 118
5. 3. 2. l No n-orog raphic PMP ••••••••••••••••••••••••••••••••••• 119
5.3.2.2 Slope intensification of r ainfall 119
5.3.2.3 Gene r alized PMP estimates ..................................... 119
5 . 3. 3 PMP for draina ges up to 250 km 2 in t he Tennessee river basin ••• 120
5.3 . 3.l Outstanding r ainfalls ••••••••••••••••••••••••••••••••••••••••• 121
5 . 3.3.2 Local to pographic classification•••••••••••••••••• •••• •••••••• 123
5 . 3. 3.3 Broad-scale topographic effects ••••••••••••••••• •• •••• • ••••• •• 123

CONTENTS
IX
Page
5.3.3.4 PMP depth-duration curves for 15 km2 •••••••••••••••••••••••••• 124

5.3.3 . 5 Adjustment for moisture and latitudinal gradient •••.•••••••••.• 126
5-3.3.6 Six-hour 15 km2 PMP index map ................................. 128
5.3.3.7 Time distribution of rainfall ................................. 130
5.3.3.8 PMP for specific basins ....................................... 131
5.3.4 PMP for drainages from 250 to 8 000 km2 in the Tennessee river
basin ......................................................... 132
5.3.4.l Derivation of non-orographic PMP •••••••••••••••••••••••••••••• 132
5.3.4.2 Orographic influence s on PMP .................................. 132
5.3.4 .3 Areal and time distribution ................................... 136
5.3.4.4 PMP for specific basins ••• • .•••••.••.••••••••••.••..•••••••••• 136
5.3.5 PMP for the Lower Mekong river basin in south- east Asia 139
5.3.5.l Mean seasonal precipitation map 139
5.3.5.2 The typhoon as a PMP prototype •••••••••••••••••••••••••••••••• 143
5.3.5.3 Adjustment of U.S. tropical storm rainfalls •..•.•.•.•.•.•••••• 143
5.3.5.4 Adjustment of Vietnam tropical storm rainfalls •.•.•.•..•.•.••• 143
5.3.5.5 Generalized estimates of PMP ..........•..••..••••......•.•.... 151
5.3.5.6 Time distribution ............................................ . 151
5.3.5.7 Areal distribution•••••• ••••••••••••••••••••••••·············· 151
5.3.5.8 PMP for specific basins ••••.•••••••••••••••••••••••••••••••••. 153
5.3.6 Generalized thunderstorm PMP estimates for north-western United

States ....................................................... . 155
5.3.6.l Introduction ................................................. . 155
5.3.6.2 PMP depth-duration relation ............................ ...... . 155
5.3.6.3 Seasonal and regional variations •••••••••••••••••••••••••••••• 158

x CONTENTS
Page
5.3.6.4 Elevation adjustment e e e e e e I •• e e e e ••• e • e • e I e I e I I I I I I I I I e I I I I I I I 158
5. 3.6.5 Depth-area relation ........................................... 158
5.3.6.6 Time distribution ............................................. 159
5.3.6.7 Thunderstorm PMP for specific basins .......................... 159

5.4 Cautionary remarks ............................................ 162
Refere nces I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 163
ACKNOWLEDGEMENTS ••••••••••••••••••••••••••••••••••••••••••••••••••••••••• 165
BIBLIOGRAPHY ••.••••••••••••••••••••••••.••••••••••••••••••.•••••••••••••. 167
ANNEX l - TABLES OF PRECIPITABLE WATER IN A SATURATED PSEUDO-ADIABATIC

ATMOSPHERE 169
ANNEX 2 - GREATEST KNOWN RAINFALLS ••••.•••••••••••••••• •. ••••• • •.•••••••• 175
SUBJECT INDEX ............................................... ............. 181

FOREWORD
Having in mind the need for guidance material on the procedures for estimat-
ing probable maximum precipitation for hydrological forecasting and design purposes,
the president of the Commission for Hydrology and the WMO Executive Committee Panel
of Experts for the Inte r national Hydrological Decade recommended that a manual be pre-
pared describing the techniques that have been found generally applicable in middle
latitudes for basins of various sizes subject to both orographic and non-orogra phic
effects.
Arrangements were accordingly made by WMO for the preparation of this report.
It has been written by J. L. H. Paul hus, Gonsulting hydrometeorologist in co- operation
with the Office of Hydrology, National Weather Service, National Oceanic and Atmos-
pheric Administration, U.S . Department of Commerce. Principal contributors from that
office were J . F. Miller, J. T. Riedel, F. K. Schwarz and C. W. Cochrane. Porti ons
of the text were taken from materi al written by V. A. Myers.
I should like to express the gratitude of WMO to Mr. J. L. H. Paulhus and to

those of his colleagues who contributed to this excellent report on a very complex
subject, which is now published as the first of a new series of WMO publications en-
titled "Operational Hydrology" .
.
~·
D. A. Davies
Secretary-General
SUMMARY
Probable ma xi mum precipitation (PMP) is defined as t he greatest depth of pre-

cipitatio n for a given duration meteorologicolly possible for a given basin at a parti -
cular time of year, with no allowanc e mode for long-term climatic trends . Current
knowledge of storm mechan isms an d their precipitation-producing efficiency is inade-
quate to permit precise evalu a tion of limiting values of extreme precipitation. PMP
estimates must be considered th erefore, at least for the present, as approxi mations .
The accuracy, or re liability, of on estimate depends basically on the amount and qu al-
ity of da ta available for applying various estimati ng procedures.
Procedures for estimating PMP cannot be standardized as they vary with amo unt
and quality of data available, basin size and location, basin and re gional topography,
storm types producing extreme precipitation, and climate . There ore many r egions in
various ports of t he world for whic h PMP hos never been estimated. It would be impos-
sible at this time to prepare a manual that would cov3r all problems that might possi -
bly be encountered . Nor would it be practicable to prepare a manual that would cover
all situations that hav e cr i s en in deriving post est imat es. For th ese reasons, this
manual discuss es procedures that have been found generally applicable in the middl e
latit udes for basin s i zes up to about 50 000 km2 in orographic a nd non - orographic
regions.
The procedures are described by examples f r om actual studies by t he National
Weather Service (formerly U.S. Weather Bureau), Nat i on a l Oceanic an d Atmospheric Ad-
ministration , U.S. Deportment of Commerce. Several count ries have made equa ll y valid
studies . The chief reasons fo r using the exa mples desc ribed were that: (1) they re-
presented a variety of problems, (2) they were from stu dies published in widely dis-
tributed reports re latively accessible fo r re ferenc e, and (3) read y availability o f
basic materia l, such as photographic prints of many illustrations, mi ni mi zed time and
cost of preparing this manual . The examples given cover estimates for specific basins
and generalized estimates, and include PMP esti mat es for th unde rstorms, general storms,
and tropical storms .
All procedures describ ed exc e pt one are based on th e meteorological , or tra-

di tionol, approach. The one exception is a statistical procedu re. The traditional
approach consiits essent ially of moistu re ma ximi zat ion and transposition of observed
storms . Wind maximi zati on is sometimes used. Storm transposition involves adjust-
ments for elevation, moisture- inflow barriers, and distance f r om the moistu re source,
These adjustments are founded on hypothetical sto r m models . A variation of the t ra-
ditional approach is the use of an orographic computation model in mo~ntainous re gions .
Methods are described for de termining the seasonal variation and chronological and
areal distribution of PMP .
Tables of precipitable water in a saturat ed pseudo- adiabatic atmosph e r e ar e

included for making various adjustments involving atmospheric moisture. Also includ-
ed are world record and near-record rainfalls that may be used for making rough assess-
ments of derived PMP estimates.
XIV SUMMARY
The manual was written under the assumption that t he user would be a meteoro-
lo gist . No attemp t was made to define or discuss basic meteorological terms or pro-
cesses. It is beli eved that the procedures described are presented in sufficient de-
tail to permit the professional meteorologist, especially one with hydrological train-
ing and ingenuity, to proceed with their application to t he usual problems involved in
estimating PMP.
RESUME
La hauteur maximale probable des precipitations (HMPP) est definie comme

etant la hauteur ma ximale de la l a me d'eau qui peut meteo r ologiquement s'accumuler
en un temps donne, sur un bassin donne, a une epoque determinee de l'annee, . sans
qu'il soit tenu compt e des tendances climatiques a long terme. Nos connaissances
actuelles sur le mecanisme des perturbations et sur l a quantite de precipitations
que celles -ci peuvent effectivement donner sont insuffisantes pour nous permettre
d'evaluer avec precision les valeurs ex tremes des precipitations exceptionnelles . Les
estimations de la HMPP doivent done etre considerees, tout au mains pour le moment,
comme des approximations . L'exactitude, ou la fiabilite, d'une estimation depend
fondamentalement de la quantite et de la qualite des donnees dont on dispose pour
appliquer les diverses methodes d'estimation.
Il n'est pas possible de normaliser les methodes employees pour estimer la

HMPP etant donne qu'elles varient en fonction de la quantite et de la qualite des
donnees disponibles, de la superfici e et de la situation du bassin, des caracteris -
tiques geographiques du bassin et de la region, de la nature des perturbations res-
ponsabl es des precipitations extremes et, finalem ent, du climat. Dans diverses par-
ties du monde, il existe de nombr e us es reg io ns pour le squelles la HMPP n'a encore
jamais ete estimee. Il serait impossibl e actuellement d'etablir un manu e l qui traite
de taus les problemes susceptibles de se poser . Il n' e st pas possible non plus de
preparer un manuel qui fass e e tat de toutes les situations rencontrees dons le passe
lors de l'etablissement d'estimations de la HMPP . Pour ces raisons, le present manue l
analyse les methodes qui se sont revelees generalement utilisables aux latitudes
moyennes pour des bassins d'une SUperficie egale OU inferieure a 50.000 km2, en region
montagneuse et en region de plaine.
Ces methodes sont exposees en recourant a de s exe mples d'etudes reellement

faites par le service meteor ologique national (anciennement le U.S. Weather Bur ea u),
l'Administration nationale de l'ocean et de l'atmosphere et le ministere du Commerce
des Etats-Unis. Plusieurs autres pays ont effectue des etudes tout aussi valables.
Les principales raisons qui ont motive le choix des exemples retenus sont : 1) ceux-
ci correspondent a differentes sortes de problemes, 2) ils sont tires d'etudes qui
ont ete publiees dans des rapports largement diffuses qu'il est assez facil e de se
procurer si l'on veut s'y referer, et 3) le temps de preparation du manuel et les
frais d'impression de celui-ci ont ete reduits du fait que l ' on disposait du materiel
necessaire, par exemple des cliches photographiques d'une grande partie des illus-
trations . Les exemples donnes concernent des estimations etablies pour des bassins
particuliers et de s estimations de caractere general; ces exemples comprennent des
estimations relatives a la hauteur maximal e des precipitations engendrees par des
orages, par des perturbations de type classique et par des temp e tes tropicales.
A une exception, toutes les methodes exposees sont fondees sur la technique
meteorologique dite traditionnelle . La seule exception concerne une methode statis -
tique. La technique traditionnelle consiste essentiellement a maximiser le contenu
en vapeur d'eau et a extrapoler en partant des averses observees. On a parfois
XVI RESUME
rec ours egalement a la maximalisation du vent. L'e xtrapolation a partir des averses
observees necessite de tenir compte de l 'a ltit ude , des barrieres s'opposant au trans-
port de la vapeur d'eau et de l'eloignement du lieu par rapport a la source de vapeur
d'eau. Les ajustements apportes pour tenir compte de ces parametres sont fondes sur
des modeles hypothet iques des averses. Une variante de la technique traditionnelle
consiste a utiliser, dons les regions montagneuses, un modele de calcul orographique.
Le manuel expose diverses methodes pour determiner la variation saisonniere ainsi que
la distribution spatio-temporelle de la HMPP.
Des tableaux de l'eau precipitable au sein d'une atmosphere pseudo -adiaba-

tique s aturee ont ete inclus pour permettre d'effectuer divers ajustements en ce qui
concerne le contenu de !'atmosphere en vapeur d'eau. Le manuel comporte egalement
une liste de valeurs record et quasi record de la hauteur des precipitations qui peut
servir a evaluer grossierement les estimations de la HMPP obtenues.
Le manuel a ete redige a l'intention des meteorologistes. Il n'a done pas

ete juge necessaire de definir et d'expliquer les termes et les processus meteorolo-
giques fondamentaux. On estime que les methodes decrites sont exposees suffisamment
en detail pour que les meteorologistes, pour peu qu'ils soient ingenieux et qu'ils
aient re~u une formation hydrologique, puissent les appliquer pour resoudre les pro-
blemes courants que pose l'estimation de la HMPP.
PE3IOME
BepoBTHhle M8Kel1M8nbHhle oe8AKl1 orrpeAensroTeH K8K Ha116onhm11A enoA oeaA-

KOB 38 ABHHyro ITPOAOn~11TenbHOeTb, MeTeoponor11qeeK11 B03MO~Hyro An H A8HHoro
6aeeeAHa B onpeAe neHHo e Bp eMH roAa 6e3 yqeTa AOnroepoqHhlX Kn11MaT11qeeK11x
TeHAeHu;11A. Cym,eeT Byrom,11 e 3H8Hl1B MexaHl13MOB IJ;lfilmHoB 11 11x n11BHeo6pa3y1om,eA 3cpcpeK -
Tl1BHOeTl1 HeAOeTaToqHhl, q To 6hl IT03BOnl1Tb rrpo113BeeTl1 ToqHyro ou;eHKY 3KeTpeManb-
HhlX oeaAKOB. B eB B311 e 3Tl1M ou;eHKa BepoBTHhlX MaKel1M8nbHhlX oeaAKOB, no
KpaAHeA Mepe B HaeTosm,ee BpeMB,AOn~Ha paeeMaTp11B8TbeB K8K np116n11~ e HH 8B.
ITp11MeHeH11e pa3n11qHhlX rr pou;eAYP AnH onpeAe n eH11s ToqHoeT11 11n11 HBAe~ H oeT11 ou;eH-
K11 38Bl1el1T B oeHOBHOM OT o6~ eMa 11 KaqeeTBa 11Merom,11xes ABHHhlX.
MeTOAhl ou;eHKl1 Be poBTHhlX M8Kel1M8nhHhlX oeaAKOB He MoryT 6hlTb eT8HA8P-

Tl1311pOB8Hhl, T8K K8K OHl1 113MeHBIDTeB B 38Bl1eMMOeTl1 OT o6~eMa 11 KaqeeTBa
11Merom,11xes ABHHhlX, pa3Mepa 11 reorpacp11qeeKoro nono•eH11s 6aeeeAHa, Tonorpa cp1111
6aeee AHa 11 paAoHa, xapaKTepa lliTOpMoB, A8IDIIJ,11X 3KeTpeManbHhle oe8AKl1, 11 Knl1-
M8T8. B pa3n11qHhlX qaeTsx 3eMHoro mapa eym,eeTByeT MHoro paAoHoB, AnH KO-
Tophlx Hl1KOrA8 He npo113B OAl1ne s paeqeT BepoBTHhlX M8Kel1M8nbHhlX oeaAKOB.
B Ha eTosm,ee BpeMB 6hlno 6hl HeB03MO~Ho noAroTOBl1Th noeo611e, oxBaThlBarom,ee
Bee npo 6n eMhl , KOTOphle MoryT 6hl Tb BeTpeqeHhl. Ehlno 6hl TBK~ e HeB0 3MO•Ho noA-
roT OBl1Tb noeo611e, KOTopo e OXB8 ThlBano 6hl Bee e11 Tya u;1111, KOTOphle B03Hl1K8nM
rrpM BhlBeAeHMM ou;eHOK B rrpomnoM. ITo 3TMM rrp11q11HaM B HaeTosm,eM noe o61111
paeeMaTp11B8roTeB Me TOAhl, KOTOphle,KaK rrpaB11no, 6hlnl1 npM3H8Hhl npMMeHMMhlMM
B epeAHMX m11pOT8X AnH 6aeeeAHoB p83MepaMM np11MepHo AO 50 000 KM 2 B rop-
HhlX 11 paBHMHHhlX paAoHax.
MeTOAl1K8 onMehlBaeTeB e noMOIIJ,bID rrp11M ep oB cpaKTMq eeKMX 11een e AO B8Hl1K,

npoB e AeHHhlX Hau;110HanhHOM MeTeoponor11qeeKoA e n y~6oA (paHee Ero po nor oAhl CWA ) ,
Hau;110HanhHo e ynpaBneH11e no oKeaHy 11 aTMoecpepe, ~enapTaMeHT ToproBnM CWA.
B H ee~onh Kl1 X eTpaHax 6hln11 npoBeAeHhl aHanor11qHhle 11eeneAO B8HMH. On11eaHHhle
rrp11Mephl 11enonh30B8nMeh no eneAYIDID,11M np11q11HaM: (1) B Hl1X np e AeTaBn e Hhl
pa3n11qHhle npo6n e Mhl, (2) OHM B3BThl 113 11een eAOB8Hl1M, KOTOphle 6hlnM orr11eaHhl
B m11p oKO paenpoeTpaHeHHhlX AOHn8A8X 11 KOTOphl e n erKo MoryT 11erronb30B8TbeH
AnH eehlnOK, ( 3) nerKo MO~HO rronyqMTb oeHOBHOM M8Tep11an, T8K OM KBK ~OTO
rpacp11qeeK11e oTneqaTKM MHor11 x 11nnroeTpau;11A, qTo yeKop11no 11 YAemeB11no noAro-
TOBKY HaeTo sm,ero rroeo611s. ITp11BeAeHHhle rrp11Mephl oxeaThlBaroT ou;eHK11 rr o
KOHKpeTHblM 6a ee e AHaM 11 o6o6m.eHHb!e ou;eHKl1 11 BKmoqaroT ou;eHK11 ,B epoH THb!X M8K -
eMM8nbHhlX oeaAKOB AnH rpo3, o6hlqHhlX n11BHeA 11 Tpon11qeeKMX nl1BHeA.
Bee on11eaHH hle enoeo6hl, 38 Me Knroq eH MeM OAHOro, oeHOBhlB8IDTeH Ha Me -

Teopo nor11qe eKo M,11n11 TPBAMIJ;MOHHOM n oAX OA e. EA11H e TB e HHhlM 11eKnroq e H11 e M HBnn -
eTeH eT8Tl1eT11qeeKMM MeTOA. Tp8AM IJ;l10HHhl M Me TOA eoeTOMT B oeHOBHOM B M8 K-
e MMM38IJ;l111 BnaroeoAep•aHMH 11 TpaHeIT03MIJ;l111 Ha 6nroA8 e MhlX nMBHeK. HH OrAa 11C-
nonh3yeTeB 11 M8KeMMM38IJ;MB BeTpa . TpaHeno311u;11s n11BHB rr peAyeMaTp11eaeT yYeT
XVIII PE3ID1VIE
BhlCOThl npenHTCTBHH ~na npHTOKa BnarH H paCCTOHHHH OT HCToqHHKa BnarH.

M e TO~hl
yqeTa 3THX ~aKTOpOB nonyqeHhl Ha OCHOBe rHn oTeTHqecKHX MO~eneM
nHBHeM. BapHaHToM T pa~H QHOHHoro no~xo~a HBnHeTCH npHMeHeHHe oporpa~H
qecKoM Mo~enH pacqeTa B ropHhlX paMoHax . OnHChlBBIDTC H MeTo~hl ~na onpe ~ en e
HHH cesoHHhlX H3MeHeHHM H pacnpe~eneHHH BepoHTHhlX MBKCHMannHhlX oca~KOB
Bo BpeMeHH H no nno~a~H.
~na yqeTa pas nHqHhlX ~aK TopoB , Kacaro~HXCH aTMOC~epHOM BnarH, np HBO-
~HTCH Ta6nHQhl o6~ e ro KOnHQ eC TBB nap a B HBChl~eHHOM ncee~oa~Ha6a T HQeCKOM
B TM OC~epe , KOTopoe MO• eT BhlnaCTb B BH~e oca~KOB. B noco6He BKnroqeHhl TBK•e
MHpOBhle peKo p~Hhle HnH 6nH3KHe K peKop~HhlM oca~KH, KOTOphle MoryT Hcnonn 30-
BBTbCH ~nH rpy6oM npoBepKH pacqeTHhlX 3HBQeHHM BepoHT HhlX MBKCHMBnbHhlX
oca~ Ko B.
Ti oco6 He COCTaBneHo B pacqeTe Ha TO, qTo HM 6y~yT nonn30B8TbCH Me-

Teopo norH . OcHoBHhle MeTeoponorHq ecKHe Te pMHHhl H npoQ eCChl He onpe~enaro T -
ca H He o6cy•~aroTCH. Tip e ~nonar ae TCH, qTo M e TO~hl onHC8Hhl ~OCTBTOQHO n o ~
po6 Ho , qTo6hl noseonHTb npo~ eccHo HannHOMY MeTeoponory, B oco6eHHOCTH MeTeo-
ponor y, npome~m e M y n o ~roToBKY no rH~ponorHH, npHM eHH Tb 3TH MeTo~hl K o 6hlqHhlM
npo6neMaM, Kacaro ~HMCH OQeHKH Be poHTHhlX MBKCHMannHhlX oca~KOB.
RESUMEN
La precipitacion maxima probable se define como la mayor cantidad de pre-

cipitacion meteo r ologicamente posible que corresponde a det e rm i nada duracion en una
cuenca dado y e n determinada epoca del ano, sin tener para nada en cuenta las t e n-
dencias climaticas que s e producen a la r go plazo. Los conocimi e ntos que actualmen -
t e se poseen sobre el me canismo de los t e mporal e s y su e ficacia pa r a pr oducir preci -
pi taciones resultan insuficientes para poder evaluar con precision los limites de lo s
valores extremos de la precipitacion. Las estimaciones de la pr e c ipit a cion ma xima
probabl e hon de ser pues consideradas, al menos por el momento, como apro ximaciones .
La precision y seguridad de una estimacion depende fundamentalmente de la cantidad
y calidad de los datos disponibles para su aplicacion a los diferentes pr ocedimien-
tos de estimacion.
Los procedimi e ntos de estimacion de la precipitacion maxima probable no

pueden ser normalizados ya que va rian con la cantidad y calidad de los da tos dispo -
nibles, con el tamano de la cuenca y su emplaza miento, con la topografia de la cuen-
ca y de la region, con los tipos de temporales que producen precipitaciones extremas
y con el clima . Existen numerosas regiones en varias partes del mundo en las que
jamas se ha estimado la precipitacion maxima probable. Se r i o imposible en este mo-
mento redactor un manual en donde se estudiaran todos los pr oblemas que a este res -
pecto pueden plantearse. Tampoco seria factible resumir en un manual todas las si -
tuaciones que se plantearon al deducir las estimaciones anteriores. Por estos mo-
tivos, en el presente Manual se estudian los procedimientos que se consideran de apli-
cacion general en las latitudes medias en las cuencas cuya extension sea de hasta
50 . 000 km2 aproximadamente, en regiones montanosas y llanos.
Los procedimientos se desc r iben en forma de ejemplos tomados de los estu -

dios reales ll e vados a cabo por el Se r vicio Meteorologico Nacional de Estados Unidos
de America, dependientes de la Administracion Nacional del Oceano y de la Atmosfera,
del Departamento de Comercio . Varios poises hon efectuado tambien valiosos estudio s.
Las principales ra zones que hon motivado la utilizacion de los ejemplos a ntes descri -
tos son los siguientes : 1) estos ejemplos representan diferentes problemas; 2) hon
sido tomados de estudios publicados en informes de amplia distribuci6n que resultan
de acceso bastante facil para usarlos como referencia; y 3) facil disponibilidad de
la documentaci6n basica, tal como las fotografias de numerosas ilustraciones y reduc -
cion al minimo de los gastos y tiempo necesarios para la preparacion de este Manual.
Los ejemplos citados se refieren a estimaciones para determinadas cuencas asi como
de caracter general y en ellos se incluyen valores estimados de la precipitaci6n ma-
xima probable procedente de las tormentas, de los temporales en general y de las tor-
mentas tropicales.
Todos los procedimientos descritos, excepto uno, est6n fundados en plantea-

mientos meteorologicos o tradicionales. La unica excepcion es el procedimiento es-
tadistico. El planteamiento tradicional consiste fundamentalmente en la maximizacion
de la humedad y en la transposici6n de los temporales observados. Algunas veces se
xx RESUMEN
utiliza tambien la max1m1zac1on del viento. La transposici6n de los temporales im-

plica ajustes de altitud, de las barreras contra el flujo de humedad entrante y de
la distancia a partir de la fuente de humedad. Estos ajustes se fundan en modelos
de temporales hipoteticos. Una de las variaciones del planteamiento tradicional con-
siste en la utilizaci6n de un modelo orografico de calculo, aplicable en las regiones
montanosas. Se describen metodos para determinar la variaci6n estacional y cronol6-
gica asi como la distribuci6n zonal de la precipitaci6n maxima probable.
Se incluyen tambien tablas de agua precipitable en una atm6sfera saturada

pseudoadiabatica, con objeto de hacer varios ajustes en los que interviene la humedad
atmosferica. Tambien se incluye un registro mundial y un registro aproximado de la
lluvia que pueden ser utilizados para hacer evaluaciones no muy aproximadas de las
estimaciones deducidas de la precipitaci6n maxima probable.
Este Manual ha sido escrito suponiendo que el usuario es meteor6logo. No

se ha trata do de definir o discutir los terminos o procesos meteo rol6gicos. Se cree
que los procedimientos descritos hon sido expuesto con detalle suficiente para que
el meteor6logo profesional, especialmente con formaci6n hidrol6gica y dotado de cier-
ta iniciativa, pueda aplicarlos a los problemas habituales que se plantean para esti-
mar la precipitaci6n maxima pro bable.
C HA P T E R l
INTRODUCTION
1.1 Definitions of probable maximum precipitation (PMP)
1.1.l
The use of meteorological knowledge to de r ive limiting precipitation values

for hyd r ological design purposes began to gain fa vour in th e middle 1930's . There ore
varying degrees of limiting design values depending on the purpos e fo r which they ore
required. Precipitation associated with the uppermost limits is known as the proba-
ble maximum precipitation ( PMP), which is defined [3=/ as the theo r etically gr ea test
depth of precipita tio n for a give n du r ation that is physically possible ove r a pa r~
ticulor drainage basin at a particu l ar time of year. Such is the conceptua l defini -
tion of PMP. The values deri ved as PMP under this definition ore subject to change
as knowledge of t he physics of atmospheric processes increases. They ore also sub-
ject to change with long-term climatic variations, such as would result from changes
in solar radiation intensity. Climatic trends, however, progress so slowly that
their influence on PMP is small compa r ed to other unce r tainties in estimating these
extreme values. Climatic trends ore therefore ignored.
1 . 1.2
In addition to the conceptual definition of PMP, on operational definition

may be considered as consisting of the steps followed by hydrometeorologists in arr iv-
ing at the answers supplied to engineers for hydrological design purposes . Whatever
the philosophical objections to the concept, the operational definition leads to
answers that hove been examined thoroughly by competent meteorologists and enginee r s
and judged as meeting the requirements of a design criterion. The result of apply-
ing the operational definition over on entire region is to approach uniformity in
design, safety and cost.
1.1 . 3
Probable maximum pr ecipitation (PMP) was once known as maximum possible pre-
cipitation (MPP), and this lotter term is found in most reports on estimates of ex-
treme precipitation mode prior to about 1950. The chief reason for the name change
to PMP was that MPP carried a stronger implication of physical upper limit of pre-
cipitation than does PMP, which is preferred because of the uncertainty surrounding
any estimate of maximum precipitation. Procedures for estimating PMP, whether mete-
orological or statistical, ore admittedly inexact, and the results ore approximations .
Different, but equally valid, approaches may yield different estimates of PMP . Fo r
this reason various levels of PMP may be considered, as discussed in section 1.2.
2 ESTIMATION OF PROBABLE MAXIMUM PRECIPITATION
1.1.4
PMP for all durations and sizes of area in a specific basin is usually deter-
mined by several types of storms . For example, thunderstorms are very likely to pro-
vide PMP over an area smaller than about l 000 km2 for durations shorter than 6 hours,
but controlling values for longer du rati ons and larger areas will be derived almost
i nvari ably from general storm s. For short durati ons , thundersto rms can produce heavi -
er ra infall than can gene ral storms, but they are relatively short-lived, and indi-
vidual storms cover relatively small areas . General storms, although they often in-
clude thunderstorms, produce less intense rainfall on the average, but the ir longer
l ife and gr eater areal coverage resul t in greater ra infall amounts fo r du rations of
about 6 hours and longer, and for large areas.
Norma lly, it wou ld appear illogi cal to assume that PMP fo r all durations and
sizes of area co uld be realized from one storm, but this is not necessarily so. PMP
for small basins may be, and i s often ass umed to be, obtainable f r om a single stor m.
In such case s, PMP and PMS are synonymous , but this is not always so . PMP va lues fo r
al l ranges of duration and sizes of area in a basin are always understood to represent
limi ting ra in fal l am ounts without regard to storm type. In other words, PMP values
enve l op t he probable maximum amounts that might be realized from any t ype of storm
that could produce heavy precipitation over the basin. PMS, on the othe r hand, may
refer to any maximized observed or hypothetical storm that is equal to PMP for at
least one duration and size of area. The term has been applied also to a hypotheti-
cal storm that would produce PMP for all durations at the total basin area and somewhat
lesse r values for smaller areas within the basin.
1.2 Lower and upper limits of PMP
1.2.1
That the procedur es described he re for deriving estimates of PMP yi eld results
to the nearest millimeter or tenth of an inch should not be taken as on indication of
the degree of ac curac y of the estimates. There is no objective way of assessing the
general level of PMP estimates derived by the procedures described here or by any ot her
known procedures. Judgment based on meteorology and experience is most important.
Obviously, estimates subsequently exceeded by observed storm r ainfall were too lo w.
There is no way, however, that an estimate can be labelled with certainty as being too
low or too high at the time it is mode. Thei r accuracy may be assessed, however, by
consideration of the following factors: (1) excess of estimated PMP over the max imum
observed rainfall values for the project basin and surrounding region; (2) number and
severity of record storms; (3) limitations on storm t ra nsposition in the region; (4)
number, character, and interrelationship of maximizing steps; (5) reliability of any
model used for relating rainfall to other meteorological variables; and (6) probabili-
ty of occurrence of the individual meteorological variables used in such models, with
care being taken to ovoid excessive compounding of probabilities of rare events.
Subsequent chapters show that various steps in the procedures require mete-
orological judgment. Consequently, the resulting estimates con be conservative or
liberal depending on decisions affecting the degree of maximization used in their de-
rivation. Thus, in effect, lower and upper limits to PMP can be estimated, although
only one set of values is usually derived.
INTRODUCTION 3
1.2.2 Confidence bands
The delineation of lower and upper limits to PMP is somewhat analogous to the
confidence bands used in statistical work. It would be nice if a confidence band
could be placed about a PMP estimate in an objective manner, similar to the standard
statistical method, but this is not possible because PMP is not estimated by formal
statistical methods. This limitation, however, does not invalidate the concept of a
confidence band, but it means that its limits must be based in considerable measure on
judgment, as is the PMP estimate itself . Factors influencing such judgment are the
same as those for assessing the general level of PMP listed in th e preceding paragraph.
1.3 The manual
l. 3.1
The following statement was published in a UN/WMO report [4}: "The practice
of hydrometeorology has not been r educed to a handbook. No one can furnish a set of
rules, graphs, and procedures whereby one can proceed step by step and necessarily de -
rive an acceptable estimate of probable maximum rainfall. The lectures will discuss
only certain principles. Handbooks work best in solving uniform problems from data
that ore uniform and ample. None of these three conditions is the rule in probable
maximum rainfall estimates - neither problems nor data are uniform, and the data are
certainly not ample."
There is no disagreement with this statement . No two basins present exactly

the same problems as they vary in size, shape, orientation and other geographic fea-
tures. Also, the type, amount and quality of meteorological data available differ
from basin to basin. Nevertheless, it is believed that a useful purpose would be
served by some description, in as much detail as possible, of the more commonly used
procedures for estimating PMP . It is for this reason that this manual has been pre-
pared. With the procedures and examples presented here, the hydrometeorologist with
some ingenuity should be able to make the necessary modifications to adapt the pro-
cedures to his particular problems.
1.3.2 Scope
The manual describes the more common meteorological approaches for estimating
PMP in orographic and non-orographic regions and for regions with and without adequate
meteorological data. It also describes a statistical procedure for small basins.
Generalized estimates, storm transposition, and seasonal variation are discussed also.
The text is restricted to methods for estimating PMP, and does not include procedures
for deriving maximum seasonal snowfall accumulations, optimum melting rates, etc.
Estimation of PMP for very large basins is usually a complicated problem

[2, 3_}. The continued deposit for several days throughout an area of hundreds of
thousands of square kilometers of precipitation at a rate computed from a sustained
maximum Jnflow of moist air with maximum moisture content, and released by repeated
development of storm mechanisms of maximum efficiency, would be many times greater
than what is experienced in the situations producing the maximum floods, and would be
an unreasonably excessive estimate of maximum precipitation. For this reason, the
various meteorological procedures described here are considered most applicable to
basins up to about 50 000 km2, although they have been used for much larger basins.
The meteor ological procedures discussed are more suited to middle latitudes
than to the tropics. In the tropics the heaviest rainfalls are associated with very
high atmospheric moisture, which prevails most of the time during the rainy season.
Hence, there is neither theoretical nor empirical r eason to relate rainfall potential
to the mino r fluctuations in humidity that occur. It is for this reason that the
meteorological procedures presented are conside re d to be generally inapplicable to the
derivation of PMP estimates for the tropics.
1.3 . 3.
Examples from published reports on e sti mates of PMP fo r various basin sizes
in regions with different climates and topography ore used in the following chapters
to describe the more generally applicable procedures for making such estimates. There
are two main reasons for using such examples. One is that they are real estimates
mode for real situations, and thus should inspire more confidence in the procedures
than would descriptions of hypothetical situations. The second rea son is that the
published r eports from which the examples were taken provi de more detail than can be
given in this manual, and they are available for reference. The information pre-
sented in this manual, however, should be adequate for describing procedures. While
estimates of PMP have been made by various countries, the examples used are from re-
po r ts published by the U.S. Weather Bureau, renamed National Weather Service in 1970 .
It should not be inferred that the procedures and results presented in these reports
are superior to those derived by other agencies or nations. The chief reason for using
examples from reports prepared by the Weather Bureau (now National Weather Service)
is that it hos issued published reports, particularly in its HydrometeorologicalReport
series, giving detailed descriptions of over four dozen PMP studies made for various
ports of the world. Most, of course, are for the United States, which, because of
its wide variety of climate and topography, presents a wide range of problems involved
i n the derivation of PMP estimates. Some reports are on studies made for specific
river basins, while others present generalized estimates. Both types are discussed
here.
The examples presented are not intended for direct application in deriving PMP
estimates. They serve merely to show how PMP has been estimated in a number of dif-
ferent situations involving different basin sizes, topography, climate, and data avail-
ability. It should not be inferred that the example given for any particular situa-
tion represents the only solution. Equally valid results migh t have been obtained by
other approaches. The examples should thus be looked upon as sug~estions on how to
approach derivation of PMP estimates. Particular attention should be paid to the
cautionary remarks at the end of each chapter.
Although barely mentioned in the manual, the importance of meteorological

studies in preparing PMP estimates cannot be over-emphasized. Such studies give
guidance to regional, seasonal, durational, and areal variations and to topographic
effects.
INTRODUCTION 5
References
l. American Meteorological Society, 1959: Glossary of Meteorology p. 446.
2. Myers, V. A., 1959: Meteorology of hypothetical flood sequences in the Missis-

sippi river basin. Hydrometeorological Report No . 35, U. S. Weather Bureau.
3. Schwarz, F. K., 1961: Meteorology of flood-producing storms in the Ohio river

basin. Hydrometeorological Report No. 38, U.S. Weather Bureau.
4. United Nations/World Meteorological Organization, 1967: Assessment of the magni-

tude and frequency of flood flows. Water Resources Series No. 30, p. 15.
C HA P T E R 2
ESTIMATES FOR NON-OROGRAPHIC REGIONS
2.1 Introduction
2.1.l
The theoretical interrelationship of donvergence, vertical motion and con-

densation is well known. If the convergence at various heights in the atmosphere or
the vertical motion (averaged over some definite time and space) is known or assumed
with a given degree of precision, then the other can be calculated to an equal pre-
cision from the principle of continuity of mass .
Observations confirm that the theoretical pseudo-adiabatic lapse rate of

temperature of ascending saturated air from which precipitation yield is calculated is
closely approximated in deep precipitating clouds. The higher the specific humidity,
the greater the precipitation yield for a given decrease in pressure. All these
factors are basic to the formulation of a convergence model, and several such model s
have been postulated ~' 10, l.lj7.
2.1.2
There is a problem in estimating probable ma xi mum precipitation (PMP) with a

convergence model. Maximum water vapour content can be estimated with acceptable
accuracy for all seasons for most parts of the world by appropriate interpretation of
climatological data. However, there is neither an empirical nor satisfactory theo-
retical basis for assigning maximum values to either convergence or vertical motion.
Direct measurement of these values has been elusive. The solution to this dilemma
has been to use observed storm rainfall as an indirect measure.
Extreme rainfa lls are indicators of maximum rates of convergence and vertical
motion in the atmosphere, which are re fe rred to as the storm, or precipitation-produc-
ing, mechanism . Extreme mechanisms for extreme storms may then be determined fo r
basins under study without the necessity of actually calculating the magnitude of t he
convergence and vertical motion . The procedures used for maximizing observed storm
rainfall to estimate PMP involve moistu re adjustments, storm transposition and envelop-
ment, and these are discussed in the following sections.
2.2 Estimation of atmospheric moisture
2.2.l
Since many of the extreme, or major rec orded storms occurred before extensive
networks of upper-air temperature and humidity soundings had been e stablished , any in-
dex of atmospheric moisture must be obtainable from surface observations. Also,
curren t upper-air observational networks are still too spa rse to define adequately the
moisture inflow into many storms, especially those limited to areas of the size con-
sidered in this report .
Fortunately, the moisture in the lower layers of the atmosphere is that most
important for producing precipitation, both because most of it is in the lower layers
and because it is dis tr ibuted upward through the storm early during rainfall [5 , 7_}.
Theoretical computations show that, in the case of excessive rains , ascensional ra tes
in the storm must be so great that within an hour or so air or iginally near the sur-
face has reached the top of the layer from which precipitation is falling . In the
case of severe thunde r storm ra infall, surface air may reach the top in a matter of
minutes.
The most realistic assumption seems to be that the air ascends dry-adiabati-
cally to the satu r ation level and thence moist- adiabatically . For a given surface
dew point, a column of air will contain more moisture the lower the level at which the
air r eaches saturation, the greatest precipitable water occurring when this level is
at the ground . For these reasons, hydrometeorologists generally postulate a satur-
ated pseudo-adiabatic atmosphere for extreme storms.
2.2 . 2
Moisture maximization of a storm requires identification of two saturation

odiobots . One typifies the vertical temperature distribution in the storm to be
maximized, and the other is the warmest saturation odiobot to be expected at the same
place and time of year as the storm. It is necessary to identify these two soturoy
tion odiabats with some indicator, and the conventional label in meteorology for satu-
ration odiobats is the wet-bulb potential temperature, which corresponds with the dew
point at l 000 mb . Tests have shown that storm and extreme values of precipitable
water may be approximated by estimates based on surface dew points when saturation and
pseudo-adiabatic conditions are assumed (7}.
Surface dew points representative of the moisture inflow into the storm
identify the storm saturation adiabat. The moist odiobot corresponding to either the
highest recorded dew point for the location and season or the dew point for some speci-
fic return period, soy, 100 years, is considered sufficiently close to the probable
warmest saturation adiabat. Both storm and maximum dew points are reduced pseudo-
adiabatically to the l 000 mb level (Figure 2.1) so that dew points observed at sta-
tions at different elevations are comparable. This permits construction and use of
tables showing atmospheric moisture as a function of the l 000 mb dew points (Annex 1).
2.2.3
As- the moisture hos an appreciable effect on the storm, precipitation must be
that which persists for hours rather than minutes . Also, any single observation of
dew point may be considerably in error. There is, then, merit in basing dew-point
values to be used in estimating storm and probable maximum moisture on two or more con-
secutive measurements separated by a reasonable time interval or a continuous auto-
matic record of dew point over a period of time rather than on a single reading. The
so-called highest persisting 12-hour dew point is generally used. This is the highest
ESTIMATES FOR NON-OROGRAPHIC REGIONS 9
4,0
3, 5
3.0
2.5
........
]
'-' ·
.µ 2.0
t1
•.-!
Q)
~
1. 5
LO
0.5
Temperature (oc}
Figure 2.1 - Pseudo-adiabatic diagram for dew-point reduction

to l 000 mb at height zero
value equalled or exceeded at all observations during a 12-hour period. For example,
the following is a series of dew points observed at 6-hour intervals:
Time: 00 06 12 18 00 06 12 18
Dew point (°C): 22 22 23 24 26 24 20 21
The highest persisting 12- hour dew point for this series is 24°C, which is
obtained from the period 18 to 06. However, if the air temperature had dropped below
23°C during the period 00 to 06, the highest persisting 12-hour dew point would then
be 23°, which is obtained from the period 12 to 00. Hourly dew points may be used, of
course, but such records are sparse, and they add a great deal of work to the surveys
for persisting values, especially in the case of maximum persisting 12-hour dew points,
which are discussed in section 2.2.5.
2.2.4
To select the saturation adiabat representing the storm moisture, the highest
dew points in the warm air flowing into the storm are identified from surface weather
charts. Dew points between the rain area and moisture source should be given primary
consideration . Dew points in the rain area may be too high because of the precipita-
tion, but they need not be excluded if they appear to agree with dew points outside the
area. In some storms, particularly those with frontal systems, surface dew points in
the rain area may represent only a shallow layer of cold air and not the temperature
and moisture distributions in the clouds releasing the precipitation.
14
•
Heavy rain area
16
•
19 .
•
Figure 2.2 - Determination of maximum dew point

in a storm. Representative dew point for this
map time is average of values in boxes
Figure 2.2 illustrates schematically a weather map from which the storm dew
point is determined. On each consecutive weather map, say, for 6-hour intervals dur-
ing t he storm, the maximu m dew point is averaged over several stations, as illustrated
in t he figure. Occasionally, for lack of data, it is necessary to rely on the dew
point at only one suitably located station. The single or average maximum dew points
selected from eac h map for m a series, and the max i mum persisting 12-hour storm dew
point is then selected, as described in section 2.2.3. The selected dew point is then
re duced pseudo-adiaba ti ca lly to the l 000 mb level.
If the originally observed values plotted on the we ather maps are for stations
differing appreciably in elevation, the redu ction to l 000 mb should be made be fo re
averaging . However, elevation differences between dew-point stations in the moist-air
inflow are usually small and are generally neglected in the selec t ion of the storm dew
point.
2. 2.5 Maximum persisting 12-hour l 000 mb dew points

----------------------------------------------
Maximum values of atmoipheric water vapour used for storm maximization are
usually estimated from maximum persisting 12-hour l 000 mb dew points. These dew
points are generally obtained from surveys of long records, say 50 or more years, at
several stations in the problem area. In some regions, the maximum dew points for
each month of the year or critical season may be adequate to define the seasonal varia-
tion of maximum atmospheric moisture, but it is generally advisable to select maximum
12- hour dew points by semi - monthly or 10-day intervals .
Dew-point records appreciably shorter than about 50 years are unlikely to

yield maximum values representative of maximum atmospheric moisture . The usual
practice in such cases is to make a frequency analysis of the annual series of mon th ly
or shorter interval maximum persisting 12-hour dew points. Since values for the 100-
year return period have been found to approximate maximum dew points obtained from su r-
veys of long r ecords, it is the 100-year values that are generally used for defining
the seasonal variation curve, although 50-year values are sometimes used.
Certain precautions are advisable in the selection of maximum dew points in-
tended to be indices of maximum moisture for storm maximi zation. These precautions
apply regardless of whether the maximum dew points are used directly as surveyed or
subjected to frequency analysis. In certain places and seasons characte r ized by
ample sunshine, sluggish air circulation, and numerous lakes, rivers and swamps, a
local high dew point may result f r om local evaporation of moisture f r om the surface
and may not be at all representative of atmospheric moistu re at upper levels . Such
dew points should be discarded. To eliminate dew points so affected, the surface
weathe r charts for the dates of highest dew points should be examined and the dew
points discarded if they appear to have occurred when the observing station was clear-
ly in an anticyclonic or fair weather situation rather than in a cyclonic circulation
with tendencies towards precipitation.
All values of maximum persisting 12-hour dew points selected directly from
surveys of long records are plotted against date observed, and a smooth envelope
drawn, as illustrated in Figure 2 . 3 . When dew points from sho r t records are subject-
ed to frequency analysis, the resulting values are ~sually plotted against the middle
28
•
•
26 • •
,,-...
•
u
0
• •
+'
.,.,i::: 21. •
0
p,
)I
GJ
A
22 Note: 20 July dew point found to have been observed

during fair we ather situation and is und e r c ut.
5 15 25 5 1525 5 15 25 5 1525 5 1525 5 1525 5 1525 5 1525
Figure 2.3 - En veloping maximum dew points at a station
day of t he interval for which the series is compiled. Thus, for example, if t he fre -
quency analysis is for the series of semi- monthly maximum persisting 12-hour dew po ints
observed in the first half of the month, the result ing 50- or 100-year va lues would be
plotted against th e ei ghth da y of the month.
The preparation of monthly maps of maximum per sisti ng 12- hour l 000 mb dew
points is advisable, especially where numerous estimates of PMP are required. Such
mops not only provide a ready, convenient source of maximum dew points but also aid in
main taining consistency between estimates for various basins. The maps are based on
mid-month dew-point values read from the seasonal variation curves and adjusted to the
l 000 mb level . These va lues are plotted at the locations of the observing stations,
and smooth isopleths are then drawn, as in Figure 2.4 .
Some regions have no dew-point data, or a period of record so short as to pre-

clude reliable frequency analysis. Since the chief source of moisture inflow into
ma jor storms is water evaporated from the seas or oceans, sea-surface temperatures pro-
vide a logical base fo r estimating maximum dew points. In fa ct, sea-surface tempera-
tures may be more representative of atmospheric moistu re in depth t han are inland dew
points, which, as mentioned earlier, may be affected by local c onditions .
Estimation of maximum dew points f r om sea-surface temperatures is relat i vely

simple for coa stal regions since there is little modification of the moist air by
passage over land surfaces. In the coasta l regions of the Gul f of Mexico, for example,
maximum persisting 12-hour l 000 mb dew points ra nge f rom about 1°C to 2°C below upwind,
offshore mean monthly sea-surface temperatures. The difference increases with dis-
tance inland.
Figure 2,4 - Maximum persisting 12-hour l 000 mb dew points

for August
The rate of decrease of maximum dew points with distance inland depends upon
the season of the year, direction of moisture flow during periods of maximum humidity,
topographic barriers, and other geographic factors. The decrease must be determined
for each month and for each region of interest in order to obtain a reasonably reliable
seasonal variation curve. The gradients indicated by maps of maximum persisting 12-
hour l 000 mbdew points prepared for areas with adequate data provide the most useful
guidance in determining such dew points for areas with very little or no data . The
map of Figure 2. 4, for example, would be useful for estimating maximum persisting dew
points for regions of similar geography .
2.2.6 Precipitable water

------------------
This is a term, used mostly by hydrometeorologists, to express the total mass
of water vapour in a vertical column of the atmosphere . A statement, for example,
that the air contains 3 cm of precipitable water signifies that each vertical column
of l cm2 cross section contains 3 gm of water in vapour form . If the water vapour
were all condensed into liquid water and deposited ~t the base o'. the colu~S' the ac-
cumulated liquid would be 3 cm deep, since the density of water is l ~m cm •. _Pre-
cipitable water is, in fact, a misnomer, because no natural process ~ill precipitate
all the water vapour in the atmosphere. For this reason, the.substitute terms,.
liguid eguivalent of water vapour or, simply, liguid water equivalent, are sometimes
used.
Tables of precipitable water for saturated air with a pseudo-adiabatic lapse

rate between the 1 000 mb surface and various heights or pressure levels as a function
of the 1 000 mb dew point are presented in Annex 1. These tables are used for mois-
ture adjustments.
2.3 Moisture maximization
2.3.l Seasonal limitation;
Seasonal variations in storm structure place a limitation on moisture maxi-

ization . For example, a wi nte r storm would never be adjusted for the moisture content
indicated by the maximum persisting 12-hour dew point fo r the year if it should be in
summer, which it almost always is . In practice, the moistu re ad j ustments are made on
the basis of the maximum persisting 12-hour dew point fo r the same time of year as the
storm occurrence or , more often, the maximum persisting 12-h our dew point within 15
days. Thus, for example, if the maximum dew point for maximizing a 15 May storm was
being selected from the curve of Figure 2.3, one would use the higher dew point in-
dicated for 30 May . Similarly, the maximum dew point indicated for 15 Septembe r wou ld
be used generally for maximizing a 30 September storm.
2.3.2
The t ab les presented in Annex l show depth of precipitable water f rom the
l 000 mb surface to various alt itudes or pressure levels as a f uncti on of the l 000 mb
dew point. In maximiz i ng storm rainfall, only the depth of precipitable water from
the ground to some arbitrarily selected l evel from 400 to 200 mb is used. The 300 mb
level is accepted generally as the top of the storm, but it makes little difference
which level f rom 400 mb on up is selected, as there is very little moisture at those
heights, and the effect on the moisture adjustment is negligible. In cases where a
mountain barrier lies between the storm area and the moisture source, the mean eleva-
tion of the ridge, or crest, is generally selected as the base of the moisture column.
In most cases, it is advisable to select the storm and maximum dew points between the
barrier and the storm location.
2.3 . 3
The dew points from a single station or set of stations used to obtain a rep-
resentative persisting 12-hour storm dew point are unlikely to be in the most intense
moisture inflow for much more than 12 to 24 hours, after which the stations where the
dew points were observed are very likely to be in the cold air because of the displace-
ment of the storm. The seleition of different repres entative 12-hour dew points for
every 12 hours of a storm is a very tedious task, especially for storm durations of 72
hours and longer. Comparisons of storm rainfall values adjusted on the basis of 12-
hour dew points from different sets of stations and those from a single set indicate
that differences are too small to justify the additional time required for obtaining
representative 12-hour dew points for different storm intervals.
It should be noted also that the use of different representative dew point~
for a storm requires different maximum dew points for the maximizing procedures des-
cribed below. Tests of the use of re presentative storm dew points over time intervals
up to 72 hours, e.g., 24- , 48- and 72-hour dew points, for adjusting storm r ainfall
values showed only small differences from the results obtained from the use of the 12-
hour representative storm dew point. The general practice is to use a single repre -
sentative persis t ing 12-hour dew point for adjusting the storm rainfall for all dura -
tions and sizes of area.
2.3.4 Maximization of storm in place

------------------------------
Moisture maximization of storms in place, i.e., without change in location,
consists simply of multiplying the observed storm rai nfall amounts by the ratio ( r )
of the maximum precipitable water (W ) indicated for the storm location to the pre-m
cipitable water (W ) estimated for t~e storm, or
s
r
m
= Wm/w s
(2.1)
Thus, for example, if the representative persisting 12-hour l 000 mb storm dew point
is 21°C and the m~ximum is 24°C and the rain area is at an elevation of 400 m above
mean sea level (always assumed to be at l 000 mb) with no intervening topographic
barrier between the rain area and moisture source, the mois ture maximizing ratio (r )
is computed from precipitable water values obtained from the tables in Annex l: m
W = 74 - 8 = 66; W = 57 - 7 = 50; and r = 1.32. The precipitable water values
m s m
used in determining W and W are for a moisture column with base at l 000 mb and top
at 300 mb minus the p~ecipi¥able water in a column with base at l 000 mb and top at the
elevation of the rain area, i.e., 400 m.
If it is now assumed that there is ·an extensive, relatively unbroken range of

hills with a mean crest elevation of l 200 m, m.s .l., between the rain area and moisture
source, r would then be determined as follows : W = 74 - 23 = 51; W = 57 - 19 = 38;
m m s
and rm = 1.34. Here, the precipitable water in the l 000 to 300 mb column 1s
decreased by that in a column with a base at l 000 mb and top at l 200 m, i.e., the
elevation of the barrier crest and not that of the rain area. Whenever possible,
however, representative storm dew points on the lee side of the barrier should be used.
This is especially advisable in the case of local storms, which do not necessarily
require a strong, widespread moisture inflow but may utilize moisture that may have
seeped into and accumulated in the storm area during an interval of several days or
longer of sluggish circulation prior to the storm.
Moisture maximization in transposing storms 1s somewhat more complicated and

1s discussed in section 2.5.
2.4 Wind maximi zation
2.4.l Introduction
Wind maximization is most commonly used in orographic regions when it appears

that observed storm rainfall over a mountain range may vary in proportion to the speed
of the moisture-beari ng wind blowing against the range. Wind maximi zat ion in such
regions is discussed in sections 3.3.l.l and 3.3.1.2. In non -orographic regions, wind
maximization is used only infrequently; storms can be transposed hundreds of kilo-

metres to synthe s ize an adequate storm history for a project basin. It is reasoned
that moisture inflow rotes recorded in extreme storms are at a maximum or near-maximum
for precipitation-producing effectiveness, and there is generally no need to maximize
wind speeds .
This reasoning appears logical since storms with the highest wind speeds do
not necessarily produce the most intense prec ipita tion . While it is true that hu rri-
canes, or typhoons , with their high wind speeds tend to produce heavier rainfall
amounts than do the most vigorous extratropical storms, it should be noted that their
moisture content i s much higher. Also, whether hurricanes with the highest wind
speeds produce more ra infall than weaker hurricanes is uncertain, since they generally
reach full strength over seas . It is known, however, that rainfall from hurricanes
over land is not proportional to their wind speeds.
2.4.2
Wind maximization is sometimes used in non-orographic regions when moisture

adjustments alone appear to yield inadequate or unrealistic results. In regions with
limited hydrometeorologicol data, for example, wind maximization may be used to com-
pensate portly for the sho rt period of record. The reasoning here is that the limited
data available ore unlikely to include extreme values of dew points or outstanding
storms equivalent to those that would be observed over a long period of record. The
heaviest storms recorded may be relatively weak, and their moisture inflow rates are
likely to be l ess than those associated with maximum precipitation-producing effective-
ness. Increasing both wind and moisture yields a higher degree of maximization than
would moisture adjustment alone, and this compensates, in port at least, for on in-
adequate sample of observed data.
Wind maximization is sometimes used also when the seasonal variation of

ITTaxi mum 12-hour dew points gives a false indication of the seasonal variation of PMP.
This is most likely to occur in regions where summers ore dry and all major storms are
experienced in the cold half of the year . The dew- point curve almost always peaks in
summer, .and the seasonal variation of maximum wind speeds must be considered in devel-
oping a representative sea sonal variation curve of PMP (sections 2.10.3 and 2.10.4) .
In coses where this is done, individual storms ore maximi zed for both moisture and
wind, as described in sections 2.4.3, 2.4.4 and 2.9.2.
2.4 . 3 Winds representative of moisture inflow in storms

-------------------------------------------------
Low-level winds are generally used to estimate moisture inflow in storms be-
cause most of the moi sture usually ente rs the storm system in the lowest l 500 metres.
The winds in this bottom layer can be obtained from pilot-balloo n or rawinsonde ob-
servations, the winds at 1 000 and 1 500 metres perhaps being the most representative
of moisture inflow. Upper-air observations, however , have r elatively short records
and cannot be used for maximizing the olde r storms. Also, pilot- balloon observations
cannot be made in storms . Another shortcoming of upper-air wind observations is that
they are made at considerably fewer stations than are surface-wind observations and are
often inadequate for determining moistu re inflow into small-area storms. For these
reasons, surface data are generally used as an index of wind movement in the cri tical
moisture-bearing layer.
2.4.3.l Wind direction
The first consideration in developing wind adjustments is the wind direction

associated with moisture inflow during major storms. Only winds from critical direc-
tions are considered in deriving wind-adjustments ratios. If more than one direction
provides moist-air inflow, separate seasonal maximum wind-speed curves should be con-
structed for each direction. This is particularly advisable if the different wind
directions bring in mois ture from different source regions.
Various measures of wind speed have been used to develop wind maximization
ratios. Among them are: (1) average wind speed through the moisture-bearing layer
computed from representative soundings; (2) average speed in the moist layer computed
from two or three consecutive 6- or 12-hourly soundings; and (3) average speed or
total wind movement for a 12- or 24-hour period at a representative station, the 24-
hour period being preferred beca use of diurnal variations. Only wind speeds from
critical directions are conside red (paragraph 2.4.3.1). Wind observations during the
24-hour period of maximum rainfall are usua lly the most representative of moisture in-
flow to storms of that or longer duration . For storms of shorter duration, average
winds need be computed for the actual duration only.
2.4.4 Wind maximization ratio
The wind maximization ratio is simply the rat io of the maximum average wind
speed for some specific duration and critical direction obtained from a long record of
observations, say 50 years, to the observed maximum average wind speed for the same
duration and direction in the storm being maximized. The monthly maximum average
values obtained from the records are usually plotted against date of observation, and
a smooth seasonal curve drawn so that storms for any time of the year may be maximized
readily (Figu re 2.12, part C) . The ma xi mum wind speeds used for maximization are
read from the seasonal curve.
Wind records appreciably shorter than about 50 years are unlikely to yield
maximum speeds reasonably representative of those to be obtained from a long record.
Frequency analysis is advisable for such short records. The computed 50- or 100-
year values, usually the former, are used to construct the seasonal variation curve of
limiting wind speed.
Sometimes the moisture values (precipitable water), both ma ximum and storm-
observed, are multiplied by the corresponding wind speeds to provide a moisture-inflow
index . The advantage in this is that the resulting moisture-i nflow index curve pre-
sents a more readily visualized seasonal variation of PMP (Figure 2.12, part D) than
when moisture and wind-speed curves are examined separately. Also, when the seasonal
variation curves are expressed in terms of percentage of the peak or other value, the
moisture-inflow index curve provides a single percentage value for adjusting PMP values
for any particular time of year.
2.5 Storm transposition
2.5.l Definitions
The outstanding rainstorms in a region surrounding a project basin are a very

important part of the historical evidence on which a PMP estimate for the basin is
based. The transfer of storms from locations where they occurred to other areas where
they could occur is called storm transposition.
Transposition limits refer to the outer boundaries of a region throughout

which a storm m6y be transposed with only relatively minor modifications of its rain-
fall amounts. . The area within the transposition limits has similar, but not identi -
cal, climatic and topographic characteristics throughout. · More restricted trans-
position limits may be defined if a region has a long record of precipitation measure-
ments from a relatively dense network of gauges and has experienced several outstand-
ing storms. Where the record of storms is more limited, either because of a sparse
raingauge network or because of very infrequent occurrence of severe storms during the
period of record, then more liberal, though perhaps less reliable, transposition limits
must be accepted.
A transposition adjustment is a ratio by which the storm rainfall amounts are

multiplied to compensate for differences between conditions at the storm site and those
at the project basin.
2.5.2
The transposition procedure involves the meteorological analysis of the storm

to be transposed, the determination of the limits of transposability, and the applica-
tion of the proper adjustments for making the modifications required by the change in
storm location. The procedure may be divided into four steps, as in the following
paragraphs.
2.5.2.l The storm

The first step in transposing a storm is to identify clearly when a?d where
the heavie st rainfall occurred and the approximate causes in terms of synoptic meteo-
rology. An isohyetal chart, a few key mass rainfall curves, and weath:r maps ser~e
these purposes. The isohyetal chart may be a simple one, since its primary func!i?n
is to identify the storm location. Routinely avail~b~e w:ath:r maps may be s~fficient
to identify the storm causes, especially if the precipitation i~ closely associa!ed
with either a tropical or an extratropical cyclone. In other instances, a detailed
analysis may be necessary to identify causes.
2.5.2.2 ~f_s!o:m_t~p~
-Region
- - -of- -influence
--
The second step is to delineate the region in which the meteorologi~a~ st?rm
type identified in step l is both common and important as a producer of precipita!ion.
This is done by surveying a long series of daily weather charts. Tracks of tropical
and extratropical cyclones are generally available in published f orm, and these may be
used to delineate the regions frequented by the various storm types .
The third step is to delineate topographic limitati on s on transposability.

Coasta l storms are transposed alo ng the coast, but only a limited distance inland.
Transposition of inland storm s is restricted to areas where maj or mountain barriers do
not block th e inflow of moisture f rom the sea unless such bl ocking prevailed at the
original s torm site. Adjustments for tra nsposit ion behind moderate and small barriers
are discussed in section 2 .6. 3. Some limitation is placed on l at itudi na l transposi-
tion in order not to involve excessive differences in air mass characteristics. Figure
2.5 shows the transposition l imits for a summer storm in Kansa s , U.S .A. In estimating
PMP over a specific basin, it is only necessary to determine if a particalar storm can be
transposed to the problem basin, and delineation of the entire area of tra nspos abi lity
is not required. It is required, however, in the preparation of generalized esti-
ma t es , which are discussed in Chapter 5.
Figure 2. 5 - Transposition limits (heavy dashed

line) of 9-13 July, 1951 storm . Locations of
synoptically similar summer storms marked X.
Light lines indicate maximum persisting 12-hour
l 000 mb dew points (°C) for July .
The final step in transposition is the application of ad~ustments discussed

in the following section.
2.6 Transposition adjustments
2.6.l
Simply stated, the moisture adjustment is merely the multiplication of the

observed storm rainfall amounts by the ratio of the precipitable water (W ) for the
2
enveloping, or maximum, dew point at the transposed locat ion to that, w, for the
1
representative storm dew point, or
(2.2)
where R is the observed storm rainfall for a particular duration and size of area, and
R is t~e storm rainfall adjusted for transposition. Equation 2.2 incorporates both a
2
transposi tion adjustment and a moisture maximization . The storm depth-area-duration
array of rainfall values, such as in Table 2.1, is multiplied by this ratio. There is,
of course, no need to adjust values for areas exceeding the basin size. The moisture
adjustment may be either graeter or less than unity, depending on whether the trans-
position is toward or away from the moistu r e source and whether the elevation of the
transposed location is lower or highe r than that of the original storm site.
For reas ons given in section 2.2. 4, dew point~ between the rain area an d mois-
ture source tend to be more representative of the atmospheric moisture content, or pre-
cipitable water, flowing into the storm than dew points within the rain area. Such
representative dew points may be a few hundred kilometres away from the storm centre.
In ma~imizing for moistu r e , the maximum dew point used is for the same location as that
of th e -.presentative sto r m dew point. In transposing, the same reference distance is
laid out on the same bearing from the transposition point, as show n in Figure 2.6. The
r eferenced dew-point location is then used for obtaining the maximum dew point from the
maximum dew-point chart for calculating the maximization and transposition adjustments.
2.6.2
An increase in surface elevation decreases the moisture that may be contained

in a column of the atmosphere. However, many storms receive most of their moisture
in a strong low- level flow 1 to 1.5 km deep, and this inflow is not necessarily affect-
ed appreciably by relatively small changes in ground elevation. Ranges of low hills
or gradually rising terrain may actually stimulate convection and inc rease rainfall.
This effect on precipitation may more than compensate for the decrease in precipitable
water with increasing ground elevation . Elevation adjustments for PMP estimates for
non-orographic regions in the middle latitudes are discussed in the next two paragraphs.
ESTIMATES FOR HON-OROGRAPHIC REGIONS 21
Project basin Transposed

_...,. storm __...... 22 •
-- --
\ /
I /
I ,.,,.,.
I _...
I_.-
'-----'~"'""'"'-)
- \...._ - -
- -- ---- ---
. _...!.1feference dew point bearing 170°
23
-- -- --- --
and distance 200 km
-- - -- --
. 24
-- --- -- - - --
-- --
N
25
Storm centre
. - --- - - - ---
--
Observation site of represent-
26 - ative storm dew point. Reference
bearing 170° and distance 200km
Shoreline
Ocean or other
moisture source
Figure 2.6 - Example of storm transposition.

Long dashed lines indicate maximum persisting
12-hour l 000 mb dew points (°C) for the same
time of year the storm occurred or within 15 days
according to common practice (paragraph 2.3.1)
2.6.2.l General storms
Because of unce r tainty as to the effects of relatively small or gradual eleva-

tion changes on pr ecipitation, there are differences of opinion as to whether or not
elevation adjustments should be made for storm transposition over broad, gradually
sloping plains. Decision as to the use or non- use of an elevation adjustment is based
on comparisons of major storms in the vicinity of the actual site of the storm to be
t ransposed with those in the area surrounding the project site. For example, if ob-
served major storms for the two sites showed differences in ma gnitude ascribable only
to differences in moisture, not in vo l ving elevation differences, omission of an eleva-
tion adjustment would be justified. If it is decided to omit adjustment for eleva-
tion, W2 of equation (2.2) is com puted fo r the maximum dew point at the referenced
location (paragraph 2 . 6.1.1) fo r the project site and the same column height as for W •
If an adjustment is used, w 2 is computed for the same maximum dew point jus t describea
but for the column above the ground at the project site, which may be lower or higher
than the site of the observed storm. Regardless of whether or not an elevation ad-
justment is used, transposition involving elevation differences of more than 700 m is
generally avoided.
2.6.2.2 Local th understorms
Intense local thunderstorms are not adjusted for elevation when transposition
involves elevation diffe ren ces of less than about l 500 m. Since this chapter deals
with non-orographic regions, it can be stated, simply, that no elevation adjustment is
made for local thunderstorms. Elevation adjustment for s~ch storms is required in
orographic regions, however, and they are discussed in sections 5.3.3.l and 5.3.6.4.
2.6.3
Transposition of a storm from the windward to the leeward side of a topo-

graphic barrier normally requires an adjustment for the height of the barrier . This
is a common situation, because basins upstream from a proposed dam site are often
rimmed by mountains or hills. Transposition of storms across barriers higher than
about 700 m above the elevation of the observed storm site is generally avoided be-
cause of their dynamic influence on storms. Also, barrier adjustments are not used
in transposing local, short duration, intense thunderstorms, which can draw in moisture
entrapped by the barriers prior to the storm. The example of storm transposition pre-
sented in the next section includes a barrier adjustment.
2.6.4 Example of storm transposition and maximization

-----------------------------------------------
Assume that synoptic weather charts associated with major storms indicate that
the hypothetical storm pattern shown in Figure 2.6 is transposable to the project basin
shown in the same illustration. The average elevation of the storm area is 300 m, and
that of the moisture-inflow, or south, side of the basin is 700 m, with no intervening
orographic barriers . The represe ntative persisting 12-hour storm dew point (section
2.2.4) is 23°C, which was observed at a site (Figure 2.6) located at an elevation of
200 m and 200 km from the storm centre on a bearing of 170° (paragraph 2.6.1.1) . Re-
duction of this dew point to the l 000 mb level (Figure 2.1) yields 24°C.
The adjustment factor, or ratio, is computed as follows:
(2.3)
where the subscripts within parenthes es refer to the l 000 mb dew points for which the
precipitable wate r W is computed, and the subscript s outside parentheses refer to the
various pertinent ground elevations forming the bases of the atmospheric columns for
which W is computed . Thus, the term (w /w ) represents moisture maximization at
26 24 300
the storm site ; (W ~/w )
2 26 300 is the adjustment for the difference in maximum dew points
of the original and transposition locations ; and (W ) c/(W 23 ) is the elevation ad-
23 70 300
justment. Multiplication of all th ese terms leads Io a simple result that all the r e-
quired adjustments are implici t in the sin gle term (W23) 700/(W24)300 • Referr ing to
Tables A.l.l and A.1 . 2 for a co lumn top of 300 mb, (W23) 700 = 67 - 13 = 54,and (W24)300
= 74 - 6 = 68mm . Hence, r = 54/68 = 0 . 79 .
Table 2.1 - Maximum average depth (mm) of rainfall

in storm of 20- 23 May 1927
Area Duration (hours)

(km2)
6 12 18 24 36 48 60 72
25* 163 208 284 307 318 328 343 356
100 152 196 263 282 306 324 340 353
200 147 190 251 269 300 321 338 352
500 139 180 234 250 290 315 336 351
l 000 133 171 220 235 278 304 328 341
2 000 124 160 202 215 259 284 308 322
5 000 107 140 172 184 218 241 258 274
10 000 91 118 140 151 182 201 215 228
20 000 66 87 104 114 143 158 173 181
*Assigned area for maximum s tation precipitation.

If an extensive orographic barrier (section 2.6.3) of, say, l 000 metres in

mean elevation lay between the observed storm site and the project basin, (W 2 3) l 000
would be substituted for (W23)700' and ration r would then be (67 - 18)/(74 - 6), or
0.72. The appropriate rations then applied to the storm depth-area-duration data like
those of Table 2.1. Other storms are adjusted similarly by appropriate ratios, and
the results are then treated as described in sections 2.8 and 2.9.
2.7 Sequential and spatial maximization
2.7.l Definition
Sequential and spatial maximization involves the development of hypothetical

flood-producing storms by combining individual storms or rainfall bursts in individual
or separate storms. The combination is effected by hypothesizing critical sequences
with minimum time intervals between individual events (sequential maximization), which
also may be repositioned, or transposed, geographically (spatial maximization).
2.7.2
Sequential maximization is the rearrangement of observed storms or portions

thereof into a hypothetical sequence such that the time interval between storms is at
a minimum. The storms may have occurred in close succession, or they may have occurr-
ed years apart. The procedure is most often used for large basins, where outstanding
floods result from a sequence of storms rather than from a single event. For small
basins, where rainfall for one day or less may produce the maximum flood, sequential
maximization may involve the elimination or reduction of the time interval between
successive bursts in the same storm or in separate storms.
The initial step for sequential maximization is the same for large or small
basins. In each instance, a thorough study of the meteorology of major storms in the
area of interest is required /"l, 8, 9_}. Storm types associated with heavy rainfalls
in or near the project basin are determined. Movements of surface and upper-air lows
and highs are examined; depth, breadth, and direction of .moisture inflow are deter-
mined; vorticity advection is investigated; etc. It is usually impossible to study
all major storms with the same degree of detail. In the case of older storms, for
example, upper-flow patterns must be estimated from surface data.
The next step is to determine the sequences of storms in and near the project
basin. For large basins, storm sequences should be examined to determine the shortest
reasonable time interval between individual storms of various types. The minimum time
interval, usually measured in days, should be determined for each combination of storm
types producing heavy precipitation. This interval is a critical factor in the hypo-
thetical storm sequence established. For small basins, the procedure, though similar,
concentrates on the interval, usually measured in hours, between bursts in individual
storms. In some instances, the combination of bursts from separate storms is a possi-
bility, and the time interval between similar storms should be considered.
After storms have been examined and reasonable minimum time intervals between
them determined, pairs or sequences of storms or bursts are developed. Each pair of
storms, or for small basins individual bursts within a storm, is examined carefully to
insure that meteorological developments following the first storm or burst, i.e., move-
ment of lows and highs, over-running of the basin by cold air, etc., would not prevent
the succeeding storm or burst from occurring within critical time limits.
If all the important features of the weather situation at the beginning of the
second storm can be developed in a logi~al manner over a sufficiently large area, the
necessary conditions for its onset will have been met. The successive hypothetical
synoptic weather maps for the interval between storms or bursts are patterned to the
greatest extent possible after the actual maps following the fir st storm or bu r st and
preceding the second. Synoptic features, such as highs, lows, and fronts, are allow-
ed to move and change, as indicated by experience, at a somewhat faster than average,
but not excessive, rate. The resulting hypothetical storm sequence is intended to
depict a critical, meteorologically possible transition from one storm or burst to an-
other.
While the derived hypothetical storm sequence often consists of two unadjust-
ed observed storms, the proba ble maximum storm (PMS) is sometimes selected as the
second storm of the sequence. In other words, the second storm has been maximized
for moisture and perhaps wind so that it equals PMP for at least one du ratio n and size
of area (sections 1. 1.4, 2.11.2 and 2.11.3). Sequences of two probable maximum storms
are never developed, however, for two reasons. One is that a properly derived PMS has
a very low probability of occurrence, and the probability of two such storms occurring
in unusually close succession is extremely remote . The second reason is that the
first PMS would be followed by a meteorological situation unfavourable for the rapid
development of the second, and the longer transition period between the two would very
likely make the sequence less critical hydrologically than a sequence of lesser storms
with a shorter time interval between them.
2.7.3
Spatial maximization involves the transposition of storms that occurred in or

near a project basin to one or more critical locations in the basin so as to obtain
maximum runoff. The procedure consists of determining if particular storms can be
transposed to critical locations within specified time intervals and combined t o pro-
duce maximum runoff rates or volumes. Again, as in sequential maximization, the re-
quirement is a thorough knowledge of the storms causing heavy precipitation.
The following example of spatial maximization is based on a series of heavy,

localized rainfall bursts in eastern Colorado, U. S.A ., during the period 14-18 June,
1965 . During this period, a persistent large-scale circulation maintained a pronounc-
ed inflow of moist unstable air into the storm area. Fronts and related synoptic
features played a minimal role as did high-level factor~, such as vorticity advection
8J.
Two distinct, severe six-hour bursts occurred on successive days, 16 and 17
June. Isohyetal maps for the two bursts are shown in Figure 2 .7. The burst on the
sixteenth was centred over Plum Creek Basin (1 100 km2) while that on the seventeenth
was centred about 40 km SE . It is reasonable to assume that the rainfall centres
could have occurred over the same location since the weather situation was very much
the same on both days. Combination of the two isohyetal patterns on the basis of
this assumption resulted in the pattern of Figure 2. 8. In combining the patterns,
i
Denver
•
,,,...- .....
, / "
- 5"'
,
I,,,.-, ,.10 - . . .
I I I' ) I
,,. ,-
J
"'" / "-10 I
/
(
\
\
I
"
I
'--
/
,,,"" "10
.... ,,,... ... "
/
/ _, __ .,,..--...... /
-5--
/ ,,.. - \ I
/ 10"'1'15- .... -, """I I
I / I /' I I//
I I ,..1 ,..,, - .I ,/1 I
I t { l '-._ 11 I
I I I \ )/It
I I lO"f"/1
I I I 20 - .... l
I I ,, ' ' I "" ,,.. - _,I I
I
I
I 11
I ,, I'
I I I ,,, - - "
(I
I I ,, I I I I
I I ll ....;:::::.:::J I
Colorado Springe• \.,, I \
1
I I 15'? "' I
I I ,1,,.-
1/
I "' ,,,"" I
0 10 20
1 /
1 '
I I I I I I
I I km
I l~ /
/I
l .._ - /
I
Figure 2.7 - Isohyets (cm) for the 6-hour

afternoon storms of 16 June 1965 (solid lines)
and 17 June 1965 (dashed lines) in eastern
Colorado, U.S.A.
t
/15,
I \
I I
I I
I I
I /
/ I
.-/ I
/,. I
/ 20
, t
+
0 10 20 30
•Colorado Springs
I I I I
km
Figure 2.8 - Isohyets (cm) resulting from

combination of the patterns of the 6-hour
storms of 16 and 17 June 1965 shown in
Figure 2.7
the principal centre of that on the seventeenth was super imposed on that of the six-
teenth, and the pattern was rotated about 25 degrees counter- clockwise for better
agreement with the orientation of the pattern on the sixteenth . In this region, -
such a rotation is realistic for this type of storm. In oth e r regions and for other
storm types, examination of many storms might show that such r otation would not _be
permissible.
2. 7.4
Sequential and spatial maximizations are generally used in combination, i.e . ,

storms or bursts wi t hin storms may be repositioned geographically in addition t o short-
ening the time int e r val between them. In the study /"3_/ from which the example of
section 2 . 7 . 3 was t a ken, the two rainfall bursts we r e not only maximized spati ally by
supe r imposing centres and rotating one of the isohyetal patterns, but also the time
inter val between them was shortened.
The actual times of the bursts depicted in Figure 2. 7 were l p.m. to 7 p.m.,
16 June, and 2 p.m. to 8 p.m . , 17 June. Examination of a large number of similar
stor ms occurr ing in rela ti vely close succession indi cated that the interval between the
two bursts could be reduced to 12 hours. This shortening of the t ime interval result-
ed in assigning an overall duration of 24 hours to the total rainfall for the two
bursts, or seven hours less than the observed total storm period of 31 hours.
Examples of the use of sequential and spatial maximization in deriving hypo-

thetical maximum flood-producing storm sequences for large basins may be found in ref-
erences 2 and 4.
2.8 Envelopment
2.8.l Introduction
To maximize a single storm and transpose it to a basin is a demonstration that

a certain precipitation volume could fall over that basin. Nothing about the relation
of this pr ecipitation volume to PMP is revealed, and it could be far l es s than PMP
magnitude . To consider only two or three storms or storm sequences, no matter how so-
phisticated the maximization and transposition adjustments might be, gives no assurance
that the PMP level has been obtained.
The question of adequacy of storm sample for estimating PMP is a difficult

one, es pecially with limited data. It seems logical, however, to expect that an enve-
lope of rainfall values maximized and transposed to a basin is very likely to yield
values indicative of PMP magnitude. This is especially true since no single storm is
likely to yield extreme rainfall values for all durations and sizes of area. It is
for these reasons that envelopment is considered a necessary final step in estimating
PMP.
2.8.2 Envelopment
-----------
Envelopment is a process for selecting th e largest value from any set of data.
In estimating PMP the maximized and transposed rainfall data are plotted on graph paper,
and a smooth curve is drawn through the largest values. Figure 2.9 shows an envelope
of transposed, maximized precipitation values for durations up to 72 hours over a 2 000
km2 area. The variables are cha nged in Figure 2.10, which is an envelope of trans-
posed, maximized 24-hour rainfall values for areas ranging up to 100 000 km2. In de-
veloping a full array of PMP depth-area-duration data for a basin, it is necessary to
envelo pe both ways, as in Figures 2.9 and 2.10. Values re~d from the envloping curves,
such as shown in these two figures, are then used to construct a set of depth-area-dura-
tion curves, as shown in Figure 2.11.
600
500
x
~
400
~
,-... •
~
~
n
n
~
300
~
~
~
~
~
• St orm A
~ Storm B
200
w Storm c
(e ) St orm D
,00 • Storm E
0 12 24 36 46 60 72
Duration (hours)
Figure 2.9 _ Depth-duration envelope of trans-

posed maximized storm values for 2 000 km2
50000
••
10 000
5 000
N
~
.......,1 000
.. ~ .
!l
ti
500 •
•
Storm A
Storm B
' ..
(•)
)( Storm C
100
( •)
~
Storm D
Storm E
. ..
(•)
50
• (•) .•
10
0 200 400 600 800
24-hour rainfall (mm)
Figure 2.10 - Depth-area envelope of trans-

posed maximized 24-hour storm rainfall values
L.B 72
50 000
10 000
5 000
..-...
(\J
~ 1 000
Cll
Q)
F-1
«I
• Storm A
y Storm B
)( Storm c
100
(•} Storm D
Al Storm E
50
10L---'-~--'-__J'----'-~-'-------'-~--'-~-'-------'-~_._~..___._~~~~~~~~~~
0 10 200 300 L.00 500 600 700 800 900
Probable maximum precipitation (mm)
Figure 2.11 - Enveloping-depth-area-duration

curves of probable maximum precipitation for
a hypothetical basin
It should be noted that the controlling points determining each curve are usu-
ally from different storms. On Figure 2.11, for example, with the exception of the 6-
and 12-hour curves, the points controlling the curves at about 2 500 km2 are typically
from different storms than those at 100 000 km2. Similarly, the points controlling
the short-duration curves are usually from different storms than those controlling the
long-duration curves.
2.8.3
The data used in constructing an envelope curve are not of equal accuracy of
reliability. For example, with r eferenc e to charts like those of Figures 2.9 and
2.10, the basin under study may lie definitely within the transposition limits of some
of the transposed storms, but it may lie just within the fringes of the transposition
limits of other storms, which leads to an element of doubt as to their transposability
to this particular basin. Under these circumstanc~s, it may be justified to place the
curve at somewhat lower values than the extremes in the dubious category. This is
called undercutting.
2.9 Summary outline of procedure for estimating PMP
2.9.l Introduction
The steps outlined below for estjmating PMP over a project basin are applic-
able only for a non-orographic region with hyd rometeorological data. For most reli-
able estimates, data should include: (1) relatively detailed 6-hourly or daily weather
maps; (2) long rec ords, say, 50 years or more, of hourly and daily rainfal l data from
precipitation networks of sufficient density to permit reliable determination of time
and spatial distribution of storm rainfall; (3) long records of temperature, dew-point
and wind data both at the surface and, if possible, aloft, although upper-air data are
not absolutely required for the procedure outlined here. It should be kept in mind
that the procedure described generally applies only to middle-latitude basins of no
more than about 50 000 km2. Also, since it is very unlikely that a project basin will
have experienced all the outstanding storms of the region in which it lies, storm
transposition is almost always required.
2.9.2
Step 1. Using weather, topographic, and preliminary total-storm isohyetal

maps, determine the transposition limits of storms, as described in section 2.5.
Step 2. Survey precipitation records to obtain outstanding storms of record

within the region of transposability.
Step 3. Make depth-area-duration (DAD) analyses of the storms selected in

step 2, as described in "Manual for depth-area-duration analysis of storm precipita-
tion", WHO-No. 237. TP. 129. The results of the analysis for each storm are tabulated
as shown in Table 2.1. (The DAD analysis of storm precipitation is a lengthy and
ESTIMATES FOR NON- OROGRAPHIC REGIONS 33
tedious process even when done by computer. A ready file of storm DAD data is a real
convenience in making PMP estimates, and some countries maintain a continuing program
of DAD analysis for accumulating a file of such data both for old storms of record and
for new storms as they occur . DAD data for storms in the area of transposability may
be selected readily from such files, thus eliminating steps 2 and 3 . )
Step 4. Determine the representative persisting 12-hour dew point for each
appropriate storm, as described in section 2.2.4. Since this dew point is usually
outside the rain area (Figure 2.2), its distance and direction, or bearing, from the
storm centre should be specified (paragraph 2.6.l.l). If wind maximization is indi-
cated (section 2.4), select also for each storm the maximum 24-houraverage speed of the
wind from the moisture-inflow direction. Multiply the precipitable water (W), corre-
sponding to the representative storm dew point, by the wind speed to obtain the repre-
sentative storm moisture-inflow index (Figure 2.12).
Step 5. Determine t he highest maximum persisting 12-hour dew point of re-

cord for the location of the reference dew point for the transposition site, as des-
cribed in sections 2.2.5 and 2.6.l.l. Since several storms of different dates and
with different reference dew-point locations must be transposed, it is recommended that
the maximum dew points for the entire storm season and for the project basin and su r -
rounding areas be determined at one time, as described in section 2.2.5. Preparation
of maximum persisting 12- hour l 000 mb dew-point maps, such as shown in Figure 2.4, is
advisable. Such maps have an additional advantage in that they yield some indication
of the geographic variation of PMP values in a plains area.
If wind maximization is required, survey storm wind data for highest maximum
24-hour average speed from direction of moisture source. Multiply the precipitable
water (W) corresponding to the maximum persisting 12-hour l 000 mb recorded dew point
for the storm date, or within 15 days, by the maximum 24-hour average recorded wind
speed for the same date to obtain a maximum moisture-inflow index, as in Figure 2.12.
Here, again, it is rec ommended that the maximum moisture-inflow index be determined
for the entire storm season at one time.
Step 6. Compute the combined transposition and maximization ratio of the

precipitable water (W) for the maximum persisting 12-hour l 000 mb dew point of step 5
for the storm date, or within 15 days of it (paragraph 2.3.1), to that fo r the repre-
sentative persisting 12- hour l 000 mb dew point for the storm, as described in section
2.6. If wind maximization is involved, compute the ratio of the maximum moisture-
inflow index to the representative storm moisture-inflow index.
Step 7. Multiply the DAD array, such as in Table 2.1, for each storm by the
appropriate precipitable-water or moisture-inflow index ratio, as determined in step 6.
Step 8. Plot the transposed, maximized DAD values of step 7 on diagrams,

such as shown in Figures 2.9 and 2.10, and draw envelope curves. Use envelope curve
values to construct DAD curves of PMP, as shown in Figure 2.11. Although not manda-
tory, storms providing control points on the PMP curves should be identified, as indi-
cated in Figure 2. 11, for convenience in selecting actual storm patterns for dete r min-
ing the time and spatial distribution of the PMP in '. calculating the design flood .
Maximum persisting 12-hour 1 000 mb dew point at Mosul
Precipita ble water in layer between 1 070

• barri{lr and 200 mb level (based on above
urveJ
Maximum 24-hour average wind speed at

Mosul (SE and SW quadrants)
Moisture inflow precipitation water x wind speed
Nov Dec Jan Feb Mar Apr May
Figure 2.12 ~ Seasonal variation of probable

maximum precipitation in the upper Tigris
river basin in Iraq
2.10 Seasonal variation of PMP
2.10.l Introduction
In those regions where the maximum flood is likely to result from a combina-
tion of snowmelt and rainfall, it is necessary to determine the seasonal variation of
PMP so that various combinations for different times of the melting s eason can be eva l u-
ated in order to obtain the most critical. For example, in a particular region, maxi -
mized June storms may provide the controlling points for PMP but optimum combinations
of accumulated snow on ground and melting rates may be found in April. It is then
necessary to estimate PMP for April. Since it is not known exactly what time of year
is most critical for the maximum snowmelt and rain - flood, the usual procedure is to
determine the seasonal variation curve of PHP for the entire snowmelt season. The curve
then permits a ready adjustment of PMP for use in assessing flood situations at various
times during the melting season in order to determine the most critical flood .
There are various ways of determining the seasonal variation of PMP . The
more common procedures are disc ussed here. Selection of a procedure depends on data
available. Whenever possible, it is advisable to use several procedures in develop-
ing a seasonal variation curve. Cautionary remarks on the representativity and use
of seasonal variation curves are given in section 2.13.4 .
2.10.2 Observed storms
The best way for determining the seasonal variation of PMP requires a relo-
ti vely large number of storms for which DAD data are available and which are fairly
well distributed throughout the melting season. Different variations are usually
found for small and large areas and for short and long durations. It is, therefore,
important to base the seasonal variation on data consistent with the basin size and
critical rainfall duration. Because of this, it is often advisable to construct a
set of curves rather than a single one. The storm rainfall for a particular size of
area and duration is then maximized for moisture, as described in sections 2.3 and
2.6. The maximized data are then plotted against date of storm occurrence, and a
smooth envelope curve is then drawn. The rainfall scale is usually converted to a
percentage scale expressing the PMP as a percentage of the peak value or the value for
some particular time of year.
2.10.3
The seasonal variation of maximum persisting 12-hour dew points may be used
also to determine the seasonal variation of PMP. This procedure is more applicable
to localized thunderstorm PMP than to PMP for large areas and long durations. Pre-
cipitable water is computed for the individual maximum 12-hour dew points throughout
the critical season, or it may be computed for values read from their seasonal varia-
tion curve, like that of Figure 2.3. A shortcoming of this procedure is that it will
almost always indicate a peak PMP value in summer, even in regions where summers ore
dry and major storms occur in winter. It cannot be used under these conditions un-
less wind is considered also (see next paragraph).
2.10.4 Moisture inflow
In those regions where summers are dry and major storms occur only in the cold
half of the year, the seasonal variation of maximum precipitable water (paragraph
2.10.3) gives a false indication of the seasonal variation of PMP when used by itself.
Awind factor is then required to develop a representative seasonal variation of PMP.
Figure 2.12 shows a seasonal variation curve developed for PMP in the upper
Tigris Rive r Basin, where in summer there is very little rain. While the maximum dew
point and precipitable water curves tend to show minimum values during the cold season
climatological records show t hat in this region all major general-type storms occu r in
that season. Weather charts indicate that the heaviest precipitation occurs with sur-
face winds in the south-east and south-west quadrants. A survey of a long record of
surfac e winds yielded the maximum 24-hour wind curve of part C of the figure, which
sh ows peak values in January and February. Multiplication of precipitable water
values by wind speed resulted in the so-called moisture-inflow index curve of part D.
The double peak was confirmed by outstanding recorded storms.
2.10.5
An indication of the seasonal variation of PMP may be outlined readily from

mon t hl y maximum daily station rainfall amounts. The use of average maximum values for
several stations rather than from a single station is advisable fo r the larger basin
sizes. In the usual periods of rapid weather transitions, such as early fall and late
spring, it may be advisable to select maximum rainfall values by half-month or 10-da y
periods. Here, again, the maximum values are plotted against date of occurrence, and
a smooth seasonal envelope curve is then drawn. The rainfall scale is usually con-
verted into terms of percentage, as in section 2.10.2.
2.10.6
Occasionally, special summaries of precipitation data may be found which can

be used to derive the seasonal variation of PMP. One such summary is of average week-
ly precipitation for given areas, as determined by averaging station precipitation
within each area for each week of the year over a long period. The seasonal variation
curve of PMP may be based on an envelope of these weekly values. Obviously, a season-
al variation curve thus developed would be more applicable to PMP for long durations
and large areas.
2.11 Areal distribution of PMP
2.11.l Introduction
Once the PMP values for a particular location have been derived and presented
to the. hydrologist in the form of a table or enveloping DAD curves, as in Figure 2.11,
he still has the problem of determining areal distribution over the project basin. It
is not generally recommended that the PMP values be considered as applying to one storm,
especially for the larger basins. Direct use of the PMP values may be unrealistic for
the most critical design storm for two main reasons. First, the storm producing ma xi-
mum ra infall over small areas within a project basin is usually of a differen t type
from that producing maximum rainfall over the same basin as a whole. Similarly, dif-
ferent types of storms may obtain for different durations over the same basin. Second,
the shape and orientation of the basin may be different from those permissible for the
controlling isohyetal patterns.
2.11. 2 Observed storm pattern

-----------------------
F~r the above reasons, the hydrometeorologist makes recommendations regarding
the storm isohyetal patterns that may be applied to a basin. One or more transposed
storms may pr~vide a suitab~e pattern or patterns . Such a choice applies especially
when both ba~in and storm site are topographically similar. A limitation may be placed
on the rotation or displacement of the isohyetal pattern. If, as often happ~ns, the
transposed or basin storms select ed provide points on the PMP DAD curves, no further
adjustment may be required . If not, they may be maximi zed as in Figure 2.13. Current
practice, however, favours bringing average depths for all durations of the storm to
PMP levels, as described in section 2 ~ 11.3 for an idealized pattern. In applying the
procedufe to actual storms, care must be exercised to ensure that rainfall depths for
are~s smaller than the basin do not exceed PMP. If they do, the storm depth-area re-
lations must be altered so that depths nowhere exceed PMP.
2.11. 3
An alternative method for fixing the areal distribution of PMP over a basin is
based on the assumption that the PMP values for all durations at the total area of the
basin could occur in a single storm. This usually introduces an additional degree of
maximization, because controlling values for all durations at a particular size of area
are generally f rom several storms . In order to counter this, the precipitation values
for the smaller areas within the basin are maintained at less than PMP, usually being
patterned after the depth-area relations of major storms that have occurred over or
near the project basin . For example, the dashed "within-basin" curv~s (only two
shown) of Figure 2.14 set the concentration of rain within a 3 000 km basin for the
6- and 24- hour durations. These curves are generally drawn for all durations by 6-
hour intervals.
2.11.3.l Areal distribution
The areal distribution of basin PMP involves the shape and orientation of its
isohyetal pattern, and this may be based on observed storms. For basins up to about
20 000 km2 in flat terrain, an oval-shaped pattern with almost any orientation is
adaptable and the pattern is usually centred over the basin. For larger basins up to,
and even above, the limiting si ze considered in this report , in the middle latitudes of
the northern hemisphere, the orientation of the pattern tends to be in a general south-
west - north-east direction over flat terrain. The pattern may or may not be centred
over the basin, depending on what the history of major basin storms indicates.
38 . ESTIMATION OF PROBABLE MAXIMUM PRECIPITATION
100 200 500

10
100 (17 5)2 00 (350) 500 ( 875)1000
Probable maximum precipitation (mm)
Figure 2.13 - Maximization by sliding technique. Storms

not providing control points on PMP DAD curves may be
maximized by plotting to same scale on separate sheets of
logarithmic paper the storm and PMP DAD curves. The sheet
with the storm curves is then superimposed on the oth er and
is slid to the right until the first apparent contact be-
tween curves for the same duration is effected. The ratio
of any PMP scale value to the superimposed storm scale value
is the maximizing factor. Obviously, this factor adjusts
the observed storm for greater rain-producing efficiency as
well as for maximum moisture. Above illustration for a
5 000 km2 basin shows point of first contact occurring be-
tween the 72-hour curves at about 2 000 km 2 , but different
time and spatial distributions might show point of first
contact for another duration and/or size of area
10 000
Duration (hours)
5 000
,,-..
000
N
PMP DAD curves
~
.."4 500
Cll
Q)
f-1
<
100
50
10
0 50 100 150 2 00 250 300 350
Depth (mm)
Figure 2.14 - Example of enveloping depth-area-

duration curves of probable maximum precipitation and
within-basin storm rainfall depths for a 3 000 km2
basin
2. ll. 3. 2 Example
-- - -
The critical storm pattern is usually constructed on the assumption that the
largest volume of rain over the basin will produce the most critical design flood.
Hypothetical isohyets are drawn more or less congruent to the basin boundaries (Figure
2.15), and the rain values, or labels, for the isohyets are determined by a procedure
that is essentially a reversal of the usual DAD analysis. For example: given the
6-hour PMP and "within-basin" DAD curves of Figure 2.14, determine the isohyetal values
for the critical storm pattern superimposed on the outline of the 3 000 km2 basin of
Figure 2.15. Table 2.2 shows how the isohyetal profile is computed, and the results
are shown in Figure 2.16. The required isohyetal values are obtained as shown in
Table 2.3.
40 ESTIMATION OF PROBABLE MAXIMUM PREC IPITATION
... Basin outline
Figure 2.15 - Critical isohyetal pattern

over 3 000 km2 basin
0'--~~~5~~-1~0~~---:':~:--~--,l20=--~~25L,,-~__j30
radius [km]
Figure 2.16 - Isohyetal profile constructed from

data in columns 6 and 8 of Table 2.2
Table 2.2 - Isohyetal profile computation ·
(1) (2) (3) (4) (5) (6) (7) (8)

Total Net Average Accumulated Net rain ~Volume Average Equivalent
area area depth rain volume volume 6 Area area circle
radius
km 2 km 2 mm km 2 mm km 2 mm mm km 2 km
10 10 122 1 220 l 220 122 10 1.8

40 30 113 4 520 3 300 110 25 2. 8
60 20 110 6 600 2 080 104 50 4.0
80 20 107 8 560 l 960 98 70 4.7
100 20 105 10 500 l 940 97 90 5.3
200 100 100 20 000 9 500 95 150 6.9
400 200 92 36 800 16 800 84 300 9.8
600 200 88 52 800 16 000 80 500 12 .6
800 200 84 67 200 14 400 72 700 15 . 0
1 000 200 81 81 000 13 800 68 900 16.9
2 000 1 000 71 142 000 61 000 61 1 500 21.9
3 000 l 000 64 192 000 50 000 50 2 500 28.2
Column 1. Standard size areas.
Column 2. Successive subtraction of column items.
Column 3. Maximum average depths from 6-hour "within- basin" curve of Figure 2.14.
Column 4. Product of columns l and 3.
Column 5. Successive subtraction of column 4 items.
Column 6. Column 5 divided by column 2.
Column 7. Average of two consecutive areas in column 1.
Column 8. Radius of circle with area of column 2.
Data of columns 6 and 8 are then used to construct the curve of Figure 2.16.
42 ESTIMATION OF PROBABLE MAX I MUM PRECIPITATION
Table 2. 3 - Evaluation of isohyet labels of Figure 2.15
(1) (2) (3) (4)

Isohyet Enclosed area Equivalent radius Isohyet value
2
km km mm
A 10 1.78 122
B 200 7.98 89
c 500 12.65 77
D 750 15.50 70
E 2 000 25.20 55
F 3 000 30.98 48
Column l. Refers to isohyets of Figure 2.15 .
Column 2. Areas enclosed by isohyets of Figu r e 2. 15.
Column 3. Radii of circles equivalent in area to values in column 2.
Column 4 . Labels for isohyets of Figure 2.15 as indicated by entering Figure 2.16
with radii of column 3.

2.12 Time distribution of PMP
2.12.l Order of presentation
PMP values, whether presented in tabular form or by DAD curves, are generally
given with the maximum accumulated amounts for any duration preceding all other values
fo r the specified duration. In other words, the 6-hour PMP amount given is the maxi-
mum 6-hour increment to be found anywhere in the PMP sequence, Similarly, the amounts
for 12, 18, 24 hours and longe r are the maximum for the sequence. This order of pre-
sentation, however, is rarely representative of the chronological order found in actual
storms. Furthermore, it often is unlikely to produce maximum runoff for the amounts
of rainfall involved.
2.12.2
A more realistic, and genera lly more critical, chronological order is usual ly
obtained f rom some storm producing critical runoff amounts and rates in or nea r t he
project basin. Table 2.4 presents an example of how the order of the 6- hour PMP in-
crements might be rearranged to agree with the chronological order of a critical ob-
served storm. Note that this procedure leads to mu c h higher rainfall amounts, hence,
higher runoff than would the use of a storm maximized as in paragraph 2.11.2, where
usually only one maximized value equals PMP.
When it is thought that there might be more critical possible arrangements of

rainfall increments than indicated by observed storms, various realistic arrangements
are examined, and the more likely ones are specified. It is the responsibility of the
hydrologist to determine which arrangement will produce maximum runeff.
2.13 Cautionary remarks
2. 13.l Importance of adequate storm sample

-----------------------------------
Transposition and maximization of a few storms are unlikely to yield reliable
PMP estimates, It is important that all outstanding stor ms recorded over the project
basin and areas of transposability be used in making such estimates. If comparison of
storms in the areas of transposability with those outside indicates that only a few
storms within the areas reach the magnitude of the generally greate r storms outside the
areas, the transposition limits should be re-examined and relaxed, if at all possible
to include storms in the marginal areas just outside the limits originally determined:
Storm surveys and analyses should be extended to meteorologically comparable

regions no matter how far removed from the project basin. If synoptic storm types are
kept in mind, far distant areas of the world may sometimes provide better clues to PMP
than nearby areas. This not only applies to precipitation data but to other factors
instrumental in developing concepts basic to understanding of storm precipitation-
producing mechanisms,
Table 2.4 - Chronological distribution of PMP

for a hypothetical 3 000 km2 basin
Duration PMP 6-hour increments Maximum

hr mm accumulation
PMP Arranged*
6 284 284 16 284
12 345 61 28 345
18 384 39 20 384
24 419 35 12 419
30 447 28 39 431
36 467 20 61 451
42 483 16 284 479
48 495 12 35 495
54 505 10 5 500
60 513 8 8 508
66 521 8 10 518
72 526 5 8 526
*Increments in fourth column assumed to be arranged according to sequence of incre-

ments in critical storm producing maximum runoff in project basin. Note that maxi -
mum summation of increments in last column for any given duration may be less than or
equal to, but not more than , the summation of PMP increments for the same duration.
Thus, for example, the maximum 24- hou r amount in the last column is equal to the PMP
value of 419 mm (39+61+284+35 ), but t he maximum 30-hour value is only 431 mm (12+39+
61+284+35), whereas the 30-hou r PMP value is 447 mm . Actually, in this example,
only the 6-, 12-, 18-, 24-, 48- and 72-hour accumulations equal the PMP values.
ESTIMATES FOR NON- OROGRAPHIC REG IONS 45
The greater t he number of caref ul l y se l ected extreme storm s t r ans pose d and
maxi mized, t he greater the r eli a bility of t he r esu ltin g PMP es ti mates . Unde r id eal
conditions, s ome two doz en ma jor storms migh t be cri tic a l for dete r min i ng PMP. Of
these , pr obab ly f ewer th a n half a do zen mig ht pr ovide cont r ol poi nts on t he PMP DAD
curves.
2. 13 . 2
The final results of any PMP esti mate should always be compared with observed
record values . The world r ecor d values of point rainfall, pr esented in Annex 2, very
probably approach PMP magnitude , and estimates appreciably exc ee ding these valu es, say
by 25 pe r cent or more, are li kely to be excessive . Most estima t es of point PMP woul d
be lower th a n these r ecord values for durations of, s ay, four hou r s and longer since
few basins ar e so f avou r ably located as to expe r ience rainfalls of these r ecord magni -
tudes.
Table A. 2. 3 presents enveloping values of DAD data from ove r 700 storms in the
United States . Note th a t all but one value are f r om storms in the southern portion o f
the country near the mo is tu r e sourc ~, which is the Gulf of Mexico. These enveloping
values f r om such a la r ge sample of major storms ve r y probably approach PMP magnitude
for this region, especially for areas larger than about 25 km2. On the other hand,
they exceed PMP magni t ude in those regions farther r emoved f r om the moisture source .
2.13.3
PMP estimates for various basins in a climatically homogeneous region should

be compared for consistency . Appreciable differences should be studied to see if they
are supported by climatic or geographic factors . If not, i t ca n be concluded that the
diffe r ences are not valid and the various steps involved in t he pr ocedu r e for estimat-
ing PMP should be re - examin ed thoroughly . When PMP estimate s are made basin by basin
at various tim e s, consistency is difficult to maintain. For achieving consistency,
the generalized es timates approach, described in Chapter 5, is recommended.
2 . 13.4 Seasonal variation
Any one of the procedu r es described in section 2. 10, except possibly that de-
scribed in paragraph 2. 10 . 2, may result in seasonal curves of PMP that are obviously
misleading . For this reason, it is advisable to try several procedures to see if
there is agreement between the resulting seasonal variation curves . Judgment on
whether a de r ived curve is representative or not should be based on a compa r ison with
actual storms observed at various times during the critica l season .
As mentioned in section 2 . 10, the seasonal va r iation of PMP varies with dura -
tion of storm rainfall and size of area, and several seasonal variation cu r ves may have
to be derived for various durations and areas. Also, a seasonal variation curve does
not imply that maximized storms can be transposed in time without regard to seasonal
limitations on storm types . The curve may be used only to adjust the level of PMP to
various times of the year. Storm types and patterns, however, differ fro m month to
month, and a July storm, for example, is rarely adaptable to Ap r il conditions. Storm
transposition in time is usually li mited to 15 days, but a longer period, sa y, one
month, ma y be justified whe n storm data are sparse.
2.13.5 Areal distribution
Two methods of establishing the areal distribution of what may be termed the
PMP storm we re desc ribed in section 2.11. The first, which involves the use of on ob-
served storm pattern maximized by the "sliding technique" (section 2.11.2), yi elds con-
servative values, since the storm thus maximized usually equals PMP for only one dura-
tion and size of area . The second me thod, which is used with idealized storm patterns,
requires PMP values for the basin area to be equalled for all durations (section
2.11.3). For a large basin, it is unlikely that any one storm would provide PMP
values for all durations, so that, in effect, the assumption that it could is on over-
maximizotion. In order to compensate for this, values for areas smaller than the
total basin area are set at le ss tha n PMP by th e use of "within:...basin" depth-area curve~
shaped according to observed storms. The larger the basin, the larger is the differ-
ence between PMP and "within-basin" curve values for any given area smaller than the
basin (Figure 5. 31) . Conversely, the difference decreases as basin size decreases, so
that for basins of no more than a few hundred square kilometres, the areal distribution
is usually accepted as conforming to the PMP curves.
If meteorological conditions are the some there is no reas on why t he rainfall

potential over, say, a 100 km2 area in a 25 000 km~ basin should be less t han that over
a 100 km2 area in a 5 000 km2 basin. The reason that "w ithin-ba sin" curves indicate
lesser small-area depths as basin size increases is that they ore patterned ofter actu-
al storms and reflect actual distributions. The effect of small- area depths on total
basin rainfall volume decreases as basin size increases.
An important restrictio n on construction of depth-area curves is that their

slopes should nowhere indicate a decrease in ra infall volume with increasing area.
This applies to all depth-area curves, including PMP.
While most examples of PMP estimation presented in this manual involve areal
distribution based on "within-basin" curves, it should not be inferred that this method
is recommended. Whether the areal distribution fs based on an observed storm maxi-
mized by the "sliding technique," on "within-basin" curves, on PMP depth-area curves,
or on other methods depends on the safety factor required in the design of a hydro-
logical structure. The areal distribution to be used is usually selected by the
hydrological engineer. If he wants the most liberal design values, he will select
areal dist ribution based on PMP curves. If not, he will select another method yield-
ing lower de sign values. In making his selection, the engineer receives guidance from
the hydrometeorologist . For example, the storm patterns used for maximizing by the
"sliding technique" or for deriving "within-basin" curves are selected by the hydro-
meteorologist, who may also provide advice on how the patterns may be placed on the
problem basins.
References
1. Lott, G. A. and Myers, V. A., 1956: Meteorology of flood-producing storms in

the Mississippi river valley. Hydrometeorological Report No. 34, U.S. Weather
Bureau.
ESTIMATES FOR NON- OROGRAPH IC REG IONS 47
2. Myers, V. A., 1959 : Meteorology of hypothetical flood sequences in the Missis-

sippi river basin. Hydrometeorological Report No . 35, U.S. Weather Bureau.
3. Riedel, J, T., Schwarz. F. K. and Weaver, R. L., 1969: Pr obable maximum precipi-
tation over the Sout h Plat t e r ive r, Colorado, and Minnesota river, Minnesota ,
Hyd romete orological Repo rt No . 44, U.S. Weather Bureau.
4. Schwarz, F. K., 1961: Me teo rol ogy of flood-producing storms in the Ohio river
basin , Hydrometeorological Report No . 38, U.S. Weather Bureau.
5. Schwarz, F. K., 1967: The r ole of persistence, in stability and moisture in the
intense ra insto r m in eastern Colorado, June 14 - 17, 1965. Technica l Memorandum
WBTM HYDR0- 3, ESSA, U.S. Department of Commerce ,
6. U. S. Weather Bureau, 1947 : Generalized estimates of maxi mum possible precipita-

tion over the United Sta tes east of the 105th meridi an. Hydrometeorological
Report No. 23, pp. 5 - 6.
7. U.S. Weather Bureau, 1960 : Gene ralized estimates of probable maximum precipita-
tion west of the 105th meridian . Technical Paper No . 38 , pp. 22 - 25.
8. Weaver, R. L., 1962 : Meteorology of hydrologically critical storms in California.

Hydrometeorological Report No . 37, U. S. Weather Bureau .
9. Weaver, R. L., 1968: Meteorology of major storms in western Colorado and eastern
Utah , Technical Memorandum WBTM HYDR0-7, ESSA, U.S. Department of Commerce .
10. Wiesner, C. J ., 1970: Hydrometeorology . Chapman and Hall, Ltd., London,

pp.167 - 204.
11. World Meteorological Organization, 1969: Estimation of maximum floods.

WMO-No. 233 . TP. 126. Technical Note No . 98 . pp. 9-17.
CH A P T E R 3
ESTIMATES FOR OROGRAPHIC REGIONS
3.1 Precipitation in mountainous regions
3.1.l 2:~~:~e~~=-~~!!~=~==~
The effects of topography on precipitation have been studied for many years .
Observations of precipitation and runoff in mountainous terrain in many parts of the
wo r ld show a gene r al increase of precipitation with elevation . Several features of
the increase can bi discussed separately.
First there is the increase on windward slopes due to forced lifting of ai r

over mountains. The magnitude of the effect on precipitation varies with the direc -
tion and speed of the moist air flow, and with the extent, height, and regularity of
the mountain ba r rier . Breaks in ridges, or passes, reduce the amount of lifting .
Other facto r s are extent and height of lower mountains or hills upwind of a slope .
Concomitant with increased precipitation on windward slopes is the decrease

on lee a r eas . Immediately to the lee of ridges, however, is a spillover zone, where
pr ecipitation produced by the forced ascent of moist air over windward slopes can be
as great as on the ridge. Because of the relatively slow fall velocity of snowflakes,
spillover extends much farther beyond the ridge for snow than it does for rainfall .
A second feature of orographic precipitation, indicated by theory and support -

ed by obse r vations, is that fi r st slopes or foothill regions are pr eferred loca t ions
for the initiation of showers and thundershowe r s . This e f fect results from stimula -
tion of c onvective activity in unstable air masses by an in i tial and relatively small
lift . Observational data are often too sparse to verify this phenomenon because of
the more obvious effects of higher slopes nearby . Coastal station observations some-
times exhibit the e ffects of small rises in elevation . For example, a comparison of
rainfalls at San Fr ancisco, California, and Farallon Island, app r oximately 40 km off
the coast near San Francisco Bay, showed that, in major sto r ms, r ainfall is about 25
per cent greate r at San Francisco . This effect was taken into account in a PMP study
for the north-western United States [IQ] .
3. 1. 2
Expe r ience has shown that general storm precipitation resulting from atmos -
pheric systems that produce convergence and upward motion is just as important in oro-
graphic regions as on the plains . Reports of thunderstorms and passages of weather
systems during large - area storms on high mountain ranges are an indicator of the dual
nature of precipitation in orographic regions. Radar, for example, has tracked bands
of precipitation moving across the coastal hills and Central Valley of California into
the high Sierra Nevada [f.'[J.
3.1 . 3
Mean annual and sea sonal precipitation for mountainous terrain can be influ-
enced greatly by the varying frequency of re la tivel y light rains. Some weather situa-
tions produce precipitation on mountains when little or no precipitation is observed in
valleys, and storm precipitation generally has longer durations in the mountains . Thus,
the variation indicated by mean annual or seasonal precipitation maps is not necessari-
ly a reliable in dex of geographic variation in PMP unless adjusted for these biases.
An adjustment tec hnique frequent ly used is based on the mean number of rainy days at
stati on s in the proje ct area and a map showing the average station, or point , precipi-
tation per rain y day (which is usually defined as any day with measurable precipita-
tion, but a higher th res hold value, say 2 mm 1 is sometimes use~. The most re presenta-
tive mean annual and seasonal precipitation maps are those based on other data in addi-
tion to precipitation [2., §] and such maps should be used whenever possible.
3.1.4
Because of the dual nature of precipitation in mountainous regions, the s1m1-

larity between storm precipitation patterns and topography is limited, varying with
the precipitation-producing factors involved. Nevertheless, in mountaino us terrain,
orographic influences on precipitati on usually predominate, especially in major storms ,
For this reason, caution should be exercised in transposing storms in such regions be-
cause their precipitation patterns are usually intimately linked t o the orography where
they were observed.
3.1.5
PMP estimates for orographic regions must be based on two precipitation com-
ponents : (1) orog raph ic precipitation, which results from orographic influences, and
(2) convergence precipitation, which results from atmospheric processes presumably in-
dependent of orographic influences . Both components must be evaluated in making PMP
estimates.
The orographic separation method consists of estimating each precipitation

component separatelj and then adding them, keeping in mind some necessary restrictions
on their addition f§}. The method, which is described in section 3.2, involves the
use of an orographic model for evaluating the orographic component .
Another approach is to estimate PMP for the relatively flat regions adjoining
the mountains. Modifications for terrain influences are then introduced on the basis
of differences in storm rainfall data, both in the project basin and surrounding areas,
and on sound meteorological judgment derived . from storm analyses JJ,, 4, 5, lJ]. The
procedure is described in section 3.4 and Chapter 5.
ESTIMATES FOR OROGRAPHIC REGIONS 51
The remainder of this chapter presents details on procedures used in applying

the methods mentioned in the two preceding paragraphs. The general principles involv-
ed are discussed, and examples given from published reports. Thus, t he examples neces-
sarily are for a particular set of conditions; namely, a certain amount of avail-
able data, certain terrain characteristics, and, last but just as important, the mete-
orological characteristics of the major storms in the regions for which the studies
were made.
3.2 Orographic separation method
3.2.1 Introduction
The orographic separation method for estimating PMP makes use of an orographic
model for computing orographic storm precipitation. The conditions under which the
model may be used have been found to be relatively limited, and caution in its use is
advised. Despite its limited applicability, a great deal of space is devoted here to
its description and use as these have never yet been published with the degree of de -
tail allotted to other procedures described in available reports on PMP estimates . The
evaluation of the convergence component of storm precipitation for the orographic
separation method is described in this section also.
3.2.2
Precipitation released when moist air is forced over a relatively unbroken

mountain ridge is the result of a basic process which can be idealized and treated as
a two-dimensional problem. The air passing over the mountain crest must accelerate
since there is a shallower layer within which air from a deeper upwind layer must be
passed. This process has led to an orographic precipitation model in which the air
flow, assumed to be laminar, is lifted over the mountain ridge. The model is a
storage evaporation in that the resulting precipitation is the difference between the
water vapour inflow at the base of the mountain range and the outflow above the ridge.
At some great height, called the nodal surface, air flow is assumed essenti-
ally horizontal. The height at which this occurs can be computed theoretically["~.
In general, this height is between 400 and 100 mb for moderately high .barriers. A
simplified diagram of inflow and outflow winds over a mountain barrier is shown in
Figure 3.1.
The model considers the flow of air in a vertical plane at right-angles to a

mountain chain or ridge. It is what is termed a two-d imensional model . The plane
has a "y" co-ordinate in direction of flow and a "z" co-ordinate in the vertical. The flow
may represent an average over a few kilometres or tens of kilometres in the transverse,
or "x'; direction, which does not appear explicitly in the model. The wind at ground
level moves along the surface. The slope of the air streamlines above a given point
on the mountain slope decreases with height, becoming horizontal at the nodal surface.
Outflow wind profile

Nodal surface (component)
--r--~-=---"' v
·r
Average barrier height
(mb)
--~~Generalized ground profile

1 000 rnb surface
Figure 3.1 - Simplified inflow and outflow

wind profiles over a mountain barrier
If it is assumed tha t the air is saturated and that temperature decreases

along the rising streamlines at the moist adiabatic rate, and the flow is treated as a
single layer of ai r between the ground and the nodal surface (Fig~re 3.2), the r ate of
precipitation is then:
R = (3.1)
y
1
where R is the rainfall rate in cm sec- ; Vi, the mean inflow wind speed in cm sec-!
W1, W2, the inflow and outflow precipitable water (liquid water equivalent) in cm; Y,
the horizontal distance in cm; and Llp 1 , ~ p2, the inflow and outflow pressure differ-
ences in mb.
Nodal surface
j
I
Figure 3.2 - Single layer, laminar flow, wind

model
Equation (3.1) is a storage equation, i.e., precipitation equals inflow of

water vapour minus outflow of water vapour. It may be derived as follows. Consider
the mass transport through the slice of space bounded by two identical vertical planes,
as in Figure 3.2, a short horizontal distance, s, apart. The storage equati on for
water vapour is:
(3 .2)
1
where Mr is the rate of conversion of water vapour to precipitation in gm sec- ; (Mv)1,
the rate of inflow of water vapour in gm sec-1. and (Mv) 2 , the rate of outflow of
water vapour in gm sec -1
These terms are:

Mr = RYsp, (3.3)
(Mv)1 = V1W1 sf', (3.4)
(Mv)2 = V2W2sl', (3.5)

-3
where f1 is the density of water, which is 1.0 gm cm The mass of air flowing in
equals the mass flowing out if no allowance is made for the mass of precipitation which
falls, which is relatively very small and may be negl ected. The continuity equat ion
is ex pressed by
(3.6)
Combining the last five eq uations and solving for R yields equation (3 . 1).
Greater precision requires dividing the air into several layers of flow, as
in Figure 3.3, rather than treating it as a single layer. Equation (3 .1) applies to
each of these layers. Total precipitation is then obtained by adding the rates from
all layers. With several layers, it is more convenient to use the storage equation
in the following form:
V1~Pl (q1 - q2) l (3 .7)

R =
y gl'"
where V1 and ~Pl refer to the inflow in a particular layer, and ql and q2 are the mean
specific humidities, in gm kg- 1 , at inflow and outflow, respectively. Mixing ratio,
w, is often substituted for specific humidity, q. The terms g and ,o refer respective-
ly to acceleration of gravity in cm sec-2 and density of water in gm cm-3.
Nodal surface
------~1)---- a·c
\9 .
~
Figure 3.3 - Multiple layer, laminar flow,

wind model
Equation (3.7) derives from the relation between specific humidity a nd pre-
cipitable water:
w = ~ (3 . 8)
g,,a
Substituting this relation into equation (3 . 1) yields
R = (3 .9 )
y
which reduces to equation (3 . 7) .
An approximate relation often substitut ed for equation (3.7) is:
0.0102 v1 6 p1 cw-1 W-2 )

R ~ (3.10)
y
wher e R i s the r ainfall rate in mm hr-1; v1 is the mean inflow wind s pe ed in knots;
~P l i s the pressure differer1c e between the top and bottom of an inflow lay er in mb;
w1 and w2 ar e the mean mixing rat io s in gm kg-1, at inflow and outflow, resp ectively ;
and Y is the horizontal length of th e slope in nautical miles (n miles) .
Relation (3 . 10) derives from the approximate r e lation between mean mi xing
ratio, w, and precipitable water, W:
w -;::::; 0.0102 -w Ap (3.11)
whe r e W is in mm; · w in gm kg-l; -6 p in mb; and the coefficient, 0 . 0102, has the di-
mensions mm mb-1 kg gllll Substituting this relation into equation (3.1) and using
larger units of V and Y yields relation (3.10) .
The distribution of precipitation along a windwa rd slope requires construction

mf anow and raindrop trajectories from the level of their formation to the ground .
These trajectories are considered along with streamlines of the air flow ove r a ridge,
as shown in Figure 3 . 3. Th e computation of precipitation trajectories is de scribed in
the following example of a test of the orographic model against observed storm rainfall.
3.2 . 3
The following example of th ~ 11se of the model was selected from PMP s tudie s
for the Sierra Nevada und Co ~ caclr~ Ffor1g. : near the west coast of the United States [8,
197. Figur ~ 3.4 ~ how s a maµ o f th e t ?st area with some of the precipitation stations.
Figure 3.5 shows th e smoothed a v ~ roye uround elevation profile used for th e comp uta-
ti on so Th e elevations of til e 1· rec tpi tat ion stations are plotted to show how well th ey
120°
121°
+ + 40°
1000
•
Oroville R. 5. Reno•
Wheatland 2 NE __l_ 39°

' '"
~9000 c %/.;
9\)'lo r;~/
·, .
Ridge
•Sacramento 1000
Height contoun in feet
+121°
Figure 3.4 ~ Blue Canyon orographic model test

area in California
-1
2 400
Ridge elevation 2 450 m •

2 000 -
1 600
,,
c:
......0.
0
>
~ 1 200
1.1.J •
•
•
800 • •
400
~
i
•
1
10
·~-·~,__J--+----"-----"---'--~·-_J
20 30 40 50 60
Distance (n mile)
Figure 3.5 - Precipitation station elevations

relative to adopted ground profile for test
area of Figure 3.4
fit the profile. The storm period selected for testing was the six-hour period ending
at 2000Z,22 December 1955. The 1500Z,22 December upper-air sounding at Oakland,
Calif . , approximately 160 km south-west of the inflow end (sou th-west side) of the test
area, was used for inflow data . Precipitation computations will be shown for the last
segment, or portion, of the windward slope near the crest. The following steps are
recommended in computing orographic precipitation over the slope.
3.2.3.l Ground profile

-------
Determine the ground profile of the area under consideration and divide into
segments at each break in the profile. Long segments may be subdivided . In Figure
3.6, since the slope is fairly uniform, the first nine segments, or legs, have been
made of equal length, 6 statute miles or 5.2 n miles . The length of the last leg is
4 statute miles or 3.5 n miles, so total distance from inflow to outflow is 50.3 n
miles .
300
Nodal surface at 350 mb
'S
.._,
6 0 0 1--...,,=:i=----+---
(!)
!-!
~
rn
rn
(!)
!-!
~ 700L--~=+---+-- -l---t..--=--r
. ._Crest
...
•' : . .. ·
·.· ·. . .: ··..... :. :
. . - . .· -:
. :.....· ...: ·.. .
"
~
. . . . ·. . '
~ ..
. ..
. .. ...
' . ~.
.·
.
. . . .· .. ·. .
.. . .· '• '.-·:: _.._ .·.
- .. ·.
1000 ·• . ..
i,..,;:;;::::::;;:::::::::~~:.L..:~~.o...:.....:.~.:...:L:...o.:.....~·~
:: ·., : .• .,
· ·~··~·_:_,:_--'...i.._c...:_..:__:__.:_.:__'---'-'_.:_-'---'--.:...:._._'-'-'-___:.IL-------<
0 10 20 30 40 50 60
Distance (n mile)
Leg~ +2+3+1.+s+s+1+a+9+~
Figure 3.6 - Air streamlines and precipitation

trajectories for test case
Co~vert heights of ground profile (Figure 3 .5) to pressures by means of the

pressure-height curve constructed from the inflow sounding of pressure temperature
and :elative humi~ity: Plot these pressures at the end of each leg, ~nd draw grou~d
prof~le.as shown in ~igure 3.6 . (Until some way is found to take downslope motion
~f air into accoun~ in computing precipitation, it is recommended that any downslopes
in the ground profile be drawn horizontal . ) Construct verticals at the inflow and
outflow ends of the model and at the end of each leg .
3.2 . 3 . 2 Inflow data
The inflow data used in the example are tabulated in the first eight columns
of Table 3.1. These data were obtained from the sounding. The wind speeds are the
components normal to the mountain ridge, ioe., V = V0 cosO( , where V0 is the observ-
ed wind speed from the observed direction and 0( is the angle between the observed
direction and the normal to the ridge.
3.2.3.3 Air streamlines
Space the streamlines at thi inflow vertical in the manner indicated in

Figure 3.6. There, the first streamline above the surface streamline is set at l 000
mb. Streamlines are then spaced at 25 mb intervals up to the 800 mb level, thence at
every 50 mb up to the nodal surface, which is assumed to be.at 350 mb'. Streamlines.at
the outflow vertical and intermediate verticals are spaced in proportion to the spacing
at inflow. Spacing may be done either graphically or by mathematical interpolation.
3.2.3.4 Freezing level
As the air travels along any streamline, its pressure, temperature and mixing
ratio at any point on the streamline may be determined from a pseudo-adiabatic chart.
Determine the pressure at the freezing point on those streamlines where the zero 0 c
temperature occurs between inflow and outflow. (See discussion of Table 3.2 in sec-
tion 3.2.3.5,) Plot these points on their respective streamlines, and draw freezing
line as shown in Figure 3.6. Precipitation is assumed to fall as snow above the
freezing line and as rain below .
The path followed by falling precipitation particles is determined by three

components: (1) vertical fall due to gravity, (2) horizontal drift caused by the hor-
izontal component of the wind, and (3) vertical rise resulting from the upward compon-
ent of the wind as it flows along the streamlines.
The average falling rate of precipitation particles in orographic storms

affecting the test area has been taken as 6 ms-1 for rain and 1.5 ms-1 for snow. For
computational purposes, these values have been converted to 2 160 and 453 mb hr-l ,
respectively.
The hori zonta l d!:_ift of orecipitation par!_icles while falling from one
streamline to another is Vl:::.p/rate of fall, where Vis the mean horizontal wind speed,
i n knots, in the layer between streamlines; Ap is the thick~ess of the layer in mb;
and rate of fall is in mb hr-1 . Since YAp is constant between any two streamlines,
drifts computed at inflow may be used anywhere between the same two streamlines. In
Table 3.2, horizontal rain drift (DRR) and horizontal snow drift (DRS) between stream-
lines are shown in column s (6) and (7). Drifts are in nautical miles (n miles) since
V is in knots (kn) . The effect of the upward component of the wind is automatically
taken into account by the slope of the streamlines.
Precipitation trajectories are computed f r om the ground up, starting at t he

ends of the selected legs of the ground profile. Plotting points for two trajectories
are computed in Table 3.2: one, called upper (UT), beginning at outflow, or 50.3 n
miles from inflow; and the oth er , called lower (LT), beginning at the end of the ninth
leg, or 46 . 8 n miles from inflow. Columns (8) and (9) of Table 3.2 give accumulated
horizontal drifts from the vertical passing through the ground point of each of these
trajectories. Columns (10) and (11) give corresponding distances from the inflow
vertical.
Rain drift is used below the freezing level; snow drift, ab ove. By coin-
cidence, t he lower trajectory (Figure 3. 6 ) reaches the freezing level approxi ma tely
where t he latter intersects a streamline. The upper trajectory, however, reaches the
freezing level between the 850 and 825 mb inflow streamlines. Hence, a streamline pass-
ing through the intersection of this trajectory and the 0°C line is const r ucted. This
streamline intersects the inflow vertical at 831 mb. Since the snow drift in the 831
to 825 mb layer is 0.65 n miles (Table 3.2 ), the total dr ift measured from the outflow
vertical to the 825 mb streamline would be 2.95 + 0.65 = 3.60 n miles, which would take
the trajectory below the freezing level. Hence, total drift was assumed to be 3.47 n
mile~ which means that the drift within this layer was assumed to be 0.52 n miles rather
than 0.65. Since the snow in this layer is probably very wet, the falling rate is
likely to be between that for snow and that for rain, and the above assumption appears
warranted .
After constructing the precipita t ion trajectories, compute the total volume
of precipitation under each trajectory, layer by layer. Subtract the total volume
under one trajectory from the volume under the next higher one, and divide the differ-
ence by the horizontal area of the ground on which this volume falls to obtain the
average depth over this area.
If relation (3.10) for rainfall rate is multiplied by the area, XY, it yields
the !-hour -rainfall volume. The Y's in the numerator and denominator cancel, and if
area width, X, is taken as l n mile, the 1-hour volume, R (XY), or Voll-h' under a
particular trajectory is approximately
(3.12 )
where w' is the mean outflow mixing ratio at the trajectory (see q' in Figure 3. 3).
Table 3.1 - Computation of orographic precipitation over leg 10 of Blue Canyon , California ,
test area for the 6-hour period 1400-2000Z, 22 December, 1955
(Hand computation, using 1500Z, 22 December sounding at Oakland, California, as
inflow data and assuming a nodal surface of 350 mb)
Inflow data
'-"'LT• 6v ur"
p '!' P.11 v v ii,. P ws w
I pc PLT "LT Pur "ur WI wl.T "ur "r - "Lr V~p1\;ILT "1 - "ur V6p6"ur
(1nh) (' C) (Z) (kn) (~o) (g kgl)
500 -12.3 77 61.6 2.96 2.28 475 496 2.28 495 2.28
S9 . 6 2 980 2. 70 2.70 2.70 0 0 0 0 IT1
550 -8. l nl 57. 4 3.80 3.12 529 537 3.12 536 3.12 Vl
62. 7 3 135 251 2 ~2
-I
J.61 3.53 3.52 .OS .09 H
600 - 11.2 8a 67 .9 4.65 4.09 583 575 3.94 S74 3.92 3:
62.8 . 3 140 4.6 4 4.22 4 . 20 .42 l 319 .44 l 382 )>
650 -0.6 92 57 .6 5 . 64 5.19 638 604 4.50 602 4.4'/ -I
IT1
5S . l 2 755 5.72 4.73 4.69 .99 2 727 1.03 2 838 Vl
700
750
2. 6 94 S2.6
lo9.8 2 490
6 . 64 6.24 692 630 4.95 628 4.90
6.69 5.18 5.13 1. 51 3 760 1.56 3 884 ...,
5.3 95 47.0 7.50 7.13 742 6S6 5.40 654 5 . 36 0
2 505 5 261 ;;o
SO. I 7.55 5 . 51 5.45 2.04 5 110 2.10
800 7. 9 95 53. l 8.38 7 .96 792 672 5.61 669 5.54 0
51.4 l 285 8.20 5. 75 5.57 2. 45 3 148 2. 6 3 3 380 ;;o
825 9.1 96 49.6 8. 79 8. 44 817 688 5.88 672 5.60 0
49 . 2 295 8.50 5.92 5.61 2 . 58 761 2. 89 853 G">
831
;;o
9. 4 96 48. 7 8.92 8.56 823 693 5.95 673 5.62 )>
4 7. 2 897 8. 75 6.09 5.69 2 . 66 2 386 3 . 06 2 745 "C
850 10 . 3 % 45. 7 9. JO 8.93 643 703 6.22 680 5. 76 :::c:
44 . 2 l 105 9.13 6 . 34 5.86 2. 79 3 083 3. 27 3 613 H
(")
875 11. 4 96 42. 7 9. 71 9. 32 868 718 6.45 694 5.95
42. 7 1 068 9.46 6.51 6.0Q 2.95 3 151 3.io6 3 G95 ;;o
900 12 . 5 94 42. 7 10 . 20 9.59 888 732 6.57 705 6.05 IT1
41.9 l 048 9.69 6.59 6.06 3.10 3 249 3.63 3 804 G">
H
925 13.4 93 41. l 10.52 9. 79 911 746 6.60 717 6.07 0
37 .6 94-0 9.81 6.64 6.09 3.17 2 980 3. 72 3 497 z
950 14 .2 91 34.1 10.80 9.83 929 760 6.68 729 6.10 Vl
29 . 9 748 9.63 6.57 5.94 3.06 2 289 3. 69 2 760
975 IS . O 8S 2S . 7 11.10 9.43 941 776 6.46 740 5.78
19.4 485 9.42 6.42 5.76 3.00 1 455 3.66 I 775
l 000 15. 5 8.\ 13.1 .11.20 9.41 961 790 6.37 753 5.73
11. l 56 9.55 6.48 5.80 3.07 172 3. 75 210
l 005 I S. 7 66 9 .1 11.27 9.69 971 793 6.58 758 5.87
Legend 35 841 39 979

I
RH = Re l ativ e hu mi d i t y 6-hour volume ( mm(n mile)2) = .061 2 x I 2 193 2 447
w5 = Soti.rrotion mi x ing ratio Unit-width horizontal area (n mile )2 46.8 50.3
wI = Mi xing ratio c t inf!ow 6-hour overage rain fall (mm) 47 49
Pc = Con de nsati on pr ess ur e (2 447-2 193)/(50.3-46.8) 73 mm
LT = Lower precipitation trajector y
6-hour overage rainfall over lost leg
LIT = Upµ e r prec .i pitotio n traj e c t or y
Meaning of other s ymbol s obvi o us
°'
......
The orographic model is generally used to compute rainfall by 6-hour incre-

ments, so relation (3.12) becomes
;· ), (3.13)
1
where Vol6-h is in mm (n mile) 2 ; V1 in kn; A pl in mb; (wl - w') in gm kg- ; and co-
efficient 0.0612 has the dimensions (n mile) h \6 h)-1 kg gm-1 mm mb- 1
Table 3.1 shows the computation of orographic rainfall under the two precipi-
tation trajectories shown in Figure 3.6. The following example demonstrates how the
table was prepared.
Consider the layer between the streamlines passing through inflow pressures
850 and 875 mb {Llp = 25 mb). The air at 850 mb has a temperature of l0.3°C, relative
humidity 96 per cent, and horizontal component of wind speed parallel to the sides of
the selected ground area of 45.7 kn. Plotting l0 .3°C at 850 mb on a pseudo-adiabatic
chart, the saturation mi xing ratio is seen to be about 9.30 gm kg-1. The actual mix-
ing ratio is 96 per cent of this, or 8.93 gm kg-1.
From Figure 3.6, the pressures where the streamline through 850 mb intersects
the two precipitation trajectories are seen to be 703 and 680 mb. Following the dry
adiabat through 850 mb and l0.3°C upward to where it crosses t he saturation mi xing
ratio of 8.93 gm kg-1, the condensation pressure is seen to be about 843 mb and the
temperature 9.6°C (not shown) . Since the air is now saturated, the moist adiabat is
followed upward from this point. The saturation mixing ratio on this moist adiabat is
about 6.22 gm kg- 1 at 703 mb and about 5.76 gm kg-1 at 680 mb. The mixing ratio values
on the 875 mb streamline at the lower and upper precipitation trajectories are found in
the same way.
For the 850 - 875 mb layer, Vis then seen to be 44 . 2 kn, Vei.p = 1105 kn mb,
w1 = 9.13 gm kg-1, WLT= 6.34 gm kg-1 fo r the lower trajectory, and wuT = 5.86 gm
kg-1 for the upper trajectory. The decrease in mean mixing ratio of the layer from
inflow to lower trajectory,AwLT = 2.79 gm kg-land to the upper trajectory, Li. W'uT
= 3.27 gm kg-1. For the layer, the value of VL:.pAw is 3 083 (n mile) h- 1 mb gm kg-l
between inflow and lower precipitation trajectory and 3 613 (n mile) h- 1 mb gm kg-l
between inflow and upper trajectory.
After values of V ~p .6w are computed for all layers for all trajectories,
values for each trajectory are summed and multiplied by . 061 2 (n mile) h (6 h)- 1 mm
mb-1 kg gm-1 to obtain values in mm (n mile)2 (6 h}f In Table 3.1 these values are
2 193 for the lower trajectory and 2 447 for the upper . Division by the areas over
which these volumes fall gives average depths for those areas. Since unit width is
assumed for Figure 3.6, any such area is numerically equal to the sum of the lengths
of the legs between inflow and a given precipitation trajectory. For the lower tra-
jectory this is the sum of the lengths of legs 1-9 or 46.8 (n mile) 2 , which makes the
6-hour average depth over those legs 47 mm . For the upper trajectory the volume falls
over legs 1-10 or 50.3 (n mile)2 , giving a 6-hour average depth of 49 mm. The volume
that falls on leg 10 alone is the difference betwe e n th e volumes under upper and lower
trajectories or 254 mm (n mile)2 ( 6 h )- 1 . This is distributed over 3.5 (n mile) 2 ,
which makes the 6-hour average depth 73 mm .
LSTlMATES FOR OROGRAPHIC f\E GIONS 63
Table 3 . 2 - Computation of r ain a nd snow drift for computing precipitation

trajectori es over Blue Canyon, California, test area
(Based on sounding ci f 1500Z, 22 December 1955 at Oakland, California)
Inf low do to (UT) (LT)

(UT) (LT)
50.30- 46. 80-
p v v O,P V.op D~R DRS £.DRI FT [DRIFT
£DRIFT l:DRIFT
(mb) (kn) ( kn) (mb) (n mile) (n mile) (n mile) (n mile) (11 mile) (n mil e )
~<1~)_ _(~2~)-~<~3~)- (4_) ( 5) .~(_6~)~-(~7~)~-'-(8=,)'--~~(~9~)~--'QO_)~~~(l_l~)-

350 97.7
81. 9 50 4 095 1. 90 9 . 04
400 66.l
68. 6 50 3 430 1. 59 7. 57
450 71.0 51.18• 48. 55'' -.88 -.l. 75
66.4 50 3 320 1. 54 7. 33
500 61. 8 43.85• 41. 22* 6.45 5.58
59.6 SC 2 980 1. 38 6. 58
550 57.4 37.27* 13.0 3 12.1 ~
62.7 50 3 135 1.45 6.92
600 67.9 30. 35* 27. 72* 19.95 19.08
62.8 50 3 140 1.45 6.93
650 57.6 23.42* 20. 79* 25.88 25.0J.
55.1 50 2 755 1. 28 6. 03
700 52.6 17. 34• 14. 71* 32.96 32.09
49.8 50 2 490 1.15 5. 50
750 47.0 11.84• 9. 21* 33.1,6 37.59
50.1 50 2 505 1.16 5 . 53
800 53.l 6. 31• 3.68 43.99 43. l 2
51.4 25 l 285 0.59 2. 84
825 49.6 3. 47« 3.09 46.83 43.71
49.2 295 0.14 .65
831 48. 7 2.95 2. 9 5 4 7. 35 43.85
47.2 19 897 0.42 1.98
850 45.7 2.53 2.53 47. 77 44.27
44.2 25 1 105 0.51 2.44
875 42.7 2.02 2.02 43.28 44. 78
42.7 25 1 068 0.4? 2.36
900 42.7 1.53 1.53 48. 77 45.27
41. 9 25 1 048 0. 49 2. 31
925 41.l 1.04 49.26 45.76
37 . 6 25 940 0.44 2.08
950 34.l 0 . 60 49. 70
0.60
29.9 25 748 0.35 1.65
975 25. 7 0.25 50.05 46.55
0 . 25
19.4 25 485 0.22 1.07
l 000 13.1 0.03 0.03 50. 27 46. 77
11.1 56 0.03 0.12
l 005 9.1 0 0 50. 30 46.BO
*Using snow d~ift
*AArbitrary (to keep trajectory on or above freezing line)
Legend
ORR = V.~ p/2160 = Ho riz ontal r~in drift
DRS ~ VOp/453 = Horizontal snow drift
UT w Upper prccip~t~tion trajectory
LT~ Lower prccipitatjo~ trajec tory
The above procedure has been computerized to facilitate complete computations

for numerous areas and soundings. Another computerized version of the orographic
model is somewhat more sophisticated than the one just described, Whereas in th e ex-
ample model the height of the nodal surfac e was assumed and an approximate method used
for spacing streamlines at the outflow over a mountain crest, this second computer
mod e l uses a nodal surface and stream lin e spacing based on physical law s of air flow
LY. The outflow approximations used in the above example give results comparable to
those of the mor e sophisticated model. Table 3.3 compares the resu lts yielded by the
two comput erized models for each of th e ten legs for a 6-hour pe riod and by the manual
application just de sc ribed for th e t e nth l e g.
..,..
°'
Table 3.3 - Comparison of observed and compured 6-hour precipitation for the period 1400-2000Z,
22 December 1955 over Blue Canyon, California, test area l"T'1
Vl
-I
H
3:
)>
-I
H
0
z
0
"Tl
Average ""C
Leg l 2 J 4 5 6 7 • 8 9 10 1-10 ;;o
0
Horizontcl length of leg (n mile) 5.2 5.2 5 ., 5.2 5.2 5.2 5.2 5.2 5.2 J.5 co
)>
(n mile) 5.2 10.4 15.G 20 . 8 26 .0 31.2 36 . 4 41.6 46. 8
co
Cumulctive length 50.3 r
l"T'1
Elevation at end of leg (ft) 590 l 200 l 780 2 320 3 210 4 080 4 640 5 540 6 760 8 030
366 543 707 978 l 244 l 689 3:
(m) 180 ~ 414 2 060 2 448 )>
x
Observed precipitation (mm) J 6 1J 25 38 46 55 64 67 65 37 H
3:
Machine-computed precipitation l (mm) 0 14 40 44 55 66 54 60 67 72 46 c
3:
Machine-computed precipitation 2 (mm) 1 17 44 45 56 66 55 59 67 69 47 . ""C
;;o
Hand-computed precipitation (mm) 73 49 l"T'1
("')
H
""C
H
Elevation at beginning of first leg = 200 ft (61 m) :;!
-I
H
0
~ z
Mochine-compu~e~ precipitation 1 used spacing of streamlines by a method developed by Myers (1) .
Machine-computed precipitation 2 used spacing of streamlines between surface and 350 mb nodal surface
(assumed), along any vertical, proportional to their spacing at inflow.
Hand-computed average precipitation over leg 10 and legs 1-10 based on same spacing of streamlines as
machine-computed precipitation 2.
The so-called observed precipitation used in the comparison of Table 3.3 re-
fers to the orographic component only. Ordina rily, this would be obtained by sub-
tracting from the observed total precipitation for each leg the precipitation measured
in the flat volley upwind of the test area during the 6-hour period of the test. This
volley prec ipitation (convergenc e component of total precipitation), whic h is sometimes
reduced for elevation, is attributed to atmospheric processes not directly related to
orography. In the test case described, howeve r, there was no appreciable volley pre-
cipitation so no deduction was mode from observed precipitation.
3.2.3.8 Sources of error
Differences between precipitation computed by the model and observed orog rap h-
ic precipitation (total precipitation minus convergence component).con be attributed to
two main sources: (a) errors of input to the model, and (b) sparsity and unrepresento-
tiveness of precipitation data for checking model computation .
Input to the model . Usually, no more than two upper-air observations ore
mode doily . Despite utmost core in interpolating for a particular storm period by
referring to the more frequent surface synoptic charts , the question remains as to the
representativity of instantaneous wind and moisture values for even a short period of
a few hours . Such inaccuracies lead to errors in computed amounts of precipitation.
In the example given, no allowance was mode for the fact that the upper- air
sounding station (Oakland) is approximately 160 km from t he test a re a, and moistu re
and wind values were token directly f r om the sounding . Attempts to adjust for wind
travel time (averaging less than two hours) did not improve results.
Observed orographic precipitation. The uneven distribution of storm precipi-

tation, both with respect to time and space; the sparseness of the precipitation net-
work; and the usual errors of gouge measurements make it difficult to obtain re liable
averages of storm precipitation on slopes . Also, most gouges in orographic regions
are located in narr ow volleys or on relatively flat sites unrepresentative of nearby
elevations or the generalized gr ound profile. Their measurements, while perhaps ac-
ceptably representative of actual precipitation at the gouge sites, ore unlikely to re-
present with any great accuracy the overage precipitation falling on the general slope.
These various factors make it difficult to obtain reliable values of observed storm
precipitation on a slope for comparison with model computations .
3.3 Orographic separation method for estimating PMP
Reference was made earlier to the fact that precipitation in mountainous

regions consists of two components: (1) orogrophically induced precipitation (oro-
graphic precipitation), and (2) precipitation produced by atmospheric processes un-
related to orography (convergence precipitation) . PMP is computed therefore by maxi -
mizing and adding the two precipitation components. Caution must be exercised to
avoid over-maximizing.
3.3.l
The procedure used in applying the orographic model for computing the oro-
graphic component of PMP is the same as that used in testing the model (section 3.2.3)
with the exc eption that in flow winds and moisture are maximum values.
3 . 3.l. l Maximum winds
If there is a long r ecord of upper-air winds, say 30 years or longer, an en-

velope of the highest recorded speeds for winds from critical directions for each month
or part of month is usually adequate. The probability of occu rre nce of any of the en-
velope values may be determined by statistical analysis. Such analysis may be used
also to estimate high wind speeds, say for a 50-year return period, when the record is
so short as to introduce doubt as to its maximum values being representative of those
to be obtained from a longer record. If the record is so short, say less than ten
years, as to preclude reliable frequency analysis, maximum wind speeds may be estimated
from surface pressure gradients betwee n suitably located stations. Maximum surface
winds so determined may then be used to estimate upper-air wind speeds by means of em-
pirical relations ["8_].
Figure 3.7 shows the maximum wind speed profile used for the coastal region
of California. The variation with duration (Figure 3 . 8) was based on that of geo-
strophically derived winds and that of 900 mb winds at Oakland during selected storm
periods.
300
L.00 /
/
500
.0
/
a
...__,,
Q)
600 /
fi
;j
m
~·
/
[/J
700
Q)
fi
P. / Legend
A Envelope of record winds at 4
800 / stations
B 50-year wind speed at Oakland
c Adopted profile
900 D Geostrophically derived wind
speed
1 000
0 40 so 120 160 20 0 24 0
Wind speed (kn)
Figure 3.7 - Maximum one-hour wind profile and

supporting data
'd
<1> 100
<1>
Pl
CJl
'd
~ 90
•.-!
~
!-I
;:j
0
..r:: 80
I
<1>
~
0
§ 70
a
·.-!
I><
<II
a 60
<+--!
0
<1>
Ill)
<II
+> 50
~
<1>
0
!-I
<1>
P-t 40
0 12 24 36 4S 60 72
Duration (hours)
Figure 3.8 - Variation of maximum six-hour

wind speed with time
3 3.1.2 Maximum moisture
Maximum values of moisture ore obtained from maximum persisting 12- hour l 000
mb dew points . A full discussion of these dew points is given in section 2.2 .
3. 3. 2
One method of applying the model for developing generali zed estimates of PMP
is to define terr ain profiles over the entire region of inte r est . If the topography
is relatively uncomplicated and all general windward slopes face one most critical
moisture- inflow direction, as in the California Sie r ra Nevada, application of this pr o-
cedure presents no special problems .
An alternat i ve method is to use the model to compute PMP for selected terrain
profiles and to evaluate PMP between them by means of maps, such as seasonal or pre-
cipitation- f r equency maps, adequately depicting the geographic distribution of precipi-
tation. In this approach it must be shown first that there is good correlation be-
tween computed orographic PMP on the selected computation profiles or areas and th e
values indicated by reference maps used for interpolation.
A somewhat different approach ha s been used !JQ] for regions where the opti-
mum moisture-inflow direction and orientation of slopes varied from place to place.
The procedure consists of computing PMP for terrain profiles oriented in different di-
rections and then enveloping the greatest values regardless of inflow direction or
slope orientation. Relations are then developed for adapting the envelope values to
inflow directions and slope orientations critical for a specific basin . A simple but
adequate method for making such adaptations is to use a variation with basin si ze ,
since the variety of optimum inflow directions and slope orientations tends t o increase
with size of area. This type of adjustment was used in a study for the north-western
United States /Jffl. In the California study ~'Q], the adjustment was based on the de-
crease of moisture with increasing width, or lateral extent, of inflow in observed
major orographic storms (section 3.3.3 . 3).
Generalized estimates of PMP are usually presented on an index map showing

isohyets of PMP for a particular duration, size of area, and month. Relations are
then provided for adjusting the mapped PMP values to other durations, basin sizes, and
months.
Figure 3.9 shows the January 6-hour orographic PMP index map developed in the
aforementioned California study. This particular map does not specify an area size .
In this case, the average index value for a ny specified basin is obtained by laying an
outline of the basin on t he index map and then estimating the average of the values
within the outline. No further areal adjustment is required unless the width of the
basin exposed or normal to the optimum moisture inflow exceeds 50 km (section 3.3.3.3).
3.3.3
As mentioned above, PMP varies with region, season, duration, and si ze of area.
The generalized maps show the reg ional variation, and no further discussion is required.
While the discussion of the other variations presented in this section applies partic-
ularly to the orographic separation method, especia lly as used in the California study
given as an example, much of it applies to variations of orographic PMP in general.
3.3.3.l Seasonal variation
In any region where snowmelt is likely to contribute significantly to the

probable maximum flood, it is necessary to determine the seasonal variation of PMP.
In orographic regions the seasonal variation should be determined even when snowmelt
is not involved in order to insure that the month of highest potential for total PMP
(orographic plus convergence) has not been ove rlooked. A logical procedure is to com-
pute PMP for each month on the basis of maximum values of wind and moi st~re ~n each
month. The seasonal variation of major storms recorded over a long period is gener-
ally a useful guide in delineating the seasonal variation of PMP.
Evaluation of orographic PMP by means of the model has several shortcomings.

In the transitional seasons (spring and autumn), the usual orographic influences pre-
vail, but stimulation of storm precipitation by upwind slopes or barriers is often
most effective in det erm ining pr ecipitation distribution . The need for gene r alizing
topograp hy leads t o differen ces betwe e n computed orog r aphic PMP an d that indicated by
the actual terrain. For different terrain profiles, seasonal influences may vary
with barrie r height , steepness of slope, and other features. In some cases, a com-
promise between seasonal variation indicated by computed PMP values and that based on
maximum storm rainfall amounts observed at well-exposed stations may yield the most
realistic results.
Figure 3.9 .- Six-hour orographic PMP (in) for

January. (Square delineates Blue Canyon oro-
graphic model test area)
3.3.3.2 Durational variation
Variations in maximum wind speeds and moisture with time are used to determine
durational variation of computed orographic PMP. The variation of winds in major ob-
served storms is probably the best type of information to use in establishing varia-
tions in the shape of the inflow profile with duration, and this was used in the ex-
ample study . Variation of moisture with time was based on the durational variation
of maximum persist ing 12-hour l 000 mb dew points /J~ . Moisture values at upper
levels were based on the assumption of a saturated pseudo-adiabatic lapse rate. A
common durational variation (Figure 3. 10) for all months and regions was adequate for
the example study. An additional facto r found helpful in some studies JJQJ
is the
variation of moisture with duration during major observed storms.
fQ
J..t +>
.;:.~ 110
0
J..t Pt
Q)
+> ;.
I'll Q)
;. 'd
Q) Ill) 100
r-l s::
..0 •n
cld +>
+> Ill
•n •n
Pt Ill
•n J..t
0
Q)
Q)
Pt 90
J..t
p, ..0
..... a
00
0
Q) 0
Ill)
cld r-1
80
+> (Based on highest persisting dew points)
~ ~
0 0
J..t .q
G> I
Pi N 70..._~~~~~~~~~~~~~~~~~~~~~~_.
r-l
0 12 24 36 48 60 72
Duration (hours)
Figure 3.10 - Decrease of maximum

precipitable water with duration
3.3.3.3 Areal variation
The variation of orographic PMP with basin size is controlled by the orography,
and therefore may vary greatly from basin to basin. As stated in section 3.3 .2, the
averaging of index PMP by superimposing an outline of the basin on the index map elimi-
nates the need for the usual type of depth-area relQtion. The average index PMP thus
obtained usually requires some adjustment for basin size, however, since the intensity
of moisture inflow decreases with increasing width of inflow. In the example study
/8.:J no adjustment was required for basin widths up to 50 km, but a reduction curve for
greater widths reduced the basin average index PMP by 15 and 25 per cent for widths of
160 and 300 km, respectively.
3.3.4
The procedure described here for estimating convergence (non - orographic) PMP
for co mbination with orographic PMP was developed for the coastal regions of Ca l if-
ornia ["'i[/, where the critical season for major orographic storms is October to
March . The approach, which has been used elsew here, is basically similar to those
used in estimating PMP for non-orographic regions. The greatest precipitation amounts
for various durations at stations in the least orographically influenced areas are
maximized for moisture. This is done in two steps. First, regional envelopes of
maximum persisting 12-hour l 000 mb dew points are determined for use in evaluating
maximum moisture, M, or precipitable water, W. Second, durational envelopes o f maxi -
mum P/M ratios at eac h station are determined for each month . Here, P is the storm
precipitation for a particular du rat ion ; and M, t he precipitable water for the repre-
sentat i ve persisting 12-hour l 000 mb storm dew point (section 2.2.4).
P/M ratios should be computed for several of the highest r ainfall s at any
particula r station because t he maximum rainfa ll does not necessarily yield the highest
P/M r atio . Maps of ma xi mum moisture and P/M rat io s are then drawn. Multiplication
of corresponding values from appropriate pairs of maps yields moistu re- maximi z ed rain-
fall amounts fo r any requ ired location, or (P/M)max multiplied by Mmax equals converg-
ence PMP.
3 . 3 .4. l ~o!s!u~e_(~e~ ~o!n!)_e~v!l~p!s
Maximum, or 100- year, persisting 12- hour l 000 mb dew points (section 2.2.5),
enveloped seasonally at each station (Figu re 3 .11) and smooth e d regionally (Figure 3.12)
are used to establish the level of maximum moisture available for evaluating conve r g-
ence PMP. In the example study /"8}, one mean seasonal variation curve (not shown) was
found applicable to the entire region of interest. Different seasonal trends for dif-
fe re nt portions of a region would increase only the details of application.
3 .3.4.2 ~n~e~oee:. ~f_P~M_r~t~o:

Finding suitable station precipitation data uninfluenced by orography is a
problem . In the example study, the search was confined to the large flat valley be-
tween the coastal mountains to the west and the Sierra Nevada to the eqst, and to some
coastal stations unaffected by nearby steep slopes . Except for a few short intense
r ainfalls, most data were observational-d ay or highest 24 consecutive 1- hour amount s.
Envelope cu r ves of highest P/M ratios found in the re strict ed reg ion are s hown in
Figure 3 . 13 .
Adequate data on intense rainfalls for establishing a seasonal trend in P/M

ratios would have been desirable, but there were not enough of these data in the prob-
lem area . However,many plots of maximum 24-hour precipitation at non-orographic
stations indicated no definite seasonal trend for any magnitude. On the other hand,
such trends did exist for 6- and 72-hour precipitation (Figure 3.14) .
,..-,,
0
0
20
' ' ...._Afe
...... 0 f)
......_sea
......_ - surfac -
18
- -
-------- -
e temperature
--- - - -
.....
-.......__, 12- hour dew point
16 ~..... -~
deW~n1 - - -
1
-......._ £!!.-_rear 12-hour
---- - - --
14
~~~~~~' -~~~~~ -~~~--'-~~~~-'-~--'

. Oct Nov Dec Jan Feb Mar Apr
Figure 3.11 - Seasonal envelope of maximum

observed dew points at Los Angeles, California
It was concluded that seasonal trends of moisture and P/M ratios for the 24-
hour duration must counteract each other since there was no trend in 24-hour precipi-
tation. On the basis of this concept, the greatest 24-hour P/M ratio was assigned to
February, the month having the lowest maximum precipitable water; and ratios for other
months were evaluated in proportion to their maximum precipitable water, as indicated
by their maximum persisting 12-hour dew points .
The ratios of 6- to 24-hour and 72- to 24-hour precipitation (Figure 3 .14 )

were used to establish P/M ratios for 6 and 72 hours. This was possible since 12-hour
moisture, the denominator M in the ratios, was used for all durations . The durational
variation of P/M ratios is thus the same as the durational variation in precipitation,
P. Monthly curves of durational variation of P/M ratios are shown in Figure 3.13.
3 .3.4.3 Reduction of convergence PMP for elevation
In the example study ~§], PMP values computed as described in the first two
paragraphs of section 3.3.4 were reduced for elevation. For gently rising slopes
where storm precipitation was apparently little affect ed by upwind barriers, the de-
crease in converg ence PMP was assumed to be proportional to the decrease of precipi-
table water, W, in a saturated column of air. This decreas e was computed as the dif-
ference between W in a column with bas e at the ground e levation at a po i nt 8 km upwind
ESTIMATES FOR CROGRAPHIC REGIONS 73
12 3 121 117
41
~,
' '\
''
15,5
'\ 35
I.
)
}
I
16
33
..../
117 115.
Figure 3 . 12 - Maximum pe r sisting 12-hou r

1 000 mb dew points (°C) for February
from the problem area and that with base at the ground elevation of the convergence PMP .
The 8 km distance upwind marks the average location of the formation of the storm pre-
cipitation particles falling on the problem area .
In estimating PMP by methods other than the orographic separation method, th e

usual procedure is to bas e the decrease on the difference between obse r ved storm
amounts on slopes and in valleys . In one study /JQ/, the non - orographic, or converg-
ence, PMP was reduced by 5 per cent for every 300 m increas e in elevation.
10
Feb
Jan
9
---=--
____,Nov
Mar
1
Dec
Apt
8
6'
0
•ri
.µ
Ill
F-1
5
::E:
..........
Pi
4
2 Based on maximum ratios for 14 storms (mostly for

24-hour amounts) and on 6/24- and 72/24-hour ratios
from Figure 3.13
0
6 12 18 24 30 36 42 48 54 60 66 72
Duration (hours)
Figure 3.13 - Maximum P/M ratios with

orographic storms
The amount of moisture that a column of air can contain is obviously reduced
by a shortening of the column as it crosses an orographic barrier. Convergence PMP
is therefore adjusted for the moisture depletion by upwind barriers. In making the
reductions, so-called effective barrier heights are used rather than actual heights.
Maps of effective barrier heights (Figure 3.15) differ from actual topographic maps in
that they tak e into account the effect of barriers on air crossing them . Also, since
the maps are intended for use in making generalized estimates of PMP, effective bar r ier
height contours naturally smooth out the smaller irregularities in crest height, ridge
orientation, and other orographic features . Local featur e s that would seriously
aff ect precipitation over small basins are thus smoothed out.
~
0
• Average of highest 6- and 24-hour storms for 21
·ri stations
+'
rd X Average based on generalized PMP for eastern U.S.
+'
-~ 0.70
·ri
C)
•
())
!-I
P<
!-I
::I
] 0,60
I
-.;I"
N
0
+' • •x
I
~ Q6oL-..~~~~~~--~~~~~~~~~~~~~~~~~~~~
1.60
• Average of 10 highest 24- and 72-hour rains for 50 stations
~
0
·ri
+'
rd
~ 1,50 •
P<
•ri
C)
())
!-I
P<
!-I
::I
] 1,40
I
-.;I"
N
0
+'
I
~ 1.3 0 L -_ _ _ __L__ _ _ __ L_ _ _ _ _- ' - -_ _ _ __ . __ _ _ __...._ _ _ ___.
Oct Nov Dec Jan Feb Mar Apr
Figure 3.14 - Ratios of 6- and 72- hour

precipitation to 24-hour precipitation
3.3 . 4.5 ~e~u~t~o~ ~f_p~i~tL ~r_2~ ~m:,_c~n~e:g:n:e_P~P_f~r_b~s~n_s:z:

Point precipitation data (arbitrarily accepted as representative for 25 km2)
were used in the derivation of convergence PMP described above . Ideally, the 25 km2
values would be reduced for basin size by depth - a r ea relations based on observed storms
that produced heavy convergence (non-orographic) rainfalls in the problem area. Spars -
ity of storm- centred data in non-orographic areas in the region of interest, however,
precluded the development of such relations. It was therefore necessary to develop
depth-area relations for extreme storms (excluding tropical storms) in regions where
orography had little or no influence on storm precipitation.
Figure 3 . 15 - Effective elevation and barrier

heights (1 000 ft) in northern California.
(Square delineates Blue Canyon orographic model
test area)
3.3.4.6 Cons!r~c!i~n_o! ~o~v~r~e~c~ ~M~ ~n~e~ ~ae
The steps described below for the construction of the six~hour 500 km2 con-
vergence PMP index map (Figure 3.16) for February in the example study apply equally
well to similar index maps for other durations, basin sizes and months if required.
Figure 3 . 16 - Six- hour 500 km2 convergence

PMP (in) for January and February. (Square
delineates Blue Canyon orographic model test
area)
Step 1 . After an appropriate grid had been drawn on a suitable map base, th e
maximum moistu r e for February was determined for each grid point and plotted thereon .
These maximum moisture (precipitable water) values were first obtained from the maxi-
mum persisting 12-hour l 000 mb dew points for February (Figu re 3 . 12), and then ad -
justed for effective elevation or barrier height (Figure 3.15) .
Step 2. The adjusted precipitable water value at each grid point was then
multiplied by the maximum 6-hour P/M ratio for February (Figure 3.13). The values
thus multiplied represent 6-hour 25 km2 convergence PMP.
Step 3 . The convergence PMP values computed as above were then adapted to
2
500 km by a reduction factor (0.80) obtained from the depth-area relation (not shown)
described in paragraph 3.3.4.5 . Isopleths were then drawn on the basis of these
areally reduced values to produce the index map of 6- hour 500 km2 convergenc e PMP shown
in Figure 3.16 . The factors involved in the construction of this map showed little
difference in January, so the index map was used without seasonal adjustment for both
January and February, and was so labelled.
3 . 3.4.7
The convergence PMP index map, constructed as just described, presents 6-hour
500 km 2 values for January-February. Relationships were developed for adjusting these
values for different dura.tions, basin sizes, and months. This was done as follows:
Step 1. Six-hour incremental values of maximum P/M ratios through 72 hours

were obtained for each month from Figure 3.13. These values were smoothed and ex-
pressed as percentages of the maximum six-hour P/M ratio for February.
Step 2. Durational (Figure 3.10) and seasonal variations of moisture (pre-

cip itable water), expressed as percentages of the 12-hour February moisture (based on
maximum persisting 12-hour dew points) and multiplied by the percentage variation in
P/M ratios (Step 1), yielded seasonal and durational variations for a point, or 25 km2.
Step 3. The areal variation (paragraph 3.3.4.5) was then applied to the
values obtained in Step 2 to yield a depth-area-duration relation for each month. That
for December is shown in Figure 3.17.
3.3.5
Total PMP is obtained by adding the orographic and convergence components.

Throughout the development of each component, care must be exercised to minimize the
possibility of over-estimating total PMP. In computing orographic PMP, for example,
the model should be tested against observed orographic precipitation only. Testing
may be restricted to storm periods showing little or no evidence of convergence pre-
cipitation, or the convergence component of total observed precipitation may be esti-
mated (section 3.2 .3.7) and subtracted from the total to obtain on estimate of the
orographic component.
In estimating convergence PMP, the measure of the storm mechanism, or effi -

ciency, is the P/M ratio computed from outstanding storms. As a precaution against
over-maximizing, only P/M ratios from general-type storms producing heavy orographic
precipitation should be used. Another precaution is to use only maximum persisting
12-hour dew points observed in major general-type storms for moisture maximization.
Percentages for other 6-hour periods

6-hour period : 4 5 6 7 8 9 10 11 12
18 16 13 12 11 11 10 10
December
1 000
-l (./)
-::!" fD
..... f')
0.. a
~
0..
en a>
I
I
-::!"
0 -::!"
,,.-.... a
(\J
c
..... c.....
ll "'O
Ii>
"O
..... Ii>
<11
.....
Q) a a
~ 0.. 0.
«:
25....__~..._~-'---L~---'---~..l.L-~__._~__._~-'-~--"~--'-~--'~~._._~_._~_..____,
20 40 60 80 100 120 140 160
Percentage of convergence PMP index
Figure 3.17 - Variation of convergence

PMP index with basin size and duration
for December
3.4 Modification of non-orographic PMP for orography
3.4.l Introduction
Two general approaches for estimating PMP in orographic regions were briefly
mentioned in section 3.1.5. One, the orographic separation method, was described in
detail in section 3.3. The other, as the title of this section implies, consists of
first estimating the non-orographic PMP for the mountainous problem region and then
applying modifying factor s for adjusting the non-orographic PMP for orographic effects.
The non-orographic PMP may be determined fo r the plains area in the region of inte rest,
or, if there are no br oad plains areas, it may be estimated as if the mountains did not
exist in or der to provide a working base.
While modification of non-orographic PMP is used more often than the orograph-
ic separation method, it is being described in less detail because descriptions have
been published in reports on studies made for the Hawaiian Islands ["'.J], Tennessee river
basin /J., §], and Mekong river basin [i±J. The orographic separation method could not
be used in these three problem areas for the reasons ci t ed below.
In the Hawaiian Islands, relatively isolated peaks or short ridges are rela-
tively ineffective in lifting moist air as required by the orographic model. Observa-
tions indicate that streamlines are diverted horizontally in such terrain.
The Tennessee ~iver basin includes multiple ridges at various angles to mois-
ture inflow directions. Critical inflow directions vary from south-west to south-east.
Moistu r e inflow from any direction in this range can produce heavy rainfalls in some
portion of the basin. Another obstacle to the use of the orographic model here is the
relatively large variability of storm wind direction with height, so simple wind pro-
files, as used effectively for the Sierra Nevada slopes in California["~, are not ap-
propriate.
The orographic model could not be used for the Mekong river basin for several
reasons. In regions near the tropics, precipitation variation with topography is dif-
ferent from that in middle latitudes. Atmospheric moisture is near saturation levels,
and first slopes are important in setting the locations for heavy rains. Also, atmos-
pheric iostability is generally greater. Laminar wind-flow across mountain barriers,
which results in heaviest rainfalls near the highest elevations, is not supported by
observations. Another obstacle is that typhoons, which set the level of PMP for dura-
tions up to three days, show no simple relation between wind speed and rainfall, so
that maximization for wind is difficult.
Modification of non-orographic PMP for orography as used in a study for the

Tennessee ~iver basin above Chattanooga, Tennessee /:4J, is described below. The pro-
cedure as used in generalized estimates of PMP for the Hawaiian Islands [""'.J], Tennessee
~iver basin {""~, Mekong Tiver basin [i(J in south-east Asia, and for thunderstorm rain-
fall in the Columbia river basin in north- western United States [i(jj is described in
Chapter 5~
tSTIMATES FOR OROGRAPHIC REGIONS 81
0 25 50
STATUTE MILES
Figure 3.18 - Typical orographic rain-

fall pattern for south-westerly winds.
Isolines indicate ratios of orographic
to non-orographic rainfall
3.4.2 Tennessee river basin above Chattanooga, Tennessee

--------------------------------------------------
A study J:{J for the Tennessee river basin covered the 55 000 km 2 area above
Chattanooga, Tennessee, and a 21 000 km2 sub-basin in ~he lower portion just above
Chattanooga. Topography of the larger basin varies from the rugged mountains of the
south-eastern portion with peaks above l 500 m to a re latively smooth central valley
extending from south-west to north-east. North-west of the valley lies a series of
parallel ridges extending from south-west to north-east with peaks to about l 000 m.
Chief moisture sources are the Gulf of Mexico about 600 km to the south, and the At-
lantic Ocean about 500 km to the south-east. A typical orographic rainfa ll pattern
for south-westerly winds is shown in Figure 3.18. The values shown are ratios of oro-
graphic to non-orographic precipitation as estimated from a study of several major
storms.
The approach described below is the one used for estimating PMP for these two
basins. Other approaches could have been used with equally valid results.
A major consideration in assessing topographic effects was whether they would

produce a net increase or decrease of average basin PMP as compared to that to be ex-
pected if there were no mountains. Increases, of course, would be related to slopes
exposed to moisture inflow, while decreases would be associated with sheltered, or lee ,
areas, but what would t he net effect be on the basin as a whole?
Mean annual precipitation was used first as basis for comparison. Observed
basin average precipitation indicated a net basin-wide increase of about 10 per cent
above estimates for surrounding non-orographic areas.
February, March and August were selected for estimating topographic effects
on monthly rainfall volume. The larger basin was divided into three zones (Figure
3 . 19): (A) a zone of minimal topograp hic ef fects, (B) an orographic depletion zone,
and (C) an orographic intensification zone. The average precipitation in zone A was
used as a base. The mean precipitation for each of the 3 months indicated a net topo-
graphic depletion for the winter months based on the zone B decrease overcompensating
for the orographic zone C increase.
A simila r comparison based on the mean of seven unusually wet months selected
from the January-April season in six different yea r s showed no appreciable difference
between precipitation in depletion zone B and that in intensification zone C.
Legend
A Control zone
0 25 so B Orographic depletion zon
C Orographic intensification
Statute miles zone
Figure 3.19 - Basin subdivisions for check of

topographic effects on basin-wide precipita-
tion volume
Daily station rainfalls averaged over the Tennessee river basin above and be-
low Chattanooga were used as an auxiliary indicator of net orographic effects. The
area above Chattanooga can be likened topographically to zones B and C, and the area
below, to zone A (Figure 3.19) . Comparison of the means of the series of monthly
maximum daily ave r ages showed a net deficit for the basin above Chattanooga.
Although mean annual precipitation indicated a modest orographic intensifica-

tion, the more extreme precipitation data tended to negate such intensification. The
net effects, if any, are apparently small, and it was assumed that there was no net to-
pographic effect on the volume of precipitation for the basin as a whole.
3.4.2.2 Derivation of PMP

About three dozen major storms scattered throughout the eastern half of the
country were maximized, and generalized charts of PMP were prepar~d for south-eastern
United States . It developed that March storms provided contr?lling PMP values for the
·
basins, a nd a map of 24-hour 25 000 km2 March PMP was drawn (Figure 3.20) . The PMP
2 h d f th. d d
value for the centre of the 21 000 km sub-basin was t e? rea rom is map, an a -
justed upward s l ightly, on the basis of depth-area relations of obse:ved storms, for
the difference in area size. The 24-hour Marc h PMP for the sub-basin was thus deter-
mined t o be 357 mm .
!____ _ _ __ _ _ _ _ _ _ __ _ __ _ _ _ _ _ _ _ _ _ __ '
2
Figure 3. 20 - March 24- hour 25 000 km
PMP (cm)
3.4.2.3 Seasonal variati on
Study of outstanding storms of the region indicated that, for the basin sizes
involved, a March storm would be more likely to produce PMP t ha n would summer tropical
stor ms. Tropical storms, which usually occur with near-maximum dew points, were ad-
justed to the basin location on the basis of decreased rainfall with distance inland of
observed storms. Other precipitation data , such as wettest seven-day periods and months,
rainfall-frequency data, and some unpublished generalized PMP estimates for 50 000 km2,
were used in setting the seasonal variation for the larger basin. The seasonal varia-
tion was first determined for the larger basin, because of previous studies for that
size of area, and applied to the sub-basin as described below. Figure 3.21 shows the
adopted seasonal variation of PMP for the 55 000 km2 basin as a percentage of March PMP .
100
.......................
90 .....
.....
..........
..... .....
..... ...... ... ................ . ...
Q)
~
80 .... ....
- - --
.....
---
al
+>
70
i::
Q)
0
~
Q)
p.., 60 2
Unpublished PMP estimates for 50 000 km
Average of five highest 7- day rains

50 2
Adopted seasonal variation of PMP for 55 000 km
40
Mar Apr May Jun Ju l Aug Sep
Figure 3.21 - Seasonal variation of PMP

for 55 000 km2 as percentage of March PMP
A seasonal variation curve of t he ratio of 24-hour storm ra infall fo r 55 000

2
km to that for 21 000 km 2 , the areas of the two project basins, was based on some two
dozen major storms in the south- eastern pa rt of the country. This ratio curve (Figure
3 . 22) was used to estimate PMP fo r the larger basin from that for the smaller with an
additional reduction of about 2 per cent for the north-eastward displacement ~f the
c~ntre of the large_bas~n . This sma~l adjustment was based on PMP values indicated by
Figure 3.20 . Application of the basin centre adjustmen t and area ratio for March to
the sub-basin PMP (357 mm) yielded a 24-hour March PMP of 284 mm fo r the la rger basin.
90
80
Q)
till
ro
..p
i::::
Q)
(.)
f-j
Q)
P-t 70
Mar Apr May Jun Jun Aug SGp
Figure 3.22 - Depth-area ratios

(55 000/21 000 km2) for 24-hour
rainfall
The seasonal variation curve of Figure 3 . 21 was then applied to the 24-hour
March PMP for the larger basin to obtain 24-hour PMP for April to September as shown
on line 5 of Table 3.4 . These PMP values were then adjusted for area by the recipro-
cal of the ratio curve of Figure 3.22 to yield April to September 24- hou r PMP for the
sub- basin (line 2, Table 3. 4) .
3. 4 . 2 . 4
- - -duration
-Depth- - - - - relations
----
Depth- duration relations, particularly 6/24- and 72/24- hour ratios, of over 100
outstanding storms in the eastern part of the country were examined. Although the
storms occu r red in various months during the March- July period, no seasonal trend was
indicated . The adopted depth - duration curves (Figure 3. 23) show slight differences
for basin size . These curves were used to adjust 24- hour PMP values of Table 3.4 to
6- and 72-hour amounts .
It was stated earlier that there was no net decrease or increase of basin rain-
fall as compared to surrounding areas. This does not mean that there are no topo-
graphic effects . Any examination of a number of storms shows that the distribution is
definitely affected by the topography. In rugged terrain, topographic effects result
in more or less distinct storm rainfall patterns, with appreciable differences between
patterns attributable chiefly to wind direction and storm movement .
160
120
Q)
~
~Q) 80
0
fi
Q)
fl.
40
12 24 36 48 60 72
Duration (hours)
Figure 3.23 - Depth-duration relations

in per cent of 24-hour rainfall
The PMP values of Table 3.4 merely represent overage depths of basin PMP, and
provide limiting rainfall volumes for various possible PMP storm patterns. Examina-
tion of isohyetal patterns for a number of outstanding storms over the project basins,
together with streamflow data, indicated several critical patterns for the larger basin.
Figure 3.24 presents one of these patterns for the 6-hour March PMP.
In order to minimize the work involved in determining pattern configurations

and resulting runoff, any selected pattern is generally considered applicable to all
durations, with only the isohyet values changing. Isohyet values for the pattern of
Figure 3 . 24 were obtained by the relation of Figure 3.25, which applies to the maximum,
or first, 6-hour PMP inc rement. Similar relations were developed for other 6-hour incre-
ments and for 72 hours. These relations were derived in a manne r similar to that de-
scribed in section 2.11.3, with the so-called within-basin, or typical, depth-area
curves, like those of Figure 2.14, patterned after outstanding storms in, or transpos-
able to, the project basins.
Table 3.4 - Probable maximum precipitation (mm) for Tennessee river
basin above Chattanooga, Tennessee
Line Duration Mar. Apr. Moy June July Aug. Sept.

no. (hours)
Sub-basin (21 000 km 2 )
l 6 178 177 174 171 167 167 178
2 24 357 354 349 342 334 334 356
3 72 517 513 506 496 484 484 516
Total basin (55 000 km2)
4 6 128 123 116 107 98 99 114
5 24 284 273 259 239 219 222 253
6 72 426 409 388 358 328 332 379
Isohyet values for the PMP storm pattern of Figure 3.24 are given in Table 3. 5.
The isohyet va lues for the maximum, or first, 6-hour PMP storm pattern of Figure 3.24
were obtained as follows . The total area enclosed by each isohyet was obtained by
planimetering. The area was then used to enter the nomogram of Figure 3.25 on the or-
dinate scale . The corresponding ratio of isohyet value to basin PMP was then obtained
by laying a straight-edge across the nomogram at the proper ordinate value and reading
the ratio below the intersection of the straight-edge and the appropriate basin area
curve . This ratio was then applied to the basin PMP to ob tain the isohyet value .
Isohyet values for other 6-hour PMP increments were obtained in a similar
fashion from similar ratio relations except that the ratios were applied to correspond-
ing 6-hour PMP increments. Thus, for example, the isohyet values for the second 6-hour
PMP increment were determined from a corresponding ratio relation, like that of Figure
3.25, and the second 6-hour PMP increment as indicated by the appropriate depth-dura-
tion curve from Figure 3.23.
The effect of geographic distribution of rainfall on runoff generally de-

creases as basin size decreases. The simple oval- shaped pattern of Figure 3.26 was
considered appropriate for the sub-basin. Isohyet values were determined as describ-
ed above .
85°
Figure 3.24 - Six-hour March PMP storm

pattern (mm) for total basin (55 000 km2)
3.4.2.6 Time distribution of PMP
The procedures just described yielded 6-hour rainfall incremental values or

maps for the 12 periods in the 72-hour PMP storm in any given month in the Marc h-Sept-
ember season. Ranking of first, second,etc., 6- hour increments was based on descending
order of magnitude and not on chronological sequence. Storm experience, which pro-
vides guidelines for reasonable time sequences, generally indicates a strong tendency
for several bursts of rainfall during a 72-hour storm. Within a typical burst, the
largest two or three 6-hour increments usually occur in succession. To maintain PMP
values for all durations, however, any sequence of n 6-hour increments should consist
of the n highest 6-hour values.
100 000
'
50 000 ' '' '' ' '

'' ''
..... '' '
.... ......' ' ' '
..........
.....
....
..... ' ' ' '
..... '
...... .....
.....
......
'
........ ' '
.....
10 000 .... '
..... .... ','
..... ''
''
5 000 '' '
,......._
'' ~
C\J
%
~
'-' 6'o 0-(Q
+>
Cl>
ao
>. VO
,.q _,:::"'.?
0
ta
·ri 1 000
i::
•ri
.i=
+>
·ri
;. 500
a!
Cl>
J..t
..:
100
50
2
25 km - - -- - -
10 '--~_,_~~-1--~---'-~~--'--~---'~~_,_~~.L.-~--'-~~-'--~---I.~~-'-~---'
0,1. 0,8 1,2 1,6 2,0 2/. 2,8

Ratio (isohyet value/basin PMP)
Figure 3.25 - Nomog r am for obtaining

isohyet values for maximum 6- hour
rainfall increment in pattern storms
~
0
Table 3.5 - Isohyet values (mm) for 6-hour March PMP storm pattern of Figure 3.24
f"T'1
(/)
Isohyet A B c D E F G H P1 P2
-I
1-1
3:
)>
-I
72 hours 498 470 439 378 371 333 290 241 688 584 1-1
0
z
1st 6 hours 168 142 135 117 102 86 64 41 241 206 ..,.,
0
2nd 6 hours 79 76 71 69 64 58 53 41 107 89 "

;;:o
0
OJ
)>
OJ
3rd 6 hours 53 53 51 46 43 41 40 38 71 61 r
f"T'1
4th 6 hours 41 41 38 36 33 30 28 25 56 48 ~
x
1-1
3:
2nd day* 99 99 91 61 81 74 69 61 135 114 c
3:
3rd day** 58 58 53 51 46 43 41 36 79 66 "

;;:o
f"T'1
("')
1-t
"-I
1-1
)>
-I
Total area 1-1
0
enclosed by z
isohyet (km2):7 120 l 640 18 370 27 530 39 320 55 880 78 000 107 950 2 2
* For successive 6-hour values use 32, 27, 22 and 19 per cent of 2nd day
** For successive 6-hour values use 29, 26, 23 and 22 per cent of 3rd day
The following sequence was recommended on the basis of the above guidelines.
It does not necessarily provide PMP for all durations but conforms t~observed storm
sequences. First, the four largest 6-hour increments of the 72-hour PMP storm were
grouped in one 24-hour sequence; the middle four, in a second 24-hour sequence; and
the three smallest, in a third 24- hour sequence. Second, the four 6-hour increments
within each of these three 24-hour sequences were arranged as follows: second largest
next to largest, the third larg est adjacent to these, and the fourth largest at either
end . Third, the th r ee 24- hour sequences were arranged with the second largest next
to largest, with the third at either end. Any possible sequence of the three 24-hour
periods was determined acceptable with the exception of that which would place the
smallest 24-hour increment in the middle . (Sample arrangement in Table 2.4).
3.5 Cautionary remarks on estimating PMP in orographic regi ons
The cautionary remarks of section 2.13 concerning adequacy of storm sample,

comparison with record rainfa ll s, consistency of estimates, seasonal variation, and
areal distribution apply also to .orographic regions. As stated in section 1.3.3, the
examples given are not intended for direct application.
3.5.l Basic data deficiencies

-----------------------
Precipitation networks in orographic regions ~re rela~ively sparse compared to
those in non-orographic regions, which are generally more heavily populated: Further-
more, in mountainous areas, most gauges are located in settlements at relatively low
elevations along rivers or in broad valleys . Very few are located on.s~eep.slopes or
at high elevations. To these shortcomings may be ~dded.the usua~ deficienc~es of
gauge measurements , which are likely to be at a maximum in mountainous terrain.
Consequently, precipitation data are not only relatively sparse and sometimes inaccur-
ate but are generally biased and therefore do not represent adequately the effects of
orographic influences on precipitation distribution . This shortcoming affects the
reliability of various relationships, such as precipitation-elevation and depth-area
relations, required for estimating PMP. The situation may be alleviated by referr in g to
adjusted seasonal precipitation maps [j., §7 in determining distribution of storm pre-
cipitation (section 3.1.3). Also, it is sometimes possible to use ra infall runoff rela-
tions to obtain areal estimates of storm rainfall that may be more accurate then indi-
cated by observed precipitation data alone.
3.5 . 2 Orographic separation method

----------------------------
The orographic separation method for estimating PMP (section 3.3) involve~
additional problems besides those just mentioned, since it requires enough upper-air
data to obtain reliable extreme values . Model test requirements for upper-air sound-
ings near the inflow side of the test area and for sufficient concurrent precipitation
data for the test area further limit the applicability of the model.
as· 05 83°
37° + +
84°
+ + 37°
+35
+
as·
+
83° 82
.
Figure 3.26 - Six-hour March PMP storm pattern

(mm) for sub-basin (21 000 km2)
Of the regions where the orographic model has been tested, best results were
obtained for the continuous, high and favourably oriented (with respect to moisture in-
flow) Sierra Nevada in California. The model computes orographic precipitation under
the assumption of laminar air flow. Hence, it is not well suited for regions or
seasons where or when unstable atmospheric conditions predominate. Orographic regions
where major storms occur in the cool seasons ore more likely to meet the required con-
ditions.
Some studies for regions near the tropics indicate that the laminar flow model
is unsuited for estimating PMP. Indirect approaches, such as that used for the Ten-
nessee river basin study (section 3.4. 2), are likelier to yield more reliable estimates
of PMP.
Section 3.3.5 cautioned against over-maximizing and cited some precautions.

To these may be added the use of conservative envelopment of the various factors in-
volved in the procedure whenever this technique is required.
References
l. Myers, V.A., 1962: Airflow on the windward side of a large ridge. Journal of
Geophysical Research, American Geophysical Union, Vol. 67, No. 11, pp. 4267-4291.
2. Nordenson, T.J., 1968: Preparation of co-ordinated precipitation, runoff and

evaporation maps. Reports on WMO/IHD Projects, Report No. 6, World Meteorolo-
gical Organization.
3. Schwarz, F.K. 1 1963: Probable maximum precipitation in the Hawaiian Islands.

Hydrometeorological Report No. 39, U.S. Weather Bureau.
4. Schwarz, F.K. 1 1965: Probable maximum and TVA precipitation over the Tennessee
river basin above Chattanooga. Hydrometeorological Report No. 41, U.S. Weather
Bureau.
5. Schwarz, F.K. and Helfert, N.F., 1965: Probable maximum and TVA precipitation
for Tennessee river basins up to 3 000 square miles in area and durations to 72
hours. Hydrometeorological Report No. 45, U.S. Weather Bureau.
6. Solomon, s. I. I Denouvilliez, J.P. I Chart, E.J. I Woolley, J.A. and Cadou, c./ 1968:
The use of a square grid system for computer estimation of precipitation, tempera-
ture, and runoff. Water Resources Research, American Geophysical Union, Vol. 4,
No. 5, pp. 919-925.
7. U.S. Weather Bureau, 1958: Highest persisting dew points in western United
States. Technical Paper No. 5.
8. U.S. Weather Bureau, 1961: Interim report, ~ probable maximum precipitation in

California. Hydrometeorological Report No. 36.
9. U.S. Weather Bureau, 1961: Generalized estimates of probable maximum precipita-

tion and rainfall-frequency data for Puerto Rico and Virgin Islands. Technical
Paper No. 42.
10. U.S. Weather Bureau, 1966: Probable maximum precipitation, north - west States .
Hydrometeorological Report No. 43.
11. U.S. Weather Bureau, 1970: Probable maximum precipitation, Mekong river basin.
12. Weaver, R.L., 1966: California storms as viewed by Sacramento radar. Monthly
Weather Review, U.S. Weather Bureau, Vol. 94, No. 7, pp. 416-473.
C H A P T E R 4
STATISTICAL ESTIMATES
4.1 Use of statistical procedure
Statistical pr ocedures for estimating PMP may be used wherever sufficient pre-
cipitation data are available, and a re parti c ularly useful fo r making quick estimates
or where other meteorological data, such as dew point and wind r ecords, are lacking.
The procedure described below is not the only one, but it has received the widest
acceptance. It is used mostly for making quick estimates for watersheds of no more
than about l 000 km2, but has been used for much larger areas . Its convenience lies
in that it requires considerably less time to apply than does the meteorological, or
traditional, approach and that one does not have to be a meteorologist to use it . A
major shortcoming is that it yields only point values of PMP and thus requires area-
reduction curves for adjusting the point values to various si zes of area .
4.2 Development of procedure
4.2.l
The procedure as developed /]J and later modified ~4J by Hershfield is based
on the general frequency equation ~fl:
= xn + KS n I
(4 . 1)
where Xt is the rainfall for return period t; Xn and Sn are respectively the mean and
standard deviation of a series of n annual maxima; and K is a common statistical vari-
able which varies with the different frequency distributions fitting extreme-value
hydrologic data.
If the maximum observed rainfall, Xm, is substituted for Xt, and Km for K,
Km is then the number of standard deviations to be added to ~n to obtain Xm, or
(4.2)
Records of 24-hour rainfall for some 2 600 stations , of which about 90 per
cent were in the United States, were used in the initial determination of an envelop-
ing value of Km• Values of Xn and Sn were computed by conventional procedures, but
the maximum recorded rainfall at each station was omitted from the computat i ons. The
greatest value of Km computed from the data fo r all stations was 15. It was first
thought that Km was independent of rainfall magnitude, but it was later found to vary
inve r sely with rainfall: the value of 15 is too high for areas of generally heavy
rainfall and too low for a r id areas . Values of Km for other rainfall durations we r e
later dete r mined, and its variation with Xn fo r du r ations of 5 minutes, 1, 6 and 24
hours is shown in Figure 4.1, which indicates a maximum Km of 20.
9o ESTIMATION OF PROBABLE MAXIMUM PRECIPITATION
24 h~ urs
15
J;j
15
-- -
\
\
\ 10 20 30
\ Mean annual maximum rainfall mm
\ 6'1
'\ov,..
' \ >$'
10
'\
' "\
5
''
100 200 300 400 500 600
Mean annual maximum rainfal l (mm)
Figure 4.1 - Km as a function of rainfall

duration and mean of annual series (after
Hersh field { 4 ] )
4.2.2
Extreme rainfall amounts of rare magnitude or occurrence, say, with return

periods of 500 or more years, are often found to have occurred at some time during a
much shorter period of record, say, 30 years. Such a rare event, called an outlier,
may have an appreciable effect on the mean (Xn) and standard deviation (Sn) of the
annual series. The magnitude of the effect is less for long records than for short,
and it varies with the rarity of the event, or outlier. This has been studied by
Hershfield J:'J.l using hypothetical series of varying length, and Figures 4.2 and 4.3
show the_adjustments to be made to Xn and Sn to compensate for outliers. In these
figures Xn-m and Sn-m refer respectively to the mean and standard deviation of the
annual series computed after excluding the maximum item in the series. It should be
noted that these relationships consider only the effect of the maximum observed event.
No consideration was given to other anomalous-appearing observations.
STATISTICAL ESTIMATES 97
110
100
>-!
0
+'
0
ro
..... 90
+'
>=:
<!)
s
+'
Ul
......,
;j
rrj
ro
>=:
><
0,8 0,9 1,0

Xn-m IX n
Figure 4.2 - Adjustment of mean of annual

series for maximum observed rainfall
98 ESTIMATION OF PROBABLE MAX IMUM PRECIPITATION
120
0"'
Q,
~
100 ,~
~b
,.---.
,ec,O
+'
~
~
~
Q)
0
80 Q,~
1-1 'v
Q)
p,
1-1
0
+'
0
al
'-i 60
+'
~
Q)
El
+'
rJJ
::I
'r:>
'Cl
111 L. 0
uf
20'--~-'-~----'-~~J__~---L~~L-~-L~----1~___J
0,2 0,4 0,6 0,8 1,0

S IS
n-m n
Figure 4 . 3 - Adjustment of standard deviation

of annual series for maximum observed rainfall
4.2.3 ~~i~~!~:~!-~i_!~-~~~-~~-i~:-~~~e~=-~~~=
The mean (Xn) and standard deviation (Sn) of the annual series tend to in-
crease with length of record, because the frequency distribution of rainfall extremes
is skewed to the right so that there is a greater chance of getting o large than o
small extreme as length of record increases. Figure 4.4 shows the adjustments to be
mode to Xn and Sn for length of record. There were relatively few precipitation re-
cords longer than 50 years available for evaluating the effect of sample size, but the
few longer records available indicated adjustment only slightly different from that for
the 50-yeor records.
130
125
,........
+'
i:1
Q)
() 120
~
Q)
p,
~
0
+'
()
Ill
'+-I 115
+'
i:1
Q)
a
+'
f/l
...,::l
'd
...: 110
105
100
10 20 30 L. 0 50
Length of record (years)
Figure 4.4 - Adjustment of mean and standard

deviation of annual series for length of re-
cord (after Hershfield [ 3 })
4.2.4
Precipitation data are usually given for fixed time intervals, e.g., 8 a.m.
to 8 a.m. (daily), 0600-1200 (six-hourly), 0300-0400 (hourly). Such data rarely yield
the true mQximum rainfall amounts for the indicated durations. For example, the
annual maximum observational day amount is very likely to be appreciably less than the
annual maximum 24-hour amount determined from intervals of l 440 consecutive minutes
100 ESTIMATION OF PROBABLE MAXIMUM PRECI PITATION
unrestricted by any particular observation time. Similarly, maxima from f ixed 6-

hourly and hourly intervals tend to be less t han maxima obtained from_ 360 and 60 con-
secutive one-minute intervals, respectively, unrestricted by fixed beginning or ending
ti mes .
Studies of thousands of statio n-years of rainfall data indicate that multi ply-
ing the results of a frequency analysis of annual maximum rain fall amounts for a single
fixed time interval of any duration from l to 24 hours by 1.13 will yield values close-
ly approximating those to be obtained from an analysis based on true maxima. Hence,
the PMP values yielded by the statistical procedure should be multiplied by 1.13 if
data for single fixed time intervals are used in compiling the annual series. Lesser
adjustments are required when maximum observed amounts for various durations are deter-
mined from two or more fixed time in tervals (Figure 4.5). Thus, for example, maximum
6- and 24-hour amounts determined from 6 and 24 consecutive 1-hour rain fall increments
require adjustment by factors of only 1.02 and 1.01, respective ly .
114
112
J.l
0
+>
0
110
.....al
+>
i:::
Q)
106
a
+>
m
~
.,..., 106
'd
4l
104
1 02
100
0 4 6 12 16 20 24
Number of observat i onal units
Figure 4.5 - Adjustment of fixed i nterval

precipitation amounts for number of ob-
servational units within th e interval (after
Weiss [ 8 ])
4.2.5 Area-reduction curves
The procedure described here was developed for point rainfall data. Hence,
its use requires some method for reducing the point values it yields to some required
areal rain fall ave.rages. There are many variations of depth-area relationships £2'0,
since they represent the depth-area-duration (DAD) characteristics of different types
of storms. The curves of Figure 4.6 [7..J are based on average values obtained from DAD
analyses of major general-type storms and do not show as much decrease with increasing
area as would curves based on localized cloudbursts. They do not extend beyond l 000
km2 because extrapolation of point rainfall values becomes more unreliable as size of
area increases. Necessity, however, has led to relations hi ps ~6_] relating point values
to areas in exce s s of 100 000 km2. Point values are ofte n assumed to be applicable to
areas up to 25 km2 without reduction.
r--i
r--i
al
'H
i:::
•rl 100
"'
F-1
C\J
!!
l1"'I
C\J
F-1
90
0
~
+'
i:::
·rl
0
p.,
80
·rl
!
><
al
a
Q)
r--i
.n
.n "'
0
F-1
70
p.,
'H
0
Q)
~
al
+' 60
i:::
Q)
200 L.00 600 800 1000
0
F-1 Area ( krrl)
Q)
Po<
Fig ur e 4.6 - De pth-area, or area-red ucti on,

c urves
102 ESTIMATION OF PROBABLE''MAXIMUM PRECIPITATION
4.2.6
Only daily measu rements of precipitation are available for many regions.
Various types of depth-duration relationships have been developed to show rainfall dis-
tribution within storms. Such relationships vary a great deal depending on storm type.
For example, orographic rainfall will show a much more gradual accumulation of rainfall
with time than will thunderstorm rainfall.
The maximum depth-duration relation of Figure 4.7 is based on rainfall amounts

in heavy storms averaged over areas ranging up to 1 000 km2 in Illinois, U.S.A. /"~.
This relationship arranges the rainfall increments for various time intervals in de--
creasing order of magnitude and not in chronological order. In other words, the curve
shows the greatest 3-hour amount in the first 3 hours, the second greatest 3-hour
amount in the second 3-hour period, etc. This arrangement is not intended to repre-
sent the order in which the rainfall increments occurred, nor does it do so except
perhaps occidentally for an occasional storm. Studies of chronological distribution
100
--- ----
BO _I_ - - -·- --
I
- .. ~ ... ---
I
I
I
j
~ -I
~0 60
.£: ·--1- --
I
.q-
C\I I
....0 II
•:t 40
I
II
...a I
•...
0
I
•
p.,
20
I
I
I
II
0 I
0
i
6 hl 18 24
Dur at ion (hours )
Figure 4.7 - Maximum depth-duration c urve

of rainfall within storms indicate no consistent pattern, with maximum intensities

likely to occur during any period of the storm.
The curve of Figure 4.7 is representative of convective storms. Because of

the variation of such relationships with storm type and geography, they should be de-
veloped f r om data for the same regi ons for which the PMP estimates are required. Figure
4.7 is presented here merely as an example and is not int ended for general application.
Figure 4.7, or similar re lationships, s hould be used only when rainfall data for dura-
tions shorter than 24 hours are unavailable.
4.3 Application of procedure
It i s assumed that a PMP estimate is required for a watershed of 500 km2 .

Table 4.1 lists the annual ma xi mum 1-, 6- , and 24-hour ra infall amounts (annual series)
compiled from an actual 25- year rec ord of hourly precipitation data for a station in
the problem watershed . The hou rly values are thus fo r the clock hou r, e.g., 0900-1000,
and the 6- and 24-hour amounts consist of the greatest sums of 6 and 24 consecutive
clock-hour rainfall increments, respectively. Xn-m and Sn - m ar e the mean and standard
deviation, respectively, of the annual series computed after excluding the maximum
rainfall amount in each series . Xn and Sn are for the series including all items.
Means and standard deviations are computed by conventional methods and should be com-
pared with those of nea rby stations for consistency. If inconsistent, another station
should be used for estimating PMP.
After the two means and standard deviations for each series and their respect-
ive ratios have been obtained as indicated in the table, estimation of PMP proceeds as
follows.
l. Adjust Xn and Sn for maximum observed rainfall by means of Figures 4.2
and 4.3, respectively, and for record length by means of Figure 4.4.
2. From Figure 4.1 obtain values of Km corresponding to adjusted values of

in for the various durations.
3. Compute point values of PMP, or Xm, as indicated by equation (4 . 2).
4. If basic rainfall data are for fixed time intervals, adjust upward by
applying the factor 1.13 for fixed observational periods or the factors
1.13, 1.02 and 1.01 to 1-, 6-, and 24-hour amounts, respectively, com-
piled from hourly data (section 4.2.4).
5. Use Figure 4.6 to reduce point values of PMP to the proper areal value
for the size of the basin. (Note: if only 24-hour rainfall amounts
are available, a maximum depth-duration curve, like that of Figure 4.7,
can be used to estimate PMP for the shorter durations. The 34 and 84
per cent adjustments for the 1- and 6-hour amounts, respectively, would
yield values of 155 and 382 mm, which are considerably higher than the
103 and 331 mm based on the actual data. Hence, Figure 4 . 7 does not
very well represent the depth-duration characteristics of PMP indicated
by the short-duration data for the problem basin).
Table 4.1 - Computation of probable maximum precipitation (PMP)
Annual maximum precipitation, mm (annual series)
Duration (hours)
Year
1 6 24
1941 30 62 62
1942 19 38 60
1943 15 39 57
1944 33 108 112
1945 23 49 67
1946 19 39 72
1947 32 50 62
1948 24 30 61
1949 30 39 57
1950 24 38 69
1951 28 58 72
1952 15 41 61
1953 20 47 62
1954 26 68 82
1955 42 124 306
1956 18 43 47
1957 23 39 43
1958 25 48 78
1959 28 80 113
1960 25 89 134
196i 28 33 51
1962 46 72 72
1963 20 47 62
1964 14 34 53
1965 15 40 55
xn- m 24.0
n = 25 --= 0.96 51. 3 - 0. 95 69.3 - 0.88
=---=
xn
24.9 54.2 - 78.8 -
Sn-m. 6.8
-= 0.86 19.5
--= 0.81 21.8 - 0.42
~· 7.9 24.0 51.9 -
(Continued)
Table 4.1 - Computation of probable maximum precipitation (PMP)
(Continued)
Adjustment of means (Xn) for maximum observed amount and record length:
l hour 6 hours 24 hours
From Figure 4.2: 0.99 0.98 0.91

From Figure 4.4: l.01 l.01 1.01
Adjusted Xn: 24 . 9 53.6 72.4
Adjustment of standard deviations for maximum obse rved amount and record length:
From Figure 4.3: 0 .98 0.93 0.49

From Figure 4.4 : 1.05 1.05 l.05
Adjusted Sn: 8.1 23.4 26.7
Km (Figure 4.1): 14 14 16
Unadjusted point values of PMP from equation (4.2):
l hour: PMP = 24.9 + 14(8.l) = 138 mm

6 hours: PMP = 53.6 + 14(23.4) = 381 mm
24 hours: PMP = 72.4 + 16(26.7) = 500 mm
Adjustment of PMP based on hourly data to true maximum values (see section 4.2 . 4):
1-hour PMP = 1.13(138) = 156 mm

6-hour PMP = 1.02(381) = 389 mm
24-hour PMP = 1.01(500) = 505 mm
(Note: If annual series data had been compiled from fixed observational time
intervals instead of hourly data, the adjustment factor for all
durations would have been 1.13.)
Adjustment of point PMP to 500 km2 (Figure 4.6):
l hour 6 hours 24 hours
Adjustment factors: 0.66 0.85 0.90

PMP for 500 km2 (mm): 103 331 445
4.4 Generalized estimates
Where precipitation networks are considered adequate, generalized PMP esti-

mat es of reasonable reliability may be made with relative ease. The adjusted mean
(Xn) and standard deviation (Sn) are determined (section 4.3) for each station, and the
coefficient of variation (Cy), i.e., the standard deviation divided by the mean, is
then computed. Values of Cy, which is considered a more stable statistic than Sn,
and Xn are plotted on a map, and two sets of isolines are drawn. Values of PMP for
any point on the map may then be obtained by estimating Xn and Cv from their respec-
tive isolines and using the following relation:
(4.3)
By computing PMP for a fine grid of points, a map showing PMP values directly may then
be constructed. Values of PMP, or Xm, obtained from equation (4.3) are subject to the
same adjustments described in section 4.3.
4.5 Cautionary remarks
The curves of Figure 4.1 are based on observed data. Consequently, they
imply that PMP has already occurred at those stations providing controlling values of
Km. As a matter of fact, there are at least three measurements of rainfal l made in
other than official gauges that exceed the PMP values to be obtained from the use of
Figure 4.1. The reason given for excluding these measurements in developing the pro-
cedure was that the accuracy of the measurements was somewhat questionable and that
there were no precipitation records for the locations of occ urrence from which to com-
pute Xn and Sn • Estimates of these parameters for nearby stations indicated that a
Km value of 25 would have yielded PMP values enveloping any measurements ever made in
the United States . Computations of Km for Canada [6:f indicated a maximum value of
30 associated with a mean annual maximum 24-hour rainfall amount of 15 mm.
Further studies are needed to determine more reliable values of Km• It

appears likely, for example, that Km may be related to other factors besides rainfall
duration and mean of the annual series. In using the procedure, it should be kept in
mind that the indicated Km values may be too high for some regions and too low in
others. In general, the procedure tends to yield values of PMP lower than those to
be obtained -from meteorological, or traditional, procedures.
In selecting a station for making a PMP estimate for a particular drainage

basin, it is important that its precipitation record is reasonably representative.
Comparisons of Xn and Sn or Cv with nearby stations are recommended. Odd values in
the basic data should be examined and discarded if found spurious, or the record for
another station should be used. Length of record should be considered also. A long
record will yield generally more reliable PMP estimates than will a short record of
comparable quality. Wherever possible, records of no less than 20 years should be used
and records of less than 10 years should not be used at all.
The use of general area-reduction and depth-duration curves, like those of

Figures 4.6 and 4.7, respectively, introduce additional sources of error in the PMP
estimates. Such curves should be developed for the regions for which the estimates
are to be made since they vary with rainfall type and geography.
References
1. Chow, W. T., 1961: A general formula for hydrologic frequency analysis. Trans-
actions American Geophysical Union, Vol. 32, pp. 231-237.
2. Court, A., 1961: Area-depth rainfall formulas. Journal Geophysical Research.

American Geophysical Union, Vol. 66, pp. 1823-1832.
3. Hershfield, D. M., 1961: Estimating the probable maximum precipitation. Proceed-

ings American Society of Civil Engineers, Journal Hydraulics Division, Vol. 87,
pp. 99-106.
4. Hershfield, D. M., 1965: Method for estimating probable maximum precipitation.

Journal American Waterworks Association, Vol. 57, pp. 965-972.
5. Huff, F. A., 1967: Time distribution of rainfall in heavy storms. Water Re-
sources Research, American Geophysical Union, Vol. 3, pp. 1007-1019.
6. McKay, G. A., 1965: Statistical estimates of precipitation extremes for the

prairie provinces. Canada Department of Agriculture, PFRA Engineering Branch.
7. U.S. Weather Bureau, 1960: Generalized estimates of probable maximum precipita-
tion for the United States west of the 105th meridian, pp. 52-54.
8. Weiss, L. L., 1964: Ratio of true to fixed-interval maximum rainfall. Proceed-

ings American Society of Civil Engineers, Journal Hydraulics Division, Vol. 90,
pp. 77-82.
CHAP T£ R 5
GENERALIZED ESTIMATES
5. 1 Generalized charts
The methods of e~t imating PMP discussed in Chapters 2 and 3 may be us ed either
for individual basins or for larg e regions enc ompassing numerous basins of variou s
sizes. In the latter case, th e esti mat es are re f erred to as generalized e stimates,
and are usually displayed as isohye tal maps which depict the regiona l variation of PMP
for some specified duration and basin size. Thes e maps are commonly known as genera l -
i zed charts of PMP.
The chief advantages of generalized PMP charts a r e: (1) they are a ready
source of PMP estimates for any basin in a r egion, and (2) they a r e very us e ful in
maintaining consistency between estimates made for individual basins within a region.
Within any particular reg ion, variations in topograp hy tend to increase as

basin si ze increases, and preparation of generali ze d e stimates becomes mo re complicat-
ed, espec ially in orographic regions . Because of the difficulties, generalized esti-
mates hav e been generally limited to areas under 10 000 km2, but some dev e lopmental
work has been done on such e stimate s for areas up to about 50 000 km2
5.1.l . l Scale
The choice of a suitable map base for developing and depicting generalized
estimates of PMP depends c hiefl y on the size of the region for which t he estimates are
to be ma de, the topography, and on the degr ee of detail to be shown on the final maps .
Base maps with a scale of about 1:2 500 000 may be adequate fo r man y non-orographic,
i .e. not extremely mount ainou s, regions. Regions of rugged orography require a larg er
scale, usually no less than 1:1 000 000, whil e a smaller scale, say, 1:5 000 000, might
be adequate for flat terr ain. Whatever the scale, the base maps should sh ow the top o-
graphy of the region. The final maps used for displaying the estimates may be reduced
conside rab ly, of course, but not so much as to make it difficult for t he user to locat e
a basin f or which an e stimat e is required. For this re ason, t he final maps should
show the scale, a latitude- longitude grid, boundaries of states, provinces, districts
and countries.
Once a proper bas e map is s e l ect ed, the next step is to construct a grid on
the map. The grid is usually constructed to conform with the latitude-longitude grid
of the map . The points formed by the inters ections of the grid lines (which actually
do not have to be drawn) indicate the locations to which the maximized storms ar e
transposed and the maximum values plotted . Severa l base maps are sometimes required,
the number depending on th e PMP values to be displayed. For example, one map may be
us ed for developing and di s playing 6-hour PMP over 100 km2, another, for th e 24-hour
PMP over l 000 km2, etc. Regardless of the number of maps r equired, the use of the
same grid on all maps is advisable as it will minimize the work involved in storm
transposition.
The fineness or coarseness of the grid depends on the topography. In very

flat regions, a gri d of 10 latitude degrees by 10 longitude degrees may be adequate .
In mountaino us regions, a 1-degree grid may be too coarse . It is not necessary to
have a uniform grid over an ent ire region. If a region includes both flat and moun-
tainous areas, a coarse grid may be used over the flat area and a fine one over the
mountainous sections.
5.1.2
In the preparation of a series of generalized PMP charts for a reg ion, it is

important that consistency of estimates be maintained within and between the various
charts. It is unrealistic to expect variation in PMP between different durations and
sizes of area to be irregular and erratic, and smoothing of computed PMP values i~
justified . Smoothing is in fact mandatory if consistency is to be achieved. The
smoothing techniques used are similar to those described in section 2.8.
In depth-duration smoothing, maximum adjusted rainfall amounts for various

du rat ions and specified size of area for each maximized and transpos ed storm applic-
able to a particular grid point or location are plotted on a depth-duration diagram.
Figure 2. 9 is an example of such a diagram for 2 000 km2 values at one grid point. The
data plotted are the largest maximized rainfall values for each duration, and a smooth
curve is drawn to envelop these values.
5.1.2.2
-Depth-area
- - --- - smoothing
----
Smoothing and envelopment across area sizes is similar to depth-duration
smoothing. Here, maximum adjusted rainfall values for varius sizes of area and a
specified duration for each maximized and transposed storm applicable to a particular
grid point or location are usually plotted on semilog paper, with size of area being
plotted on the log scale . Figure 2.10 shows such a plotting for 24-hour PMP. The
data plotted at 2 000 km 2 are the same data used in Figure 2.9.
Depth-area and depth- duration smoothing is sometimes performed in one opera-

tion. This is normally done by plotting the data for various durations and sizes of
area on one chart like t hat of Figure 2.11, with each plotted point being labelled with
the appropriate storm identification and duration. Smooth isopleths are then drawn.
The combined smoothing procedure is sometimes confusing because of the rela-

tively large amount of data plotted for each duration and size of area . The procedure
GENERALIZED ESTIMATlS 111
is simplified by first subjecting the data to separate depth-durati on and depth-area

smoothing as described in sections 5.1.2.l and 5.1.2.2. The values plotted on the
combination chart are then taken from the enveloping depth-duration and depth-area
curves . There is then only one value for each duration and size of area, as shown in
Figure 2.11.
5.1 . 3 Regional smoothing

------------------
Isohyets of PMP are drawn to the smoothed storm rainfall values plotted at
the grid points on the map. Limits of transposition of storms will usually result in
discontinuities between some adjacent grid points. Regional smoothing must therefore
take into account the effect of an extreme storm beyond the limits of its area of trans-
posability . In drawing smooth isohyets, meteorological factors, such as moisture
source, storm tracks, moisture barriers, etc . , need to be considered. Some plotted
values may be undercut while others may be over-enveloped. This is done when data
appear inconsistent with nearby values, and to draw for them would result in unwarrant-
ed bulges or dips in otherwise smooth isohyets. If there are geographic factors, such
as an extended range of high hills in a plains region, to support suspected incon-
sistent data, isohyets should, of course, be drawn to the data. If data at individual
grid points have been smoothed properly (section.s5.l.2.l, 5.1.2.2), little over-
envelopment or undercutting is required. Envelopment and undercutting are more common-
ly done in orographic regions.
Drawing of isohyets between grid points is often facilitated by supplementary

considerations . These considerations apply only to isohyetal gradients and patterns,
and have little or no effect on magnitude of PMP values plotted at grid points. In
other words, they provide guidance in spacing and shaping of isohyets between grid
points while giving greatest weight to plotted values.
Guidance is provided by various types of climatological data. For example,

a chart of maximum observed 24- hour point rainfall values from long observational re-
cords should show some resemblance to a generalized chart of 24-hour PMP for any size
of area up to about 1 000 km2. Rainfall-frequency charts may also be used for guid-
ance, although they are not so reliable an indicator of regional variation of PMP since
frequency is involved r ather than ma gnitude alone . Similar regional patterns may be
found also between charts of maximum observed point rainfalls for relatively long
durations, say three consecutive days, and generalized PMP charts for large areas, say
10 000 to 50 000 km2
Regional similarity of generalized PMP and precipitation frequency patterns

does not prevail in those regions where one type of storm produces a large number of
heavy ra infalls, but a different type provides outstanding amounts. An example of
this lack of similarity is found on the island of Hawaii. There, frequent heavy
showers associated with north - east trade winds produce high rainfall-frequency values,
while extreme rainfalls invariably occur with the breakdown of these trade winds, and
generally with winds from a much different direction . This climatic feature is re-
flected in differences between generalized PMP and rainfall-frequency patterns (Figure
5.1).
.......
.......
!\)
F- ,~,.. ·~
· ,.. ~ -
r---i-~;··~
~·;'k:·~\;c
- ------
_J --T· ~ @;-._ I
~
1- -- - -
'-.'-:: ,,, ____ I . --:::
.,~~
r. - - 1v-~~~
; J/ A~ .c~~ : , I
- i-
!'. ;;•. I I
...,
~ /},~~~~~~- ~
~~;;) ~/,£(}::~~\Y:~''~~:-~
1 / ,,
H
3'.
,; J? -~ ,\ \\~~ ~
)>
-I
'~'~(($( ~- ~~)\\11~~
H
0
z
!
.!\~\\\~\\lI:l(,J.·w
·~
).-,~1, ~~ c;, :_,,,
...,
0
'"C
:;:o
0
w
l1 1 '!1 :·, // ./') ' '\~1
1' · 7717 ~
0
~$1 · ~Cef!Jf£'
~0'/~ · : - ~
- ~\~1~,
CD
. "" 1 r. 1 , 1jf1.11;;; r...,
, ~
I ·.,:\ · .,.
.:: H~i11 \ t~~ j;;/~: ,_ _._,/'~
1 \ 1 1f11/1J/;,WP
1
x
H
. , /' <(,·{V1/j/
II: ~ !:1~\"\..AJ//1 /f~"
3'.
i ·;
cr-r~~ -- ,,- V'< c
3'.
i ;,'/.1.! l\ t._
, -..___///l/'r'/-ft1--
\~/! 1
'>rf/).i,;~/f i
~,,P
1
'"C
I- - I -- - • - - ••
...,
:;:o
~- ),~
1
1 ·\ //fD{j' I I ••
('")
H
1 I lfiY- I I '"C
-._._ •o~. !.
1 ,'>,.._ ( l ' ' I H
;; I ) I I I );!
I. !
·-./''.:\
-I I ' .,,. ". -I I- '«->- r
[/
( :I l~· -- ------ -- · -~
I ·- _ ___
-I
H
I
1·
i : If'
• '1 0
z
' I I
- ·- .'." . '!.... . .... . -.:":~'---~ · ... . ~ ~:,: ·.: ~-:-· ~
.. J
Figure 5.1 - Contrast in PMP and 25-year rain-
fall patterns, both for 24 hours at a point,
Island of Hawaii. Particularly note differences
on north-west coast
GENERALIZED ESTIMATES 113
In orographic regions, relations of intense storm rainfall to elevation,

slope, orientation, and other topographic factors may be developed and then used to
estimate extreme storm rainfall values. These estimated values provide a distribution
pattern that can be used to establish PMP distribution between PMP values at grid
points.
In order to maintain consistency between maps when several are to be drawn for
various durations and sizes of area, it is recommended that preliminary isohyets be
drawn first . Two successiv e maps in a series are then superimposed on a light table,
and final isohyets are then drawn so as to form consistent patterns for both maps. For
example, the map of 6-hour PMP for l 000 km2 might be superimposed on that for 1-hour
PMP for the same size of area . The 6-hour PMP isohyets should, of course, indicate
higher values at every point on the map. Also, there is usually no reason for an
isohyet on one map to show a dip, or depression, while the isohyet at the corresponding
location on another map of about the same duration and size of area in the series shows
a bulge. Of course, as differences in duration and size of area increase, there may
be gradual changes in patterns so that bulges may eventually become dips or vice ve r sa.
Maps for different sizes of area should be compared and fitted to each other
in the same manner. For example, isohyets on a map of 24-hour PMP for l 000 km2
should everywhere indicate greater depths than those for 24-hour PMP over 10 000 km2.
If maps for various months are required, as well as the all - season envelope,
seasonal smoothing is necessary. Seasonal variation was discussed in section 2.10.
5.1.4 General remarks
Much work is involved in the preparation of a series of generalized PMP charts

for different durations, area sizes, and months. The usual practice is to prepare as
few such charts as absolutely required and to provide depth-duration, depth-area, and
seasonal variation curves to adjust the chart PMP index values as required. Often,
especially for small basin sizes, a single index chart like Figure 3 . 16 is constructed
for a particular duration, area size, and month. Relations similar to those of Figure
3.17 are then developed for making adjustments for other durations, basin sizes and
months.
In one study ~Q], index charts were constructed for 1-, 6-, and 24-hour point
PMP, and depth - duvation diagrams (Figure 5.2) and area-reduction curves (Figure 4.6)
were provided for obtaining PMP values for other durations and area sizes. The depth -
duration diagrams (Figure 5.2) were based on maximized rainfall values from major
storms . A straight-edge placed on either diagram so that it intersects the first and
last verticals at the PMP values indicated on the maps for the corresponding durations
will yield the PMP value for any intermediate duration by its intersection with the
vertical for that duration. Thus, for example, if 1- and 6-hour PMP values were 250
and 400 mm, respectively, a straight-edge set at those values on the corresponding
verticals of the diagram on the left side of Figure 5.2 would show a 2-hour PMP value
of 300 mm.
180
'--
- 80 16 0
- -
70 140
- -
,-.., ,-..,
a0 60 ~120 120
i:: i::
0
·ri
- 0
·ri
+' +'
ell
+'
·ri
50 50 ~100
·ri
100
Pt Pt
•ri
0
._ - •ri
0 - 1--
Q) Q)
~ ~
Pt
L.O 40 Pt 80 80
~ §
a
•ri - a
·ri - '--
I>< I><
ell ell
a a
Q)
30 30 Q)
60 60
.-t .-t
..c .__ ..c ,__
al ell
..c ..c
0 0
~ ~
p.. 20 20 111 40 40
- - -
10 10 20 20
- - -
0 0 0 0
2 3 4 5 6 6 8 10 12 16 2 0 24
Duration (hours) Durati on (hours)
Figure 5. 2 - Depth-duration interpolation

diagrams
In anot he r study /"?], charts of generalized PMP estimates for 24 hours and
500 km2 were constructed for each month and for the all-season envelope (Figure 5.3).
The region covered by these estimates was so large as to involve several different
storm r'gimes. The region was therefore divided into zon~s, and depth-area-duration
relations like that of Figure 5.4 were developed for each zone for every month and the
all -seas on envelope.
Other examples are presented in the discussion of specific ge nerali zed PMP
studies to be found later in this chapter.
5.2 Estimates for non-orographic regions
Th e basic procedure used for making generalized PMP estimates for non-oro-
graphic regions is essentially the same as that described in Chapter 2 for individual
basins, which involves storm maximization and transposition. Hence, only the pro-
cedural modifications required to generalize the estimates are discussed here.
5.2.2 Moisture maximization
The maximum atmospheric moisture available for storm maximization throughout

a region is an important requirement for the development of generalized charts of PMP.
For reasons given in section 2.2, maximum persisting 12-hour l 000 mb dew points are
used as indices of the maximum amount of atmospheric water vapour available for maxi -
mi zing storms . Generalized charts of these dew points (Figure 2.4) are therefore re-
quired for making the various adjustments involved in developing generalized PMP esti-
mates.
5.2.3 Storm transposition

-------------------
Storm transposition (section 2.5) plays an important role in the preparation
of generalized PMP estimates . In any large region there are many areas that have not
experienced or recorded outstanding storms of the magnitude observed in adjacent areas
or elsewhere in the region, and transposable storms are adjusted to conditions in these
deficient areas to supplement the inadequate record of major storms.
In estimating PMP fo r a specific basin, major storms are examined to determine

if they are transposable to the basin . The storms are then adjusted as required by
the geographic features of that particular basin. In the preparation of generalized
PMP charts, the boundaries, or limits, of the area of transposability (Figure 2.5) of
each major storm are delineated. Each storm is then transposed within its area of
transposability to locations indicated by grid points on a suitable base map (sections
5.1.1 . 2) or to the boundaries of the area, or both . Giid points have the advantage
of allowing ready comparisons between rainfall values f rom different storms.
Transpo sition of a storm from place of occurrence to another location involves

adjustments f or differences in geographic features of the two locations (section 2.6) .
The need for elevation adjustment is minimized if the transposition limits are so de-
lineated that differences in elevation greater than 700 m within the area of transpos-
ability are avoided. When this is done, the elevation adjustment discussed in section
2.6.2 is generally omitted.
ESTIMATION OF PROBABLE MAXIMUM PRECIPITATION
116
M,-..
:) e
0 u
.J: .._,
..J.N c
0
•.-i
4- +'
0 c
+'
II> •.-i
0. 0.
0 •.-i
...... u
II> Q)
> M
c 0.
Q)
c :)
e
o e
Ill ·.-i
c )(
II> c
111 e
...... Q)
...........
<( ..c
c
I ..C
0
C'> M
• 0.
'°N
Cl> e
M~
:)
e>O
0
•.-!
u.. LC>
5.2.4
The preparation of generalized PMP estimates for non-orographic regions is

summarized as follows .
~- Construct an adequate grid system on a suitable base map, or maps

(section 5.1.1).
Step 2. Determine areas of transposability of majo r observed storms i n th e

region of interest and surrounding areas (section 5.2 . 3).
Step 3 . Maximize depth-area-duration values of the selected major storms in

place and transposed to grid points in their areas of transposability (sections 2. 3 to
2.6). It is rarely necessary to transpose all storms to all grid points since adjust-
ment of a few storms generally indicates which are likely to provide controlling (maxi -
mum) values at a particular grid point or set of grid points.
Step 4 . Data at each gr id point should be checked for du r ational, areal a nd

seasonal consistency, and smoothed (sections 5.1 . 2, 5.1 . 3).
Step 5 . Draw preliminary isohyets to the values at each grid point . In

drawing the isohyets, data at a few points may be undercut or over- enveloped if the
data appear inconsistent with adjacent values and cause unwa r rant ed bulges or dips in
the otherwise smooth isohyets. Use whatever supplementary aids a re available for
spacing and shaping isohyets between grid points and maintain consistency between maps
(section 5.1.3). Final isohyets should be smooth, with no unjustifiable dips or
bulges.
Step 6. Develop whatever relationships are required to adapt mop values to

other du r ations, basin sizes, and months (section 5. 1). One generalized chart of PMP
for a specific size of area and duration is usually used as an index. PMP for other
sizes of area and durations are then obtained from DAD relations, expressed as percent-
ages of the index, developed from all regionally smoothed maps.
5.3 Estimates for orographic regions
5.3.l Introduction
In orographic regions the problems in de r iving generalized PMP charts are much
more complex than for non - orographi~ areas. Differences in topography and its effects,
storm types, amount of data available, etc . , preclude the development of a standard
basic procedure adaptable to the wide variety of situations encountered in making
generalized PMP estimates. While such estimates are usually based on non-orographic
PMP values modified for orography, the modification procedures differ for different
situations. Since there is no standard procedure, summarized examples from actual
studies may provide some guidance on how generalized PMP estimates for orographic
regions may be made . (See cautionary remarks in section 5. 4) The examples presented
in the remainder of this chapter were selected to represent a variety of conditions.
Generalized PMP estimates made by the orographic separation method were discussed in
detail in sections 3.2 and 3.3, and are not included here .
5 000
6 24 hours
2 000
1 000
,.--,.. 500
N
]
'-'
C1l
(l)
~
C1l 200
(l)
~
al
i::
·.-i
C1l 100
~
R
50
1
25 km - - - - - - -
20
50 75 100 125 150

2
Percentage of 24-hour 500 km PMP
' ,'\.
Figure 5.4 - Depth-area-duration re l a tion-

ship for zone 7 of Figure 5.3
5.3.2 Hawaiian Islands PMP
Drainage areas in the Hawaiian Islands are generally less than 120 km2. Iso-
lated peaks extend above 3 000 m for two of the islands, and to about 1 200 m for three
other, larger islands. Numerous investigations have indicated that winds tend to flow
around rather than over the higher mountain peaks. Record-breaking rainfall situa-
tions feature complex thunderstorms and disturbances of the normally prevailing easter-
ly trade winds. The optimum situation was therefore determined /:"~ to be a rela-
tively fixed zone of convergence with imbedded regenerative smaller areas of intense
vertical motion of the size and intensity associated with thunderstorms. Examination
of 156 cases of daily Hawaiian rainfalls exceeding 300 mm disclosed that about 60 per
cent were associated with thunderstorms. Thunderstorms were thus revea led as import-
ant producers of extreme rainfalls, although, as a general weather feature, severe
thunderstorms are relatively uncommon in the Hawaiian Islands.
5.3.2.l
-Non-
- -orographic -
- - - - - PMP
A basic non-orograp hic station, or point, 24-hour PMP of l 000 mm (40 in) was
based on the following considerations: (1) the value agreed with wo r ld-wide extreme
obse rved non-orog rap hic rainfalls in tropical and subtropical re gions, wi th due con-
sideration fo r Hawaii's location and limit at ion on moistu re availability; (2) it
enveloped maximum observed rainfal l amou nts in Hawaii by a reas~nable m arg~n; and (3)
it approximated the value obtained from multiplying the envel~ping P/M ra tio.and ap~
propriate cool-season moisture~ Additional support was provid:d by an earl~ :r esti-
mate of PMP for Puerto Rico /lJ, which is at about t he same latitude as Hawaii .
An empirical relation showing rainfall intensification with slope was develop-

ed f rom observed rainfall data in somewhat comparable terrain . These data indicated
a decrease in the elevation of maximum rainfall amounts as rainfall intensity increased
and an increase of rainfall with ground slope. Precipitation da t a from various parts
of the world were used to determine the general variation in rainfall intensification
with ground slope shown in Figure 5.5.
Greatest intensification is shown for intermediate va l ues of slope (about

0.10-0 . 20) . There is almost no intensification for slopes greater tha n about 0.25 . Such
steep slopes are generally found at the higher elevations, where winds tend to circum-
vent the peaks so that there is little large-scal e lifting of air over the peaks.
The dashed lines of Figure 5 . 5 apply to a column of saturated air with a

l 000 mb temperature of 23°C, and show the depletion of moi sture with increasing ground
elevation. Thus, for any point on the intensification curve, or any given slope, the
elevation at which moisture depletion negates ra infall intensification can be deter-
mined readily. For example, the critical elevation for a slope of 0.17 is about
l 000 m. Above l 500 m, moisture depletion outweighs slope intensification for all
slopes . This is shown in Figure 5.6, which combines the effects of slope intensifica-
tion and moisture depletion to provide a slope and elevation adjustment to the basic
24-hour point PMP of l 000 mm .
5.3.2.3 Generalized PMP estimates
Generalized estimates of 24-hour point (2 km2) PMP are presented in Figure

5. 1. Climatological data showing spillover and other orographic ef f ects were used in
modifying the results indicated by the re lation of Figure 5.6.
Ratios of PMP to 100-year rainfall were examined and adjustments made to avoid
unrealistically high or low ratios . Depth-area- duration relations (Figu re 5 .7 ) for
extending the basic PMP values to du ra tions f rom 1/2 to 24 hours and to areas up to
500 km2 were derived mainly from Hawaiian storms . No seas onal variation curve was re-
quired since the greater efficiency and lower moisture of cool season storms balanced
the lower efficiency and greater moisture of summer season storms.
30
2 2
2 2
2 0 0
0 0
2 0 0 0 0 0
0 m 0 m 0
25 0
0
0 N N
~
m I
w
~
m
~
~
w 20
0
~
w
~
w
~
0 15
~
m
~
§0
~
rJ
10
Dashed lines represent percent age of

precipitable water below indicated
5 elevations in saturated column of air
with 1 000 mb temperatu~e of 23°c
I I I
I I
o,__~~~_._~~~~~~~..L..--<-~~~-'--'-~~~-W'--~~-'----'-~-<-~----'
0 10 20 30 40 50 60 70
Precipitation increase (per centage)
Figure 5.5 - Rain intensification for

ground slope, Hawaiian Islands
PMP for a specific basin is obtained by plani me tering the area within the
basin on the 24-hour point PMP chart (Figu r e 5.1) to obtain t he 24-hour basin-average
PMP. The depth-area-duration relati on of Figure 5.7 is t he n us ed to obtain PMP values
for other durations.
5.3.3 PMP for drainages up to 250 km2 in the Tennessee river ba sin
-----------------------------------------------------------
The Tennessee river basin above Chattanooga, Tennessee, roughly the eastern
half of t he entire basin, was described in section 3.4.2. The western half is rela-
tively low, with rolling hills. Generalized PMP estimates have been made ["'fl for the
entire basin for drainages up to about 8 000 km2. Because of a specific requirement
for generalized PMP estimates for small bgsins up to 250 km2 and the fact t ha t differ-
ent types of storms are likely to produce PMP over small and large a r eas, separate in-
vestigations were conducted for these small basins and for drainages between 250 and
3 500
-1 ()
----------- -
3000 -
2 500 -
,......
~
~
....0 2000 ~--...:. o ____
1 .;;.. - - - - __
+'
nl
I>
Cl>
rl
"'1 1500 Zero
1000
500 ------ - - -- - - -
+ 20
+ 30
20 30
Ground slope (percentage)
Figure 5.6 - Adjustment of non-orographic

PMP for elevation and slope, Hawaiian Islands
8 OOO -km2. Only the estimates for the eastern half of the entire basin are describ-
ed in this manual. The eastern half is referred to hereafter as the project basin.
This section deals with estimates for the small basins. Those for the larger basins
are discussed in section 5 . 3.4.
A record of 56 outstanding point rainfalls in the period 1924-1965, including

a few estimates based on run-off computations, in or near the project basin yielded a
1-hour amount and several 3-hour amounts of about 300 mm . Approximate elevations
ranging from 200 to over l 200 m were determined for most of these storms. No unique
rainfall-elevation relation was evident. This suggested a procedure for estimating
PMP that did not over-emphasize orographic influences on short-duration rainfalls.
Neither was there any noticeable definite geographic distribution of these outstanding
values.
122 ESTIMATION OF PROBAB LE MAXIMUM PRECIPITATION
500
100
,..---_
50 w
N 0
~
a
~ 3 0
...,
5 c:
l/l
c:
.....
c1I
Q)
,.,
l/l
H
«:
10
2
0 20 40 50 80 100
2
Percentage of 24-hour 2 km PMP
Figure 5.7 - Variatio n of index P~P

with basin size and durati on , Hawaiian
Islands
In order to suppiement the basin data, a survey was made of intense small-area
short-duration storms from several hundred storm studies for the eastern half of the
country. Attention was given to all storms with 6-hour 25 km2 rainfall exceeding
250 mm, particularly to those exceeding 350 mm. Some of these storms had durations
of 24 hours. A study of 60 of the more severe storms indicated that most of them in-
tensified during night-time hours. This suggested that factors more important than
day-time heating were generally re sponsible for these outstanding storms.
All information gained f r om the above investigations led to the following con-
clusions concerning small-area PMP for the project basin: (1) the PMP storm-type
situation would involve a continua tion of geographically fixed thunderstorms through-
out a 24-hour period, and (2) the PMP-type thunderstorm for durations of one hour or
less would show little, if any, orographic effect, while that for longer durations
would be likely to produce mo re rainfall on slopes and adjacent volleys than over fla t
areas with no nearby slopes.
Examination of manor storm sites by aerial reconnaissance and inspection of

large-scale topographic maps (1:24 000) led to the following topographic classifica-
tions.
Smooth: few elevation differences of 15 m in 0.5 km.
Intermediate: elevation differences of 15 to 50 m in 0.5 km.
Rough: elevation differences exceeding 50 m in 0.5 km .
Although the entire south-eastern portion of the project basin was classified
as rough, there were variations in rainfall potential across the area. Some peaks
reached up to almost 2 000 m and some ranges sheltered large valleys. The contrast
between high mountains and large sheltered valleys required additional consideration
besides roughness in order to assess topographic effects on intense summer rainfalls.
The effect of local topography on rainfall is discussed in section 5.3.3.4.
Broad-scale topographic effects on rainfall were determined by analysis of

maps of maximum observed and 100-year daily rainfalls. Mean annual and seasonal
precipitation maps were also examined. After some experimentation, the following con-
cepts evolved and were adopted.
First u slo e: a mountain slope facing the lowlands in a direction east to

south -west moistu re-infl ow directions) with no intervening mountains between the slope
and the moisture sources, viz., the Gulf of Mexico and Atlantic Ocean .
Secondary upslope: a secondary upslope high and steep enough to increase

precipitation but partially shielded from moisture sources by a lower mountain range
with an elevation between crests of at least 500 m.
Sheltered areas: these are defined as valleys having moisture-inflow

barriers of 600 m or higher.
Depression: the elevation difference between the barrier crest and a point
in a sheltered area is the depression of that point.
Terrain classifications in the project basin are delineated in Figure 5.8.

Analysis of summer rainfall amounts for the various classifications led to adoption of
the following guides on topographic effects on PMP: (1) precipitation increase of 10
per cent per 300 m from sea-level to 800 m on first upslopes, with no further increase
above 800 m; (2) increase of 5 per cent per 300 m from sea-level to all elevations on
secondary slopes; and (3) decrease of 5 per cent per 300 m of depression in sheltered
areas.
'j ·T---- 'f
N.C.
I.
~
I
s.c. I
.; ·...,. ....... .
\ LEGEND
··.... ·..~
GA. L EG END
- DAM SITES
···· ··· ·· GE N ERA !IZED 1000 Fr C...O NTOURS
e MA JO R C ITIES
- - - - BASIN 8 0 UNOA IUES
c:::::..= SE t O NOdY UPSI OPES I
~ f lRSf UPSiOPf S
SHEi TfRfD RE C. IO N r
-------··· h ·
Figure 5.8 - Topography classified on

basis of effect on rainfall, Tennessee
river basin above Chattanooga, Tennessee
5 . 3.3 . 4 PMP depth- dura t ion curves for 15 km2

- - - - - -- -- - - - - - -
Point rainfall values measured in precipitation gauges and similar containers
are likely to be less than the maximum point rainfalls experienced but not measured .
The maximum point values used were arbitra r ily cons i dered to apply to averag e depths
over 15 km2, the smallest basin size assigned for s t udy . Maximum observed point, or
15 km 2 , rainfalls for durations of up to 12 hours in the eastern half of the country
were transposed and maximized as described in Chapter 2. Outstanding maximi zed and
observed values were plotted against duration (Figure 5 . 9), and curves were drawn for
smooth and rough terrain (section 5.3 . 3.2) .
800
~ 3
4
~
0
~
~ 3
ro
~
~
~ 600
~
0
4
ID
~ • 4
~
~
~
K 400
ro
s
ID
rl
p • Observed
ro
p
x Maximized
0 1 Holt, Mo., 6/22/47
~
~
200 2 Simpson, Ky., 7/4-5/39
3 Smethport, Pa., 7/17-18/42
4 Ewan, N. J., 9/1/40
0
0 2 6 8 10 12
Duration (hours)
Figure 5.9 - Adopted 15 km 2 PMP with

supporting data, Tenne ssee river basin.
(Smooth curve applies to 100 per cent
line of Figure 5.12)
The following concepts and principles were observed in constructing the two
depth-duration curves. Over areas of a few square kilometres and durations up to
about one hour, maximum rainfall rates depend on extreme upward velocities associated
with vigorous thunderstorms. These high velocities are related to storm dynamics,
and topographic effects are negligible. Hence, the same maximum intensities may be
expected within the same air mass over various types of terrain. For longer durations,
terrain roughness becomes increagingly important. First, slopes and roughness accen -
tuate upward velocities. Secondly, intense thunderstorms tend to remain at on e loca-
tion longer over a topographically favourable site than over smooth terrain, where they
drift with the wind or propogate laterally by their own dynamics. Finally, the prob-
ability of continued rainfall after an intense thunderstorm is enhanced by terrain
roughness.
126 ESTIMATION OF PROBABLE MAXIMUM PRECIPITATI ON
150
..,,,
(6 -hour rainfalls) /
. • Smethport , Pa., 7/17-18/42 (627 mm) ,,,.'

..,,,,"'
140 T Ewan, N. J., 9/1/40 (511 mm) /
4RockHouse, N. C., 8/30-31/40 ( 241 ~/
/
Dur ation (hours )
Figure 5.10 - PMP depth- duration curve

for basins up to 250 km2 in Tennessee
river basin
Basic 6-hour 15 km 2 PMP values of Figure 5.9 are applicable to the southern
edge of the project basin . Smooth PMP in rough terrain is hypothetical but serves as
a means for consistent application of adjustments for orographic effects (sections
5.3.3.3 and 5.3.3.5).
Experience with severe storms throughout the country was useful in shaping the
depth- duration curves. The curve of Figure 5.10 was developed to extend the curves of
Figure 5.9 to durations from 6 to 24 hours.
A moisture adjustment chart was developed for the relatively smooth north-
western section of the project basin. This chart (Figure 5 . 11) was based on an assess-
ment of mean dew points and maximum persisting 12-hour dew points. Analysis indicated
a gradient of about 1°C from the extreme south-western corner of the total basin
-· 'i - - - - -- -
GA.
...
+
I
-~-------____,,_-------~---·--- --.'.
Figure 5.11 - Moisture index chart for

north-west portion of Tennessee river
(outside of area shown) to the north-eastern corner, which corresponds to a difference

in rainfall of about 10 per cent, according to the usual model for convective rain in
extreme storms [I, fil. Figure 5.11 shows the moisture index lines, in percentages,
for adjusting PMP values.
A latitudinal gradient chart (Figures 5 . 12) was developed for the mountainous
portion of the project basin. This chart was based on rainfall-frequency gradients
resulting primarily from sheltering by mountains. Moisture effects were incorporated .
'j i' 'i
LEGEND GA.
- DAM SITES
e MAJOR CITlES
- - - - BASIN BOUNDARIES
- - - - - --- ·- ---!. -·--- ---~---+i- -
Figure 5.12 - Latitudinal rainfall

gradient (in percentage) in south-
eastern portion of Tennessee river
The concepts and charts discussed above were used to develop the 6-hour 15 km2
PMP index map (Figure 5 . 13) for the project basin. Six-hour PMP values from Figure
5.9 of 650, 700 (interpolated) and 750 mm were assigned ~espectively to smooth, inter-
mediate and rough terrain categories, and multiplied by adjustment factors indicated
in Figures 5.ll ·and 5.12. Isohyets were drawn with steepest gradients corresponding
to greatest changes in elevation. This naturally placed steepest gradients where
\ LEGEND GA.
- DAMSITES
• MAJOR CITIES
- - - - BAS IN BOUNDAIUES
Figure 5.13 - Six-hour 15 km 2 PMP (in)

Tennessee river basin above Chattanooga,
Tennessee
mountains rise from valley floors. Different adjustments for south-eastern and north-
western portions of the basin (Figures 5.11 and 5.12) resulted in some discontinuity at
their common boundary, which was smoothed out in drawing isohyets. The final 6- hour
15 km2 PMP index map is shown in Figure 5.13. A depth-duration relation (Figure 5. 14)
was developed from a number of PMP depth-duration curves such as Figures 5.9 and 5.10
so that 6-hour PMP could be adjusted to other durations. A depth-area relatlon
(Figure 5.15) was constructed from intense small-area storm data for adjusting the
15 km2 PMP values to other sizes of area.
130 ESTIMATION OF PROBABLE MAXIMUM PREC IPITATION
100
Durati on (hours)
2 3 5 6 12 18 24
90
,......,
a()
...__,
80
rl
rl
"'
.....
i::
·rl
"'F-1
F-1
~
0
70
..C1
I
\D
60
50
20 L.0 60 80 100 120
Rainfall (cm)
Figure 5.14 - Depth-duration relati on

f or 24.hour PMP storm
5.3.3 .7 Time distribu t ion of rain fall
Extreme small-area storms in the project basin generally have been one-burst
events in which little rain followed the extreme 3-hour rainfall. Storm experience
pointed to the occurrence of a 24-hour rainfall in bursts. The following guidelines
were therefore suggested for critical sequences. (1) For 6-hour rainfall increments
in a 24-hour storm, the four increments should be arranged with second highest next to
highest, third highest adjacent to these two, and fourth at either end . This still
allows various arrangements, and the most critical is that which would yield most
critical streamflow. (2) For 1-hour increments in the maximum 6-hour increment, any
arrangement was acceptable so long as it kept the two highest 1-hour amounts adjoined,
the three highest 1-hour amounts adj oined , etc.
250 Duration (hours)
200
....... 150
N
!.,
.,
~
100
50
OL--'-'--'--'--'--'-'-'-_J_-'---'-~l~l-'-l-'-l~1_J_l_IL_J_l-'--l-1L-'-l-Ll_J_...L_L_l_LJ_j_~'~1--'--__l_J
65 70 75 BO 85 90 95 100
2
Percentage of 15 km PMP
Figure 5.15 - PMP depth-area curve for small

basins, Tennessee river basin
PMP for specific basins is estimated as follows.
Step l. Outline the basin on Figure 5.13, and determine mean 6-hour 15 km2
PMP for the basin.
Step 2. Use Figure 5.14 to obtain PMP for durations up to 24 hours.
Step 3. Use Figure 5.15 to adjust 15 km2 PMP for basin size.
Step 4. Construct a smooth enveloping depth-duration curve from data obtain-

ed in step 3, and determine !-hour increments for the maximum six hours and 6-hourly
increments for the remaining 18 hours.
. 132 ESTIMATION OF PROBABLE MAXIMUM PRlbCIPITATION
. Step 5 • . suggest critical time sequences (section 5.3.3.7), such as: (a)
hourly increments in maximum 6-hour period• 6 5 4 3 l 2 h
1 h · • ' ' ' , , , w ere 1 refers to maxi-
mum - our increment, and (b) 6-hourly increments in 24-hour storm.· 4 2 l 3 h
.
l now re f ers t o maximum 6-hour increment. I I I I w ere
5.3 . 4
The discussion which follows refers only to the Tennessee river basin above
Chattanooga, Tennessee J:fil. The topography and moisture sources were discussed
above, and topographic classifications are shown in Figure 5.8.
PMP was derived in the manner described in section 3.4.2. Storms for the
eastern part of the country were maximized in place and enveloping isohyets construct-
ed, thus applying an implicit transposition. PMP maps like that of Figure 3.20 were
constructed for a number of basin sizes and durations, with isohyets not only envelop-
ing the data on each chart but also showing smooth progression with varying basin size
and dura t ion. Values read from these charts for the location of Knoxville, Tennessee,
were used to develop the basic PMP depth-area-duration relations of Figure 5.16. The
24- hour 2 500 km2 chart (not shown) was converted to percentages of the value at
Knoxville (Figure 5.17) . Multiplication of the de pth- area-duration values of ~iuure
5.16 by the percentages of Figure 5.17 yielded non-orographic PMP at various locations
in the basin.
Four indicators of orographic influence on precipitation were developed. These

are summarized in this section. The first three were used to develop relationships al-
ready described.
Mean annual precipitation was one indicator. A hypothetical mean annual non-
orographic precipitation map (not shown) was constructed by eliminating the influence
of the Appalachian Chain by smooth extrapolation of isolines of mean annual precipi-
tation from surrounding non-orographic regions. This map supports the generalized
PMP percentile lines of Figure 5.17.
Charts of 2-year 24-hour rainfalls at some 600 stations in and near the basin
and of extreme monthly rains were used also to assess orographic effects.
Another indicator of orographic influence was the comparison of the small-

basin PMP chart of Figure 5.13 with the chart (not shown) reconstructed under the as-
sumption that the smooth classification applied to the entire basin. Non-orographic
PMP depth-area-duration values (Figure 5.16) are adjusted by the ratio of PMP index
chart values (Figure 5.13) to 6-hour smooth PMP (Figure 5.9) adjusted for basin loca-
tion (Figure 5.12).
10000
5000
1000
500
100
10
200 300 L.00 500 600 700 800
Non-orographic PMP (mm)
Figure 5.16 - Non- orographic PMP at

Knoxville, Tennessee
N.C.
s.c.
LEGEND
GA.
DAM SITES
e MAJOI CITIES
----BASIN IOUNOAllU
Figure 5.17 - 24-hour 2 500 km2 PMP

percentiles of Knoxville Airport value.
(Meteorological observations made at air-
port, about ten miles south of Knoxville)
The optimum inflow direction for heavy rains was another index to orographic
effect . Over a basin of no more than about 250 km2, it is presumed that the optimum
wind direction for unobstructed inflow of moist air and for accentuation of lift by
ground slope prevails during the PMP storm . In larger basins, the optimum direction
for precipitation may differ from one part of the basin to another because of varying
intensification by principal slopes. The wind direction most critical for the basin
~-----';------- - -- - : + -- - -- -- - - - - •11 -
N.C.
s.c.
SOUfHfA , /
GA. I
_J·
LEGEND SOUfH
DAM SITE S
SOUIHV.£ 5 1
• M AJ OR CI TIE S
- - - - BAS I N BOU N DA RIES Wf5f
------:---------~c----------- h
Figure 5 . 18 - Optimum wind directions

for heavy rains
as a whole is defined as the direction that is most favourable over the largest portion
of the basin. Figure 5 . 18 shows the optimum moisture- inflow directions for local
areas . The largest percentage of a pr oblem basin with the same optimum wind dir ection
is determined from Figure 5 . 18 . The orographic intensification factor is relat ed to
this percentag e value by Figure 5 . 19, which was deve lope d empiri cally after a number
of PMP estimates for specific basins had been mad e .
The entire procedure for estimating PMP for s pe cific ba s i ns i s outlin ed in

section 5.3 . 4.4.
100
Q)
~
"'
+> 95
~
Q)
0
F-t
Q)
p,
...__,,
F-t
0
+>
90
0
"'
.....
.....0
..c:p,
0
"'t, 85
F-t
0
80
100 80 60 40 20
Percentage of basin upslope - optimum wind direction
Figure 5.19 - Orographic adjustment

chart for south-eastern mountainous
portion of Tennessee river basin
above Chattanooga, Tennessee
5.3.4.3 Areal and time distribution
The relationships described above yield the volume of PMP for specified sizes
of area and for various durations. Geographic distribution of PMP within problem
basins is determined by developing an idealized or typical representative storm
isohyetal pattern and providing nomograms for obtaining isohyetal values. The pro-
cedure was described in section 3.4.2.5. Critical sequences of 6- and 24-hour rain-
fall increments may be arranged as described in section 3.4.2.6.
5.3.4.4 PMP for specific basins

-- ---- - -
For the relatively smooth north -western portion of the basin (unhatched
regions of Figure 5.18), PMP estimates (s ee cautionary remarks , section 5.4) are ob-
tained from the basic PMP at Knoxville (Figure 5.16) and the regional adju s tm ent
(Figure 5.17). The stepwise procedur e follow s.
Step 1. From Figure 5 . 16, obtain 6-, 12- , 18-, 24-, 48- and 72-hour values
of non-orographic PMP for the basin size.
Step 2. Obtain percentage adjustment indicated in Figure 5 . 17 for the cent r e

of the problem basin, and use it to multiply values obtained in step 1.
Step 3 . Construct a smooth enveloping depth-duration cu r ve from th e adjust-

ed values of st ep 2, and obtain 6-hour increments for the 72- hour PMP.
The proc edure for estimating PMP (see cautionary remarks, sec t ion 5 .4 ) in the
mountainous south - east er n r egion (hatched in Figure 5.18) is more complicated . After
the basic PMP (Figure 5 . 16) is adjusted regionally (Figure 5 . 17) : (a) multiply by
ratio of basin ave r age 6-hour 15 km2 PMP to basic smooth 6-hou r PMP (635 mm, Figure 5 . 9)
adjusted for the basin location (Figure 5.12); and (b) adjust the result fo r percent -
age of basin exposed to optimum wind direction (Figures 5.18 and 5 . 19).
The required steps may be followed more easily if it is assumed that PMP is
being estimated for a hypothetical circular 800 km2 basin centred at Fontana, Tennes -
see, with the results shown in Table 5 . 1.
Step 1. From Figure 5.16, obtain 6-, 12-, 18-, 24-, 48 and 72- hour values
of PMP for the basin size. Enter on line A.
Step 2. Determine the location adjustment factor for the centre of the basin
from Figure 5.17. Enter on line 8.
Step 3. Multiply values of line A by those on line 8 to obtain geographic-

ally adjusted PMP values . Enter on line C.
Step 4. Lay out basin outline on Figure 5 . 13, and determine basin average
6-hour 15 km 2 PMP. (Assume that this value for the example basin is 30 . 0 in, or 762
mm.) Enter on line D under 6 hours.
Step 5. Obtain the non-orographic 6-hour 15 km2 PMP from the smooth curve
of Figure 5.9. The value is 635 mm, and applies to the 100 per cent line of Figure
5.12. Multiply the value by the percentage indicated for the location of the basin
centre. (This percentage is 96 for the example basin.) Enter the product on line
E under 6 hours.
Step 6. Divide value on line D by that on line E to obtain the unadjusted

orographic factor. Enter on line F.
Step 7. Use Figure 5.18 to determine largest percentage of basin having a

common optimum wind direction. (Assume that it is 60 per cent for the example basin).
Enter on line G.
Step 8, Enter Figure 5.19 with percentage value from line G, and read cor-
responding orographic factor percentage. Enter on line H.
I-'
w
00
Table 5.1 - Sample computation of PMP for hypothetical 800 km2 basin centred at Fontana, Tennessee
rn
Ul
-I
Duration (hours) H
:::;::
)>
Line Item and sourc e 6 12 18 24 48 72 -I
H
0
z
A Non-orographi c PMP (mm) at Knoxville for 800 km2 (Figure 5.16) .•• 430 505 560 603 685 740 0
...,,..
"'tJ
B Adjus tment for basin location, in percentage (Figure 5 . 17) . ••. . .. • 102 102 102 102 102 102 :;:o
0
~
C Non-orographic PMP (mm) for ba sin (l ine A x line 8) •••••••••••••• 439 515 571 615 699 755 CJ
rrn
D Mean 6-hour 15 km2 PMP (mm) for ba si n (Figure 5.13) •• .••• •••••••• • 762 ~
x
H
:::;::
E Non-orographic 6-hour 15 km2 PMP (635 mm) from smooth c
:::::
"'tJ
:;:o
curve of Figure 5.9 multiplied by 0.96 from Figure 5.12. .......... 610 rn
(")
H
"'tJ
F Unadjusted orographic factor (line D 7 line E) • .... ..•..... . .....• 1 . 25 1.25 1.25 1.25 1.25 1.25 1-1
-I
)>
-I
G Perce ntage of ba s in exposed t o optimum wind direction H
~
(Figure 5.18) ..•.•.••.•••....•.......... . ..... . .... .... .... ... .... 60 60 60 60 60 60
H Orographic fact or perce ntage (Figure 5.19) ••••••••••• •.... .. . •. . .. 92 92 92 92 92 92
I Net orographi c factor (line F x line H) •••••••.•••. .• ..•..... .. . 1.15 1.15 1.15 1.15 1.15 1.15
J Basin PMP, in mm, (line C x line!) ••••••• .•••...•.•...•.....•.. 505 592 657 707 804 868
Step 9. Multiply values on line F by those on line H to obtain net orograph-

ic factor. Enter on line I.
Step 10. Multiply values of line C by those on line I to obtain PMP values
for the example basin. Enter on line J.
Step 11. Construct a smooth enveloping depth-duration curve from the values
of line J, and obtain 6-hour increments for the 72-hour PMP.
5 . 3.5 PMP for the Lower Mekong r i ver basin in south-east Asia
-------------------------------------------------------
Generalized esti ma tes of PMP were made /9.J for drainages from 5 000 to 25 000
km2 in the Mekong river basin south of the Chinese border at about 22°N latitude
(Figu re 5.20). This part of t he basin is referred to generally as the Lower Mekong.
The procedure used in making these estimates provides an example of how data from one
part of the world may be used to estimate PMP for a region with inadequate data.
A rough approxi mation of regional variation of rainfall potential may be

gained from mean seasonal or annual precipitation maps. A map of mean rainfall was
developed for the May-September season, the south-west monsoon period, which produces
most of the annual rainfall for much of t he Lower Mekong. Rainfall observations pro-
vided the primary basis for the seasonal map . As usual, few observations were avail-
able for mountainous areas.
Where data are severely limited in mountainous regions, as was the case in
the Mekong basin, determination of detailed effects of topography on precipitation is
a hopeless task . In such situations, relations based on extensive smoothing of topo-
graphy are the best that can be developed. Figure 5.21 shows the generalized topo-
graphy of the Mekong drainage and the locations of precipitation stations.
Topographic effects on seasonal rainfall distribution ·were assessed on the
basis of the limited data and on past experience gained from study of these effects in
regions with adequate data. Comparisons of mean rainfalls at a few paire of stations
in the Mekong river basin, critically selected to reflect different topographic effects
within each pair, provided guidance. These comparisons, plus experience, led to the
following guidelines: (1) for mountain slopes facing south to west, with no nearby
mountain barriers to moisture inflow, rainfall approximately doubles in the first
l 000 m rise in elevation. Except for extremely steep slopes extending to high eleva-
tions no further increase was indicated. (2) Upslopes near the coas t, outside the basin
but bounding it, produce spillover rainfall over limited areas in the basin. (3) Shel-
tered areas immediately to the lee of mountain barriers receive about half the rainfall
observed upwind of the barriers.
The above guidelines, plus general guidance from some streamflow data, supple-
mented observed rainfall data in the construction of the mean May- September rainfall
map (Figure 5.22). Mean rainfall maps for August and September, the wettest months,
were constructed in a similar fashion.
96° 100° 108 °
32°
LEGEND
-----WATERSHED BOUNDARY
- • -• - LOWER MEKONG BOUNDARY
0 100 200 300 400 Km .
28 ° SCALE 28 °
20° 20°
H f ." .~· c; .4 1.
96 ° 10 4° 108°
Figure 5 . 20 - Mekong river basin and

sub-basins
100°
22°
28'
20°
• 26
18°
• •• • •
• •
•
..
16°
.. .•.
• • 14°
14°
••
•
.. • •• ..
12°
D LESS TH AN
900 METERS
D 900 TO 2000
M ETE RS
~
A BO VE 2000 200 10°
METERS
10°
100° 102° 104° 108°
Figure 5.21 - Gene ralized topography of

Mekong r i ver basin with precipitation
stations
22°
20°
l!iOO
18°
16°
120
100
100° 102° 10<1° 106° 108°
Figure 5.22 - Mean May-September (south-

west monsoon season) precipitation (mm)
Typhoons are the most important producers of heavy rains for durations of
several days in the Lower Mekong fo r the range of basin sizes considered in this ex-
ample . Such storms, approaching the Mekong basin from the east, produce the heaviest
general rainfalls in the basin in spite of mountain barriers between the coast and the
eastern border of the basin. Rainfalls from typhoon Voe (21-22 October 1952), in the
southern portion of the Lower Mekong basin and Tilda (21-25 September 1964), near the
middle, are foremost examples. Large-area rainfall s from these storms, after adjust-
ment as described below, approximate greatest values from tropical storms throughout
the world.
With the idea of adapting the more abundant dep th -area-d uration rainfall data
from tropical storms along the United States coast to the Mekong drainage, the massive-
ness (si ze and intensity), speed of movement, and other featur es of tropical storms
affecting the two regions were compared. Also compared were av erage maximum 1-day
point rainfalls from tropical storms in the United Stat es and in the Pacific Ocean, in-
cluding the Vietnam coast . Values along the Vietnam coast wer e about 20 per cent
greater, but the excess was attributed to topographic influence s absent in the coastal
regions of south-eastern United States. The comparisons sugge s ted that non-orographic
tropical storm rainfall potential was about the sam e for the two r eg ions .
Two adjustments were made to the U. S. tropical storm dep th-area-duration (DAD)
data to make them applicable to the Vietnam coast. First, the s torm data we re mois-
ture maximized for a persisting 12-hour dew point of 260C, the highest valu e for U.S.
coastal regions affected by tropical storms. Second, an adjustment was made for the
decrease of tropical storm rainfall with distance inland. This adjustment is dis-
cussed in the following section. The adjusted data and enveloping DAD curves are
shown in Figure 5.23. The DAD curves were considered to represent non - orographic PMP
just off the Vietnam coast.
5.3.5 . 4 Adjustment of Vietnam•· tropical storm rainfalls
Since the non-orographic PMP DAD cu r ves of Figure 5.23 applied only to the
Vietnam coast, the indicated values had to be modified for occurrence in the Mekong
basin . The following adjustments were thus required for distance inland, moisture
source, latitude, moisture - inflow barriers and basin topography.
Adjustments for distance inland and moisture source Th e general decrease in

tropical storm rainfall with distance inland previously developed in another study {"4]
was considered applicable to south-east Asia. Approximately 60 U.S. storms in mostly
non-orographic regions were used. Figure 5.24 shows the adjustm e nt for the Lower
Mekong in percentages of the PMP values off the coast.
~
.j:..
.j:..
5000
MEKONG STORMS o ADJUSTMENTS( %)
TO LOCAL
STORM DATE IDENT BARRIER COAST LAT TERRAIN TOTAL
SEPT 21 . 25. 196' (TILDA) A 115 117 100 98 132
OCT . 21 - 22, 1952 IVAE ) 8 100 115 108 97 llO
20001 ~\'~ -;. I '!\ !Ix I ~I I
OTHER STORMS • ADJUSTMENTS (3 )
STORM DATE STQ8M L!;U:~IIQ~ MOISTURE TO COAST ~
'
,..
SEPT 3 - 7, 1950 lo YAHKEETOWH , FLA 110 100 110
JUHE 27. JULY l, 1899 2' HEARNE, TEX 116 102 118
1000 I 9 ~'~~·\~ ~ (0. ~ l "\
< I "\
\ SEPT 25. 28, 1936 . 3o HILLSBORO, TE X 116 123 1'3
JUHE 27 - JULY 2, 1936 • o BEBE, TE X 100 JO• JO• rn
(./)
SEPT 8 - 10, 1921 5o THRALL, TEX 105 11' 120
SEPT 16 - 19, 1928 60 DARLIHGTOH. S. C 121 100 121
-I
.
H
JULY 5 - 10, 1916 7: BONIFAY, FLA 110 106 117 3:
So TEXAS AND ME XICO 115 102 117 )>
500 A -I
H
N- 0
:;; z
:..'.
VI ..,,
0
§ "tl
-.....
<(
200 ;;:o
0
co
""
<(
)>
co
r
n :::,. rn
100 I I ~ I ~
"~ ]~ 1.S~ ~ .... I> .... ::,.J
K: >.j I I I I I
3:
)>
x
H
3:
sol I ? I "? I 2~~ I 2M J~~ ·~ ~' tj>, p.~ "i "\~ I I I I c
3:
"tl
;;:o
rn
(}
H
,, \, I
"tl
20 I I I I ~ I \\·~ I I \ I \~ \ \ I I H
' -I
)>
-I
H
101 I I I I 1 I
·~
' ~i I I\ ~~ \ I I I 0
z
0 100 200 300 400 500 600 700 800 900 1000 llOO 1200
RAIN DEPTH (mm)
Figure 5.23 - Depth-area-duration curves of

probable maximum precipitation on Vietnam
coast
While typhoons approach the Mekong basin from an easterly direc ti on, the wind
circulation br ings in moisture from southerly and easterly directions. The few ana-
lyzed storms in the basin clearly demonstrate multiple sources of moisture . Thus, the
distance inlan'd adjustment (Figure 5 . 24) incorporates a weighting of the generalized
decrease fo r moisture-inflow direction for the region south of 17°N. A weight of one-
third was given to distance inland from t he south coast and two-thirds to distance from
the south-east to east coasts.
Latitude Typhoon rainfall potential must decrease to about zero near the
equator . The literature reports few cases south of 10°N. It was assumed typhoons
could maintain full intensity as far south as 15°N. The need for maintaining a high
typhoon rainfall potential in southerly reaches of the basin is supported by the
October 1952 storm that occurred in the basin near 12°N. The adopted adjustment is
shown in Figure 5.25.
Barrier adjustment In addition to the generalized decrease in rainfall with

distance in non-orog ra phic r egions, it was necessary to consider decrease within the
basin due to moisture-inflow barriers . The decrease varies with height of barriers
and their uniformity, i . e ., whether continuous or with breaks, or posses. Moistu re
inflow from a southerly direction reduces the depleting effect of the eastern coastal
mountains . The eastern barrier was therefore considered to reduce rainfall to the
west by half the usual barrier reduction . Figure 5.26 shows the adopted adjustment
applicable to coastal rainfall values.
Adjustm~nt for bosi~ topography Typhoon Tilda (September 1964), mentioned

above, produced inc reased rainfall along south- west facing slopes in the basin . This
i~ con~istent.with the.assumption that moisture from southerly or south-westerly
directions, with relatively low intervening inflow barriers, must be considered in as -
sessing regional variations in PMP. As an aid in evaluating topographic effects for
the~e inflow directions, ratios of high- to low-elevation mean May-September precipi-
ta~ion were used as p:imary indices. A bias in the mean seasonal precipitation mop
(Fig~re 5.22), resul~in~ fro~ more frequent precipitation at high elevations, preclud-
ed direct use of variations in seasonal precipitation as an indication of variations
in a 3-day storm. Comparison of rainy day station values suggested an increase with
elevation of about 60 per cent over that indicated by mean seasonal values for applica-
ation to typhoon PMP.
Another ad!ustment of monsoon season rainfall ratios lnvolved consistency with

the one-half effectiveness adopted for the eastern barrier adjustment. This implied
that south-west slopes were effective for only one-half the storm duration . The ra in-
fall elevation relation thus becomes 30 per cent of that indicated on the mop. A mean
seasonal low-elevation rainfa ll value of l 200 mm was used as a basic non-orographic
value. Percentage increases for typhoon rainfall on windward slopes and decreases on
lee regions as indicated by south -west monsoon season rainfall (Figure 5.22) are shown
in Figure 5.27.
Combined adjustment Combination of the above adjustments (Figu res 5.24 to

5.27) produced the combined adjustment chart of Figure 5.28, which relates to coastal
Vietnam typhoon rainfall values equated to 100 per cent.
IOI 11'
,. ,.
"
20 10
II II
II II
" "
11
ID ID
100·· 104' IOI

"'
Figure 5.24 - Adjustment (percentage) of

coastal typhoon rainfa ll for distance in-
land
,,
IDO
"' "' "'
,. ,.
11
,,
10 10
II II
II'
"
100 ------- - -- OD
"
95
11 11
so 100 lOO
5UU. KllOMEHIS
10 '
,,.. 102 ~ 104° 106" IOI
Figure 5.25 - Latitude adjustment of typhoon

rainfall as percentage of values at 15°N
100 101
'" "' "
"
\
\
,,. \ 11
\
\
\
\
I
I ~
\ I
\ J
20' '-- ~ 20
11 '
"
11 ' II
- - - . . /.5
14'
"
,,.
"
... IO'
,,.. 102·· ,... IOI' IOI
Figure 5.26 - Barrier adjustment of

typhoon rainfall (percentage decrease)
IOD 102 ' 104 " IDB" 11"
"
22··
"
lD 20·
"
" II'
" '"
12 12-
io ·
Figure 5 . 27 - Adjustment of typhoon rainfall

for basin topography (percentage increase or
decrease relative to low-elevation south- west
monsoon rainfall over flat terrain)
11
JOO IOl
"' "'
o ~o 100 200
c:_-:-::r=:::- -: ----
ID ' scALE lllLOJllEHRS ID
104 ' IOI

"' '"
Figure 5.28 - Total adjustment (percentage)

of coastal typhoon rainfall (combined adjust-
ments of Figures 5.24 to 5.27 )
5.3.5.5 Generalized estimates of PMP
The 24-hour 5 000 km2 coastal PMP values of Figure 5.23 were multiplied by the
combined adjustment percentages of Figure 5.28 to obtain the generalized PMP map of
Figure 5.29. PMP values for basin sizes between 5 000 and 25 000 km2 from Figure 5.23
were expressed as percentages of the 24-hour 5 000 km2 PMP. These percentages were
then used to construct the curves of Figure 5.30.
5.3.5.6 Time distribution
Examination of hourly records of intense rainfalls in the Mekong Basin showed

various sequences of 6-hour increments during a storm period. Those associated with
tropical storms, for example, Tilda of September 1964, had rain bursts lasting up to
30 hours with greatest intensities near the centre of the burst. Some stations re-
ported double bursts with an intervening lull of 6 to 18 hours.
Strictly speaking, in order to maintain PMP magnitude no lulls can be allowed

in a sequence of 6-hour rainfall increments during the PMP storm. In other words , the
greatest, second greatest, etc., down to the twelfth greatest must be arranged in an
ascending or descending order such that the highest increments always adjoin. Such a
sequence is unrealistic, however, and that described in section 3.4.2.6 was recommended
as essentially conforming to requirements for the 72-hour PMP storm.
5.3.5.7 Areal distribution
Isohyetal patterns for 6-hour rainfall increments in observed storms have

various configurations. Some approach simple concentric circles or ellipses, while
others are complicated, often with centres of high and low rainfall in close proximity
to each other. An elliptical pattern, similar to that of Figure 3 . 26, was recommend-
ed for the four greatest 6-hour rainfall increments. Uniform areal distribution was
recommended for the remaining 48 hours of the storm.
Within a 3-day period, the isohyetal centre of a major storm usually moves
along the storm path. In the most extreme rainfalls, the storm may become almost
stationary. It is therefore considered reasonable to have the isohyetal centre over
the same location for a 24-hour period in the PMP storm.
Depth-area-duration relationships in the heaviest tropical storm rainfalls of

the Mekong basin and the United States were used to establish isohyetal values for the
selected pattern. Particular attention was given to maximum 6- and 24- hour rainfalls.
For these durations, consistent depth-area curves were constructed for standard area
sizes of 5, 10, 15 and 25 thousand km2 , With the 6- and 24-hour relations establish-
ed, the second and third heaviest rainfall increments were computed proportional to PMP
increments at standard size areas. The dashed curves of Figure 5.31 represent adopt-
ed depth-area relations for key basin sizes and durations. The solid curves are based
on Figure 5.23. The storm depth-area curves and PMP depth-area-duration data were
used to develop nomograms like that of Figure 5.32 for evaluating isohyetal value
Such nomograms are derived by the procedure described in section 2.11.3, the only
difference being that isohyetal values were converted into percentages of average rain-
fall enclosed by the respective isohyets and presented as a nomogram instead of in a
table.
102· 104° 106° 108°

+ + + +24°
22·+
+ + +22°
20·+
350 + + +zo·
ta·+ + +18°
16°+ + +16°
14°+ +
12·+
+ +
10°+ + + + + IO°
98° 100° 102° 104° 108°
0 50 100 200 KM
I 11111 I I
Figure 5.29 - Probable maximum precipitation

for 24 hours over 5 000 km2
25,000
20 ,000
"'"'
~ 10,000
5,000
6 12 18 24 48 72 hrs.
10 20 30 40 50 60 70 80 90 100 :o 120 130 140

PERCENT OF 24-hr. 5000 Km2 PMP
Figure 5.30 - Depth-area-duration values of

PMP in per cent of 24-hour 5 000 km2 PMP
PMP for specific basins (see cautionary remarks, section 5.4) is estimated as
follows.
Step l. Lay out basin outline on Figure 5.29 and det e rmine average 24-hour
5 000 km2 PMP for the basin.
Step 2. From Figure 5.30, read percentages of 24-hour 5 000 km2 PMP for 6,
12, 18, 24, 48 and 72 hours for the basin area.
Step 3. Multiply basin average 24-hour 5 000 km2 PMP from step l by the per-
centages of step 2 to obtain basin PMP.
Step 4. Use data from step 3 to construct a smooth de pth-duration curve, and
read off 6-hour PMP increments for the entire 72-hour storm.
100.000 ,----.,--....--,.---.--.,.--.---.-__,.--,.---.--.,.--....--.---.--.,.---....--.--....--.----r--r--.
\ ''
\ ''
\
\ ''
\
\ ''
\ ''
\
\ ~ '' "'
\
\ '' ' tf' '' ''
\
\
~
' ' ~~ ' '~ ''
\
' ~ '~
\~ ' 'J'.'~o
\'j.
\~ '' ' ,.., ,*+
,..,,..+.
\~ '' ,.,
t'v.'1'~
,~,..,
\
'' ' \
'
''
f.o
\ ' \~ \
\
\
,~
\~
\,..,
\
\
\
'' \
\ \ \
\ \ \
\ \
\ \ \
\ \ \
\ \
\ \ \
\ \ \ \
I \
100 200 JOO 400

I
~00
\
600 700
'
800 900 I000
\
1100
AVERAGE DEPTH Imm)
Figure 5.31 - PMP (solid lines) and key depth-area

curves typical of major tropical storms
Step 5 . Arrange 6- and 24-hour increments as de scribed in section 5.3 . 5.6.
Step 6. Use selected elliptical isohyetal patte r n (not shown) to distribute

the four grea t est 6- hour rainfall increments. Centre and orient pattern over the
pr oblem basin so as to obtain most critical runoff, which usuall y r e sults with great-
e st rainfall volume within the basin . Enter Figure 5. 32 with basin area, ana r ead
percentage values for each isohyet, P to E, for the maximum 6-hour increment. Multi-
ply the maximum 6- hour PMP increment of step 5 by these percentages to obtain isohyetal
values in mm. Values for second, third and fourth PMP increments are obtained in a
similar manner from similar nomograms {not shown).
25,000
E D C B A p
- ISOHYET -
20,000
.......
...
""
:z:
~ 10 ,000
""
5,000 L--~....~u..~--i.~~..__,__,_~...i..~--i...&..~.._......_._~........~~.._~.......~-"-~--'
20 40 60 80 100 120 140 160 180 200 220 240 260
PERCENT OF PMP
Figure 5.32 - Isohyetal values for maximum

six-hour increment of PMP storm as percentage
of average rainfall for area enclosed
5.3.6 Generalized thunderstorm PMP estimates for north-western United States
5.3.6.1 Introduction
Thunderstorm rainfalls usually provide the maximum amounts for small areas,
say up to about 1 000 km2, and durations shorter than about 6 hours. Extreme observ~
ed values indicate less latitudinal variation within middle latitudes than do general
storms. While severe thunderstorms are often associated with vigorous weather sys-
tems, some of those producing extreme rainfalls occur during periods of weak atmos-
pheric circulation. For this reason, and because of their small areas, it is gener-
ally impossible to determine with any reasonable accuracy the moisture inflow into such
storms. While there is no generally accepted procedure for deriving estimates of
thunderstorm PMP, the fo llowing example from a study ~fJ] for the semi-arid upper
Columbia river basin in north-western United States (Figure 5.34) may serve as a guide .
In that region, heavy thunderstorm rainfalls are rarely associated with general-type
storms, but occur generally as isolated events.
5.3.6.2
-PMP- -depth-duration
- - - - - - - relation
----
Extreme rainfall amounts for various locations in or near the project reg ion
(Figure 5.34) were moisture maximized (section 2.3) to 73°F (22.8°C), th e ma ximum per-
sisting 12-hour l 000 mb dew point for the extreme south-eastern portion of the project
region in August. The maximized values ore shown plotted and env eloped in Figure5.33,
156 ESTIMATION ~ PROBABLE MAXIMUM PRECIPITATION
which represents thunderstorm PMP values for durations to 6 hours at a point in

the extreme south-eastern corner of the project region. Also shown are observed
values from sto r ms considered not transposable to the project region . Durations of
maximized rainfalls controlling the depth-duration curve did not extend beyond one
hour, so ratios of 6- to !-hour ra infalls in moderate but longer duration thunderstorms
were used to extend the curve to six hours. On the basis of experience with dense
precipitation networks, an area of 2 km2 was assigned to the point values.
400
300 x?
)( 8
x9 Observed Storm Adjunted

200 rain Dura.ti on dBW point rairr*'
~
Station Date (mm) (min) (DC) (mm)
~
....... l . Morgan ( nr ) , Utah a.16.19sa 170 60 19.4 23?
.-i 2. Girds Crk., Ore • 7,13.1956 102 30 16.7 168
.-i
nl 3. Simon Ranch, Idaho 7. 21.1956 64 20 18.3 94
~ 4. Birch Crk. , Ore. 6.22.1938 64 20 15.0 127
·rl
nl
5, Richland, Ore , 6.15.1954 15 10 15.0 30
~ 100 6. Mitchell, Ore . 7, 13,1956 89 . 30 16.7 15
Following observed rainfalls not transposed - for comparison only :
7. Campo, Calif. 8.12.1891 292 80
8. Buffalo Gap., Sask ., 5.30. 196 1 26 7 60
Can.
• 5 9. Chiatovich Flat, 7. 19 ,1955 210 150
Calif.
o 10. Fort Mohave, Ariz. 8.28.1898 203 45
* Adjusted to 22.8°C
0 2 3 t.. 5 6
Duration (hours)
Figure 5.33 - Point (2 km2) thunderstorm PMP

for 22.8°C on extreme south-eastern border of
upper Columbia river basin
Gl~ERA L IZFD ~STlMATES
157
_____ _,,...-~--*----.-------·---· ---· ·-·

1-
,,·--·
l'-•
-+- ~ - -·
J 10'"
Basin d i stance from southe a st borde r
... 100
Q)
...
"'O
0
_o
w
Vl
June
80
0
"':>
O>
:>
<t:
..... 60
0
c:
Q)
...u
Q)
Q...
40 300 200 100 0
600 500 400
700
Distance from SE bord e r (mi.)
Figure 5.34 - Regional and seasonal

variation of thunderstorm PMP
158 ESTIMATION OF PROBAB LE MAXIMUM PRECIPITATION
5.3 .6.3 Seasonal and reg i onal variation s

-· - -
- - - ·- - - - - - - - --
Thunderstorm rainfalls provide guidance in det ermining seasonal (within summer
months) and regional variations. Amounts for short durations, say l and 3 hou rs, must
be obtained from recording- gauge stations, which usually have relatively short records .
In the example study, the upper 20 per cent of recording-gauge measurements were us e d
as one guide to the seasonal and regional variations of thunderstorm PMP . Another
guide was the variation in maximum moisture with season and region. A composite of
th e variations indicated by these two types of data resulted in the variations of PMP
indicated in Fi~ure 5.34 . The parallel lines of the upper chart, which gives distance
from the south-eastern border of the project region, are oriented approximately along
maximum pe r sisting dew-point lines. The lower diagram, whic h shows the variation of
thunderstorm PMP with month and distance from the south- eastern border, is based on
moisture variations indicated by maximum persisting dew points.
5.3.6.4 Elevation adjustment
The observed extreme values of point thunderstorm rainfall occurred at eleva-

tions from 300 to 3 000 m. Data were too sparse to indicate any distinct trend with
elevation. The much more abundant autographic record extremes, which were consider-
ably smaller, did not provide any definite indication either, although there was a
suggestion of a decrease for elevations above l 500 m. A decrease of 5 per cent for
each 300 m above l 500 m was therefore adopted on the basis of the decrease of moisture
in a saturated pseudo-adiabatic atmosphere .
None of the extreme thunderstorm rainfalls used in developing the PMP depth-
duration curve (Figure 5.33) occurred over dense precipitation networks, so the depth-
area re lation had to be based on other thunderstorms. Analysis of several such storms
with adequate data led to the depth-area curves of Figure 5 . 35.
The areal distribution of thunderstorm rainfall within a basin is often re-

quired. One way of showing the areal extent of a storm is to assume circular isohyets
and to construct isohyetal profiles of depth against distance from the storm centre, or
isohyetal radius (section 2 . 11.3). Figure 5.36, which is based on the same data as
Figure 5.35, shows the adopted isohyetal profile for thunderstorm PMP.
The idealized isohyetal pattern (Figure 5.37) was derived for a mod e l 2-hour
thunderstorm. The 2- hour duration was a compromise for each 1- hour PMP increment to
simplify procedures for application. The model thunderstorm involved the following
assumptions: (a) depth-duration re lation as i n Figure 5.33; (b) circular isohyets;
and (c) storm movement of 4 mile, or 6 km, per hour . The isohyetal pattern, toget her
with Table 5.2, is used to determine average depth of PMP over any portion of a basin.
The procedure for evaluating isohyets was described in section 2.11.3.
~
p,.,
80
N
s
~
N
'H
0
Cl>
QO
o:I
+'
~
60
Cl>
0
f-1
Cl>
p,.,
200 L. 00 600 80 0 1 0 00
2
Area ( km )
Figure 5.35 - Area-reduction curves

for thunderstorm PMP
5.3.6.6 Time distribution
Analysis of time distribution of thunderstorm rainfalls reveals many varia-

tions. The following arrangement of hourly increments was recommended: highest in-
crement in second, third or fourth hour, with next highest on either side; third high -
est adjacent, etc.; and smallest two increments at beginning and end. For example, a
possible realistic arrangement would be: 5, 2, l, 3, 4, 6, where l is the greatest in -
crement.
5.3.6.7
- - - - - - PMP
-Thunderstorm - - for
- - specific
- - - - -basins
--
(See cautionary remarks, section 5.4). If the areal distribution of thunder-
storm PMP is not required, basin average depths may be obtained as follows.
Step l. Obtain 1-, 3- and 6-hour values of point, or 2 km 2 , PMP from Figure
5.33.
~
C\J 80
~
C\J
60
20
0
0 8 12 16 20
Radius (km)
Figure 5.36 - Isohyetal profile for

thunderstorm PMP
Step 2. From the upper chart of Figure 5.34, obtain distance of problem
basin from south-east border of project region, and use this distance in lower diagram
to obtain percentage of August PMP for whatever month(s) required.
Step 3. Multip l y PMP values of step l by percentage(s) of step 2 to obtain

2 km2 PMP for location of basin.
Step 4. If lowest elevation in problem basin is above l 500 m, reduce values

obtained in step 3 by 5 pe r cent for each 300 m above l 500 m. No adjustment is re-
quired if lowest elevation in basin is l 500 m or lower.
Step 5. Use depth- area- duration c urves of Figure 5.35 to obtain percentage
adjustments for basin area, and apply to results of step 4 (or st ep 3 if elevation is
not required) in order to determine basin average PMP.
Step 6. Plot basin average PMP values of step 5 against duration, dr aw

smooth enveloping depth-duration curve, and determine hourly increments.
Step 7. Arrange hourly increments as suggested in section 5.3.6.6.

1 2 6 8
miles
I 2 .C 6 B
M - I
kilometers
Figure 5.37 - Isohyetal pattern for

thunderstorm PMP
If areal distribution of thunderstorm PMP within the basin is required, pro-

ceed as follows:
Steps 1-4. Same as above except only 1-hour PMP required.
Step 5. Lay isohyetal pattern (Figure 5.37) over problem basin outline of
same scale. Centre and rotate pattern to provide greatest average depth over basin.
Step 6. Obtain labels from Table 5.2 for isohyets up to the minimum size re-
quired to enclose basin outline completely.
Table 5.2 - Pattern thunderstorm isohyetal labels (in per cent of !-hour 2 km2 PMP)
(1) (2) (3) (4) (5) (6) (7) (8)
Isohyet Area enclosed Hourly increments of PMP in descending order
(km2) (mile2) 1st 2nd 3rd 4th 5th 6th
A 2 l 100 19 10 6 5 4
B 16 6 76 19 10 6 5 4
c 65 25 54 19 10 6 5 4
D 153 59 40 17 9 6 5 4
E 246 95 32 14 8 5 4 4
F 433 167 21 10 7 4 3 3
G 635 245 14 7 5 4 3 3
H 847 327 8 4 4 3 3 3
I l 114 430 l 2 2 2 2 3
J l 396 539 0 0 0 0 l 3
Step 7. Multiply !-hour 2 km2 value of step 4 by isohyetol percentage labels

of step 7 to obtain isohyetal values in mm.
Step 8. Determine overage dept h over basin or portion thereof by planimeter-

ing or other procedure.
Step 9. Arrange hourly increments as suggested in section 5.3.6.6.
5.4 Cautionary remar ks
Generalized estimates of PMP ore representative for individual basins having

topographic features similar to the generalized topography used in deriving the esti-
mates. PMP for individual basins with different features may be considerably differ-
ent from the generalized values, especially in orographic regions. Generalized esti-
mates are generally more representative for the l arger basins of the size range con-
sidered in this chapter. These larger basins usua lly have some topographic features
similar to those on which the generalized estimates ar e based. Smaller basins, on th e

other hand, may have topographic features entirely unlike th e general features of the
area in which they are located, and generalized estimates th er e fore tend to be less re-
liable.
The step-by-step procedures given in this manual for estimating PMP for spe-
ei f ic basins serve merely to summarize the methods used in deriving the PMP estimates
and the techniques used for applying the results to specific ba s Lns. They are not in-
tended to enable the reader to obtain PMP values for specific bu3 ins in the regions
covered by the examples. For this reason, only thos e charts a r,d tables required for
illustrating the approach used are included. Additional chart s and tables would be
required for making complete PMP estimates for specific basins.
Other, equally valid approaches besides those represented by the examples have
been used for developing generalized estimates. As mentioned earlier, the approach
used depends on the geography of the project region and the amount and quality of re -
quired data. Basic data requirements for reliable estimates are adequate precipi-
tation networks and dew-point and wind data. A thorough knowledge of meteorological
characteristics of storms likely to govern PMP limits is an important requirement. This
knowledge is most important where basic data are sparse.
The cautionary remarks of section 2.13 relative to adequacy of storm sample,

comparison with record rainfalls, consistency of estimates, seasonal variation, and
areal distribution apply to generalized estimates.
References
1. Myers, V. A., 1967: Meteorological estimation of extreme precipitation for

spillway design floods. Technical Memorandum WBTM HYDR0-5, U.S. Weather Bureau,
p. 14.
2. Riedel, J. T., Appleby, J. F. and Schloemer, R. W., 1956: Seasonal variation of

the probable maximum precipitation east of the 105th meridian for areas from 10
to l 000 square miles and durations of 6, 12, 24, and 48 hours. Hydrometeoro-
logical Report No. 33, U.S. Weather Bureau.
3. Schwarz, F. K., 1963: Probable maximum precipitation in the Hawaiian Islands .

Hydrometeorological Report No. 39, U.S. Weather Bureau.
4. Schwarz, F. K., 1965: Probable maximum and TVA precipitation over the Tennessee
river basin above Chattanooga. Hydrometeorological Report No. 41, U.S. Weather
Bureau.
5. Schwarz, F. K. and Helfert, N. F., 1969: Probable maximum and TVA precipitation
for Tennessee river basins up to 3 000 square miles in area and durations to 72
hours. Hydrometeorological Report No. 45, U.S. Weather Bureau.
6. U.S. Weather Bureau, 1960: Generalized estimates of probable maximum precipi-

tation for the United States west of the 105th meridian. Technical Paper No. 38.
7. U.S. Weather Bureau, 1961: Generalized estimates of probable maximum precipi-

tation and rainfall-frequency data for Puerto Rico and Virgin Islands. Techni-
cal Paper No. 42.
8. U.S. Weather Bureau, 1966: Probable maximum precipitation, northwest States.

9. U.S. Weather Bureau, 1970: Probable maximum precipitation, Mekong river basin.
ACKNOWLEDGEMENTS
The authors wish to express their appreciation to

Mrs . M.E. McCarty and R.E. Evans who rendered valuable assistance
in the collection and reproduction of various illustrations, and
to Mrs. C. Ebner who typed the manuscript.
BIBLIOGRAPHY
Adil, M. A. and Suffi, M. M., 1964: Probable maximum precipitation over the Tarbela
Dam basin. Scientific Note, Vol. 16, No.3, Pakistan Department of Meteorology and
Geophysics.
Alexander, G. N., 1963: Using the probability of storm transposition for estimating
the frequency of rare floods. Journal of Hydrology, Vol. l, No. l, North-Holland
Publishing Co., Amsterdam, pp. 46-57
Bell, G. J. and Chin, P. C., 1968: The probable maximum rainfall in Hong Kong. R. O.
Technical Memoir No. 10, Royal Observatory, Hong Kong, 145 pp.
Bruce, J. P., 1959: Storm rainfall transposition and maximization. Proceedings of

symposium No. l, Spillway Design Flood~ at Ottawa, Canada . National Research Council
of Canada, pp. 162-170.
Bruce, J. P. and Clark, R. H., 1966: Introduction to hydrometeorology. Pergammon

Press, London, New York, Paris, Toronto, 319 pp.
Commonwealth of Australia, Bureau of Meteorology, 1958: Conference on estimation of

extreme precipitation, Melbourne.
Dhar, O.N. and Kamte, P. P., 1969: A pilot study for the estimation of probable maxi-
mum precipitation using Hershfield techniquef~d;'!0ournal of Meteorology and Geophysics,
'· /
Institute of Tropical Meteorology, Poona, Vol. 20, No. l, pp. 31-34.
Fletcher, R. D., 1951: Hydrometeorology in the United States. Section in Compendium

of meteorology. American Meteorological Society, pp. 1033-1047.
Gilman, C. S., 1964: Rainfall, Section 9 in: Handbook of applied hydrology. Edited
by V. T. Chow, McGraw-Hill, New York.
Hounam, C., 1957: Maximum possible rainfall over the Cotter River catchment. Meteor-
ological Study No. 10, Commonwealth of Australia, Department of Meteorology.
Knox, J. B., 1960: Proceedings for estimating maximum possible precipitation. Bulletin
No. 88, California (U.S.A.) State Department of Water Resources.
Koelzer, V. A. and Bitoun, M., 1964: Hydrology of spillway design floods: large
structures ~ limited data. Journal of Hydraulics Division, Proceedings of American
Society of Civil Engineers, Paper No. 3913, pp. 261-293.
McKay, G. A., 1965: Statistical estimates of precipitation extremes for the prairie
provinces. Canada Department of Agriculture, PFRA Engineering Branch.
Moazzam, S. M., 1964: Probab le maximum precipitation for November-May season over the
Swat river basin. Scientific Note, Vol. 16, No. 5, Pakistan Department of Meteorology
and Geophysics.
Myers, V. A. 1967: The estimation of extreme precipitation as the basis for design
floods, r~sum~ of practice in the United States . Extract of Publication No . 84~ Sym-
posium of Leningrad, International Association of Scientific Hydrology.
Ogrosky, H. 0., 1964 : Hydrology of spillway design floods: small structures - limit-
ed data. Journal of Hydraulics Division, American Society of Civil Engineers, Poper
No. 3914, pp. 295-310.
Paulhus, J. L. H. and Gilman, C. S., 1953 : Evaluation of probable maximum precipita~

tion. Transactions. American Geophysical Union, Vol . 34, pp. 701-708.
Riehl, H. and Byers, H. R., 1958: Flood rains in the Bocono basin, Venezuela. Deport-
ment of Meteorology, University of Chicago.
Showalter, A. K., 1945: Quantitative determination of maximum rainfall. Section in:

Handbook of meteorology. Edited by F. A. Berry, E. Bolloy and N. R. Beers, McGrow-
Hill, New York, pp. 1015- 1027.
Singleton, F. and Helliwell, N. C., 1969: The calculation of rainfall from a hurricane
in: Floods and their computation, Vol. 1, International Association of Scientific Hy-
drology, Publication No. 84, pp. 450-461.
United Nations/World Meteorological Organization, 1967: Assessment of the magnitude

and frequency of flood flows. Water Resources Series No. 30, pp. 13-49.
Wies vier
WeisRer, C. J., 1964: Hydrometeorology and river flood estimation. Proceedings, In-
stitute of Civil Engineers, London, Vol. 27, pp. 153-167.
\Ai ' es he.r,.
~~, C. J . , 1970: Hydrometeorology . Chapmon and Holl, London, 232 pp.
World Meteor ological Organization, 1969: Estimation of maximum floods. WMO-No. 233.
TP.126, Technical Note No. 98, pp. 1-116.
ANNE X l
TABLES OF PRECIPITABLE WATER IN A SATURATED

PSEUDO-ADIABATIC ATMOSPHERE
As stated in Chapter 2, precipitable water is a term used mostly by hydro-

meteorologists for expressing the total mass of water vapour in a vertical column of
the atmosphere. It represents the depth of liquid water that would accumulate at the
base of the column if all its water vapour were condensed. The term is a misnomer
since no natural process can condense or precipitate all the water vapour in the atmos-
phere, and substitute terms such as liquid equivalent of water vapour or liquid water
equivalent are sometimes used.
The general formula for computing precipitable water, W, in cm, is:
w== q~p/gp (A.l.l)
where q is the mean specific humidity in gm kg- 1 of a layer of moist air· .6p is the
depth of the layer in mb; g is the acceleration of gravity in cm sec-2; and ,P is the
density of water, which is equal to l gm cm-3.
In much of hydrometeorological work the atmosphere is assumed to contain the

same amount of water vapour as saturated air with a saturation pseudo-adiabatic tem-
perature lapse rate. The precipitable water in various layers of the saturated atmos-
phere can be pre-computed and listed in tables or in nomogram form. Table A.l.l pre-
sents values of precipitable water (mm) between the l 000 mb surface and various pres-
sure levels up to 200 mb in a saturated pseudo-adiabatic atmosphere as a function of
the l 000 mb dew point. Table A.1.2 lists similar values for layers between the l 000
mb surface, assumed to be at zero elevation, and various heights up to 17 km.
170 'ESTIMATION ·OF 'PROBABLE MAXIMUM PRECIPITATION
Table A.1.1 'Precipitable water (mm) between l 000 mb surface and indicated pressure
(mb) in a saturated pseudo-adiabatic atmosphere as a function of the
1 000 mb dew point {°C)
. .__._ _.'t~...,.._...._...__..__..__.__....o.........._ . 1. ....z_.._1._3_1..•:.....~1...),__0-'C.__

· mb · "'__
-900 O 1 1 1 1 l 1
_ _i __ .l ___ ,_
~-- · .:_ _l__.._'_ _...__ _..__ _.__...__.__.). _ _ L .2. J. .z_
9 7Cl l 1 l 2 2 3 3 3
-·~,~·~-~_z_~-~-~-..L..-"--'---"-~~~-~---- 4
1• - 2 2 3 3 3 3 ..
- "- - -~-
5
.2-2. . _J._ _.1__2. - 3--3._ 4 .
2 "- . 4 5
3 3 3 3 5 5 6 1
~2..: .. ....l. _ _, 3 .. _i>._ __4 .• . .._ ~-~-~- . b ·7 6
91: 3 5 6 7 7 8 9
~-~-'---"---'"--""'---"--""-~"---"'---"~-"'--'- .1-11..·_ 9 )Q_
89: 5 8 8 9 9 10 11
sa: 4 _ _.__..._-i. . - L B. 9 _9_ . 10 ll 12
97C 4 5 5 b 6 1 1 8 8 9· 9 10 11 12 l3
S!.: 4 . ~ __ .i__ ,b._b __ .b , _ L _ _ L _ l l ., _9 .--2--10 11 12 12 13
e~: 5 5 6 6 1 7 6 9 9 10 11 11 12 1> !4
.a~ .• .5 i 7 R' 10 JO 1 1 _ ...l..2._l,.3._l't... . .,!.5
5 6 6 1 7 6 10 1 1 '12 !3 14 15 16
.b - ..A.__l_. _1._ ___e___e_ 10 ] ] _ _u _ _ l.2 . .1..J . 14 . l!i . 17
6 6 1 8 6 9 10 10 11 12 lJ 14 15 16 J7
ac: b __ b •. :1 . .J.. •. 8 . . .ll. -9-1..'l--1.l. . .l.L. U .. 13 .1..5 16 17 18
h 6 1 1 8 9 9 10 11 12 JJ 14 15 16 17 1;
b J __:i. _.e __ .e,__ 'C ,, 11 ~-J.'4 .1.b !7 ll:l lY
6 7 7 8 q 9 10 11 12 13 lt. l" lb ]7 117 20
.. ~ .1 .• 7 a 9_ 10 lO .. ..l~ .•n :• .. 15 . 11 18 :9 2:
1 8 8 9 lC l! 12 13 14 15 16 1 I 18 2J 21
8 9 .9. !0 11 . l2 __ l3 . 14 15 ~6 18 19 2C Z2
1 8 9 9 l-' 11 12 13 14 15 17 18 20 2J l3
.L-4. le 'l P 1' l!.......l1 l.B 20 22 . .23
8 8 10 11 12 13 14 15 16 17 19 20 22 2•
8 ll . IQ _li-----l.2... . ll...-1~ -15 16 te 19 21 z; 24
69C 8 9 ·9 10 11 12 13 14 15 J7 18 20 21 2J 25
69.: s l :o 10 l i .. 12.. ll. .15 16 17 19 20 22 ;:4 25
67: . e 9 JC 11 11 12 14 15 16 17 19 20 22 24 26
~6C _ e. . ,~_9___!.~. - lL..U . 11-..l!! '5 16 •a 19 21 23· 2• 2~
65:' . 8 8 9 10 ll 12 13 14 15 !6 18 :9 21 23 2> 27
64: e 6 2 .. 10 . JI 12 IJ .l't..-15_ .l.J 18 20 21 23 2> 27
63( 8 8 9 10 11 12 1) 14 16 17 18 20 22 24 Zb 28
62: d 9 9 10_11 12 13 14 16 17 19 20 22 24 26 26
6' - d 9 9 10 11 12 13 15 16 17 19 2C 22 24 20 28
.i.:.;:, a. _«L·__9 .,. 1 • l2 l..3--li-~~LL.i.2-.2..l 23 25 21 29
59: ~ 9 10 10 :1 :2 : .4 15 16 18 19 21 23 25 27 29
8. • ..!I. ...llL...l.!... !! .-1.l -14_.l!_ _ 16._ .l!L.19 21 . 23 25 27 JO.
9 10 1! 12 :3 14 l' 16 lS 20 21 23 2S 27 J'J
9 10 ...ll. ;2_ 13 14- .. .li_ . .l.7 . 18 . 20 .. 21 23 26 Z8 ~o
9 JC 11 12 !3 14 lS 11 18 20 22 24 26 28 JC
8 . _ 9 . !O • ..l..l .... l..2... lJ li.. .. .ll__D._ .18 . . .ZO -22. 24 26 2o 31.
d 9 I~ l: 12 lJ 14 15 17 18 20 22 24 26 Z& JI
8 _ ... 9 10 .. .ll.. J.2 _13 !4_..l.J>__l.J_ .!9 20 2Z 2• 26 " ))
B 9 lC 11 12 lJ 14 lb 17 19 20 22 24 Zb 29 31
9 , l\L.Jl 12 D l "-lb. 17 lq 20 22 24 27 29 l2
490 9 10 11 12 13 14 16 )7 19 21 22 25 27 29 32
~.e;; 5 .. 9 l.O. -1.1......JZ.._l.J ,_li .. ..l.!>. -11 19 21 23 z, 27 , , )2
4 7 ·~ B 9 IC 11 12 !3 14 16 17 11 21 23 25 27 29 32
'96..C ___ B 9 lC-11. .lZ ll...l..L!h _-17 19 21 2J 25 27 3~ l~
4~: P. 9 IC 1! 12 13 l4 fb Jl l9 21 2~ 25 27 30 32
w...::. - a _q_--1.0__l..l.......J..2 .. ....!....?. •• ...l.i__.l.A __ J.7 _.!.9 .21 z3 z~ 21 .. 1~ JJ
41:: 9 9 10 11 !2 ,., :o; :6 l7 !9 2! Z'J 25 27 JO 3l
9 10 11 12 i.~ 15 !~ !e 19 ' l 2~ 25 27 JC :3
'1 10 11 12 :: !S 16 18 19 21 23 25 27 30 '.'3
9 ia. . ..ll :z !J _:5 _ .J.ti._J5 !.9. 21 . 43 zs 2e }O J:
9 :o 11 :2 13 15 16 °:8 19 21 23 25 28 JC )3
9 )0 II :2 :J._ 15.. .lL Je J9 2: 23 2~ 28 lO n
9 :c 1~ 12 13 ls lb 18 19 'l 23 2s ~8 :-.o J)
_9 :c_ ...!..!..... __ : i _ - · 1c. .,, __!.!l u .l.3 .z~.. ~!3 J:_ .J:
3'.·: "J IC 11 12 4- "' !S :b l~ 19 21 23 25 ZB :o 3:'
I~ 11 12 13 _ I L. l b. . H 19 2..1 2J 25 ZS 3: ll
10 11 !2 lJ 15 lb l~ }Q 21 2~ 2~ 29 ., .... 1
"'
u ll 12 i~ is . .l.b...-le 19 _21 23 z~ 2e --
9 1c 11 1~ n 15 16 le 1q 21 2:? 2':> ze J~ n
__ 7_:..:::...__:..:, __ :2 . ..l.J. !..L_lb._l.-'~'lf._ .. 21 ZJ • . Z~ 4:9 ): .,,
9 •r :l ., D ~5 ~b B 19 ~1 23 ZS 26 "' " D
1: _ :i. . .iz !3 l~ •. 1e.__Je .. . l9 2~ Z3 25 4:'.e J: J;
!O 11 !2 13 l'J 16 UI 19 21 2J 25 28 , ... :')
:O ll. ..... !3 l~ - - ~~ - !S 19 2~ 2:? 2~ 'S , ... )}
.• :1 !2 !3 ~') lb !E 19 21 23 25 26 ,,., L'
_:; ... !.l . :z 1 ·4 .:...:.L ~ .:~ :9 .Z. ! ZJ 4:5 Z~ , .. ,,
9 .,. , . •, ·- !5 l~ 19 !q 2~ 2~ 2") Z ·' -- ..
7 ....... ll :z .,, :s :ei :e .. ., z.:. 23 zs ze Jc J:
9 _. . . . . :1 •. :~ !b l! tq 2! 2:' Z5 2' ., . . )~
9 ·- .. " .... ,.J :b ·o !9 Z! "tJ ~5 26 i~ 31
ANNEX l 171
Table A. 1.1 (continued)
mb .i~b-~·~7_1~0~-
~-~1~9~-~?~.\..=2.~~ - -] '' -·~~-~-~1~
'·~~~2~~~~2~"--
oc
1 112 222 22
- 2 2. _ __...._~--~-~- '"- __ J,,,.__~_ "' 5 5
4 4 44 5 . b 6 7 7
. ..:9«6...-~_.- ..5 -5,__ . .i. b.....-"----''-------'---'---"--· ::; _ ___s_ _L,O__l ,L_
95 J 6 b 1 7 10 lC 11 12 12 13
94::: 1 . J __ e . 9 'C IC 'l----l..2- . l.2 !.).. ;.i. ~5 :i
93~ B B 9 9 10 11 11 12 13 14 14 l 1l 16 17 : 0
;;!J 9 - . .9 ;.., ~ - lZ-- :3 __.....___....~_11_ __ ;,,__ - ~' 19 2J . ••
91J 10 10 11 i2 13 13 14 15 16 ~ 7 15 2 0 21 22 2J
...>..e..:i -~-----~~~-~----~~~--~~ . .;::i-.;.t.._. :..L.. -
B9J 12 12 13 :• l> 16 17 18 20 · 21 22 · 2• 25 21 2a
aau . ~· . i J l4.-.l5-l6 .1 .1 . .+0--~0-__;_.-"'J--.2"'- . 4'.t. 21 29 J!.
870 lJ I• 15 lb 18 19 20 21 23 2• 26 zB29 31 3J
86.i_ l.1.t. • . l.!'l ~~-19 . -la 21 '' . ' " '6 .,a -30 . .JZ .s.i.. ___Jb
B5J 15 16 18 19 20 21 2J 24 U 28 JO 32 3• 36 JB
_a~ H, b 1 1 , g ' ,. '1 , 1 "}:. ., b , a ., o ' 2 1 a.._40_
8JC 17 18 19 21 22 2• 26 27 29 31 33 35 38 •O o;
. 8 2 ::: . l..8 .-.lg _ .2 CL z,_2-'u'~-'2~s~_,_2~1__,_2g,._-'3,_•,__,3._,1~_,,1_,_5 3. 7 ---4..0 _ 42 . £.!I
81~ 19 20 21 23 25 26 20 30 32 34 37 39 42 4C. 47
_a.co . ...J.9. _..l.J. __ -Z...2.--.2Jt . lb. --2.a. _ u,_J~- - ~'~6~~'~'-·" l. - 44 __ 46, _ _,.9
79:, zc 22 23 2!1 27 29 31 33 35 38 40 4) .C.6 49 52
18..C......ii-2...1.~ . ?fi Vl J "l 'H • J" )? t ? 41-."'5.. . .5~
11J z: 23 2s 21 z9 Jl J3 J5 38 4: 43 46 49 sJ 5b
.7a.:l_ . ~;:i 24 2b. 20 ..l~-3..2.. -~..,l..l._Jg_u..L <. !> 4 S si ss sa
750 23 ~5 27 2~ Jl 33 35 JS 41 44 41 50 53 57 60
. 700 _ .-L4 _i_o 25 _;a _ _32.--J."---11- -~ ,,_; _ _4.JL 5: ~ 59 bl
7JV 24 26 28 30 jJ 35 38 '90 43 46 50 53 51 60 64
.J2..J._ .. ~ ., s .,b 19 <z '+ 5 5· s 5 ~b.2..--6L
71~ 26 28 30 32 :?.; J7 40 43 46 49 53 56 bO 64 60
.. 10.:. lb 28 .J: J! - ~5 30 . . u___/,,~-~1- ~o_.....sz. sa 62 &b 1:1
b9.) 27 l9 31 )4 36 39 42 45 46 52 55 59 bl 68 72
6ac 21 Jo 12 . .34 _ :n 40 ..4l.._lt£i._4..2.. 53 ... .57 &l 65 69 74
670 2s 30 33 ;5 3a 41 44 47 51 54 58 bZ 67· 11 76
_b!,;il; .. _,l9_....Jl __!l.__l.!:.._.)9 42 t.5 /,A 5") L ,J !;_4_ b!S _J..l.. ?fl. .
&s::: 29 31 1z. J7 J9 42 46 .:..9 53 s1 6! 65 10 11) 80
. blt.C 29 12 35 37 40 4.3_ 4.~i..._58 . _ &.:? &7 11 76 a!
63~ JC 12 35 33 4l 44 47 51 55 IJ9 63 6a 73 78 83
&2: 3:: -ll 30. 3& • "'2 4..!> .. -4~2-!lb..._ 60 _b..5 - 69. JM.. 1.9 - 8~
610 31 33 36 39 •2 05 49 5) 57 61 66 11 76 81 87
_b:.=. .. Jl .," · ~4.3 "O 51. 'S t' L/ '1? .1.2...~
590 12 34 37 .:..~ 4J 47 5: 55 ,9 63 &a 11 1a 84 90
~Bil-32- . 35 .2a_ 4..l.. 44 .IJi--5....!.._ ~5 6Cl 64 . b'i . 74 ao ss 91
570 32 35 38 •' •5 •8 52 56 61 65 70 75 Bl 87 91
~ .33 ; L J 9 _ il_<,; _42_ ~3 __5L_;,_L~ 71 71 8Z 88 9•
550 33 36 39 42 46 49 53 58 62 b1 12 18 BJ 90 Qb
~ 'b .,9 ~Yb !i.O . • St. i" 1,,1-6...Q.-1.J . l~f..!l. 91-97
530 34 3~ .40 4) 47
50 55 59 b4 69 74 80 66 92 99
.52il .J4 .....3..1.-..:.0.. . ..J&..3 •• . .:.7.
..!ll-. 55.--..b..O-.-...bt._ .. 70- 15 al 61. 93 ~VO
510 34 37 40 44 48
51 56 60 65 70 76 82 ff8 95 10~
.. !.lo: 3" . 37 .l..l.. . 44 <.8 12 56 ..Q..!.. .• -66..... 7l Jl . .Jll 89 96 .!Cl
470
'-9~
~---
35
35 38
38
...1o1 1o5
~--~...c..Sl ~.5."'
42 45 "9
48
52
53
57
58
61
/."'
62
66
68
12
13
78
19
84 90
67 -1~ -=.a. .. .C..S...~l--Q~
as 9l
97 ic.:.
99 :oc:i
+60 JC, ....J,.B.. . 4-2- 4~ - . ..t..9...--!...4- 58--6.J-._o...l.. . 7o. .C.O 86 93 lOO l.06.
45i:: 35 39 42 lo6 50 54 58 63 69 74 81 67 94 10\ 109
-~ J..Q.~ 1 6 SO S' S? 6' 69 - ~--Z.l.-d.8 .. 9S .. l.;2 l::.
430 36 39 42 46 ;J ;s 59 6 .. 70 76 02 63 96 !OJ il:
42.: 36 39 4J 46 i;o 55 60 65 10 76 a2 89 96 104 112
4lJ 36 39 43 47 51 ;5 60 65 11 11 83 9n 91 lj5 i:1
4).; JC:i )9 -'t) '11 -51 ;:;, .. 60. 6i_ 71 77. . d'1 'JO 98 105 1:'1
);., 36 39 43 47 51 Sb 60 6b 7: 77 B4 91 QB 106 ll'J
:w.: )ti '.\? '1) 47 51 !16 .bl bb . 72 70 .9., 92 ·J9 107 :!')
)7.: )ii toO 4) '17 'Jl 'Jb bl 6b 11 78 o5 92 !VO l;JB l l t..
1t:i.: JL tic. 1:1.:._. _v __ ;1 . .. 6 bl c· .. ~_7 'J ~-93 · -. ! C:l ::r
35:; )L 40 4) '17 !:12 5o bl 6"1 7J 79 Ut- C:::) l"' ' l r f:f. : ::i
).:...: Jt. .. c 4J .. 111 . ;.z -~o . -.b..:._0 1 . 1 3 _ 7? 6:> 93 !.i:it 1:-1 ::o
;3,; .,._ 4C 43 47 5: 56 6~ 67 73 ~G db 94 10~ !:O i.!9
J~: . .l!l 4C 1,4_ z..S 5:!. 57 62 . 67 . 71 . .J.J . .57 9 .. l ~ :!. ill :2J
J j; t.j .:.,4 t.O :io! 57 62 67 73 SJ 97 9- :.:>4: tll :.:c
r - t.,;: 41., t...:_ )4 ~:"_- !>.~!;:._}-~:' 9; .... __ ......
2·J: t.0 t.t. t.5 ;2 57 62 6d 74 bO 07 '15 ~03 112 ,.,.
40 "" <.3 . !..2. 5:' 6.Z...... 6~ :Jz. S.;) £: 5 Ii'> :JJ !12 ~!1
~-~
2 7: lb .:.a 1.,4 40 52 s·: ~z 6Ci 11. 01 fi.5 q; 1:.:. .... i::
.2.6C ":: 44 ~8 52 . 57 6~ t. i. 1"' dl 56 9o !.'.;4 !ll .....
'5.: t.C z.i.. t..d ~2 S7 b2 60 7,;. 61 63 9b tOi.. l!J i.~.!
.!A.: ,..:; z.-...._t..;._ !...! .5.7-t..Z- .. t.:.__ :: ..._. :..:._.:.::..._ 9 b ~.:..'-..!.!.l . ..:.~J
~J: '-n t.. .. i..O ;,2 57 6~ 6ti 74 ~~ C".i 96 :~.:.. lll !iJ
'~: 30 ...... 4t. <.S !12 ~7. ci2 b8 .:._ dl Ob 9tl ..... :D !.2J
n: Ji i..".) 4b ,z ;1 bl. CJh 1.:. o! .:i.i 9c:i : o ; ~l'- l:J
)0 4~ 44 .c.a s2 !d b2 a..a -1~ o!. aa 96 ~c~ 114 ~=J
Table A.1.2 Precipitable water (mm) between l 000 mb surface and indicated height (m)
above that surface in a saturated pseudo-adiabatic atmosphere as a func-
tion of the l 000 mb dew point (°C)
Height l 000 -b Temperature ( oc)

(m) 0 l 2 3 4 5 6 8 9 10 11 12 13 11; 15
200 l l l 1 l l l 2 2 2 2 2 2 2 2 2
400 2 2 2 2 2 3 3 3 3 3 l, 4 4 5
600 3 j 3 3 3 4 4 4 5 ~ 5 6 6 0 7 7
800 3 3 4 4 4 5 5 5 6- 6 7 7 3 d 9
l 000 4 4 4 5 5 6 6 6 7 7 8 9 9 10 10 !l
l 200 4 5 5 6 6 1 7 a a 9 9 10 ll 11 12 13
l 400 5 5 6 6 7 7 8 s 9 10 10 11 12 13 14 15
l 600 5 6 0 7 7 8 9 9 10 11 11 :..3 ::. 15 16
f 800 6 6 7 7 8 9 9 10 11 12 :2 13 14 15 l7 lS
2 ()()() 6 7 7 8 9 9 10 11 11 12 l3 14 16 17 13 lS
2 200 7 7 8 8 9 10 10 11 12 13 14 15 16 18 19 20
2 400 7 8 8 9 9 10 11 12 13 14 15 16 17 19 20 22
2 600 7 3 8 9 10 11 11 12 13 11. 17 l:l 20 21 23
2 800 7 9 9 10 11 - 12 13 14 15 l6 13 19 21 22 24
3 000 s 3 9 lO 10 11 12 13 14 15 17 lS 20. 21 23 25
3 200 8 8 9 10 11 12 13 14 15 16 17 19 20 22 24 "~o'
3" 4oo s 3 9 10 11 12 13 14 15 lo 13 El 21 23 1.L, 26
3 600 3 9 9 10 11 12 13 14 15 17 - 0 20 22 23 25 27
3 800 s 9 10 10 11 12 13 14 16 17 19 20 22 24 26 2C
4 000 8 9 10 11 11 12 14 15 16 li 19 2: 22 26
4 200 8 9 10 11 12 13 14 15 16 18 19 23 25 27
4 400 s 9 10 11 12 13 14 15 l~ 13 20 21 23 25 27
4 600 8 9 10 11 12 13 14 15 17 13 20 22 24 25 13
4 800 3 9 10 11 12 13 14 15 17 18 20 22 24 2.6 23
5- oOO 3 9 10 11 12 13 14 16 17 19 20 22 24 26 23
5 200 s 9 10 11 12 13 14 16 17 19 20 22 24 26 29
5 400 3 9 10 11 12 13 14 16 17 19 20 22 24 26 29
5 600 s 9 10 11 12 13 14 16 17 19 21 22 24 27 29
5 800 8 9 10 11 12 13 14 16 17 19 21 22 25 27 29
6 000 8 9 10 ll 12 " 13 15 16 17 19 21 23 25 27 30 32
6 200 8 9 10 11 12 13 15 15 17 19 23 25 27 30 32
6 400 8 9 10 11 12 l3 15 16 13 19 21 23 25 27 JO 33
6 600 8 9 10 11 12 13 15 16 18 19 2l. 23 25 27 3J 33
6 soo s 9 10 11 12 13 15 16 18 !9 2!. 23 25 27 30 33
7 ()()() 8 9 10 11 11 14 15 16 13 19 21 23 25 2G 30 33
7 200 0 9 10 11 12 14 15 16 13 19 21 23 25 28 30 33
7 400 8 9 " 10 11 12 14 15 16 18 19 z: ::3 25 :;o
7 600 8 9 10 11 12 14 15 16 18 19 23 :25 28 30 jJ
7 800 s 9 10 11 12 14 15 16 13 19 23 25 ZS JJ 33
8 000 8 9 10 11 12 14 15 16 13 19 23 25 26 30 33
8 -200 8 9 10 11 12 14 15 16 13 19 21 23 26 2S 30 33
8 400 8 9 10 11 12 14 15 16 lS 19 2:!. 23 3::; 33
8 600 3 9 10 11 12 14 15 15 13 :9 2:. 23 .:.·,;,
8 800 9 :o 11 12 14 - 15 lS l~ 19 23 33
9 000 3 9 10 11 12 14 15 lo lS 19 23 33
9 200 8 9 10 11 12 14 15 16 lS 19 23
9 400 14 15 16 :;; 19 2~
9 600 14 15 lo lS 19 21 23
9 800 14 15 l~ lS 19 23 :s
10 000 14 15 16 18 19 21 23 2J
11 000 2l 23 2:! 33
12 000 31
ANNEX l 173
Table A.1.2 (continued)
1 000 mb TemperotOJr~ (•C)

H•i;ht
(m) 16 17 13 19 20 21 22 23 24 25 26 27 29 30
200 3 3 3 3 3 4 4 4 4 4 5 5 5 6 6
400 5 5 6 6 6 7 7 8 8 9 9 10 10 . 11 12
600 7 6 s 9 10 10 11 11 12 ::.J 14 15 15 16 17
800 10 10 ll 12 13 13 14 15 l6 17 18 19 20 21 22
l 000 12 13 13 14 15 16 17 lS 20 21 22 23 25 26 23
l 200 14 15 :6 17 18 19 20 21 23 24 26 27 29 31 32
l 400 16 17 :.8 · 19 20 22 23 24 26 :!S 29 31 33 35 37
l coo 17 19 20 21 23 24 25 27 29 31 33 35 37 39 41
l 800 19 20 22 23 25 26 23 30 32 :::4 . 36 39 41 43 46
2 000 21 22 24 25 27 29 31 33 35 37 39 42 1,4 47 50
2 200 22 24 25 2.7 29 31 33 35 37 40 42 45 1,3 Sl 51'
2 400 23 2.5 27 29 31 33 35 37 40 43 45 48 51 54 57
2 600 2:, Z6 23 30 32 35 37 40 42 45 48 51 55 53 61
2 800 26 27 30 32 34 36 39 42 45 . 4S 51 54 58 61 GS
3 000 27 29 31 33 35 33 41 44 /.,} 50 53 57 61 64 68
3 200 23 32 34 37 40 42 45 49 52 56 59 63 67 7l
3 400 2S :a 33 3c, 41 44 47 51 54 SS 62 66 ;o 74
3 600 29 32 34 37 39 42 45 49 52 56 cO 64 68 73 i7
3 800 30 32 35 33 41 44 47 50 51. 53 62 ~6 70 75 60
4 000 31 33 36 39 42 1,5 4S 52 56 60 64 68 73 73 83
4 200 31 34 37 40 43 46 49 53 57 6l 65 10 75 eo ss
4 400 32 34 37 40 44 47 51 54 55 63 67 72 77 82 S7
4 600 32 35 33 41 44 48 52 56 60 64 69 74 79 84 90
4 800 33 36 39 42 45 49 53 57 6: 65 70 75 31 86 92
5 000 33 30 39 42 46 SC 54 53 62 67 72 77 82 G~ 94
5 200 37 40 43 47 50 54 59 63 66 73 78 34 so %
5 400 .h 37 40 1;7 51 55 60 S4 69 71, 60 es 92 93
5 600 35 3S 41 44 48 52 56 60 65. 70 76 81 87 93 lOO
5 800 35 33 41 45 4S 52 57 6l 65 7:. 77 82 83 95 101
6 000 35 33 42 45 49 53 57 62 67 71 7S e4 90 SS 103
6 200 35 3.3 42 45 49 54 53 63 63 73 79 85 91 S3 104
6 400 35 39 42 46 50 54 53 63 6S 74 so 86 92 99 lCo
6 600 30 39 42 46 50 54 59 64 69 74 80 87 93 100 107
6 Boo 30 39 42 46 50 55 60 65 70 75 61 87 94 101 108
7 000 39 43 46 55 60 65 70 76 82 SS 95 102 llO
7 200 36 39 43 47 51 55 60 65 71 75 S9 96 103 lll
7 400 36 39 43 47 51 5.J 61 . 66 71 77 83 90 97 104 112
7 600 36 39 43 47 51 56 61 66 72 77 S3 90 93 lGS ll3
7 800 36 39 43 47 51 56 61 66 72 73 S!, 9l 98 1 V6 114
3 000 36 40 43 47 52 56 61 . 67 72 78 es 92 99 107 115
8 200 36 40 43 47 52 57 62 67 73 7S :.lS 92 :oo 10$ 115
8 400 36 40 43 47 52 57 62 67 73 79 SS 9:! lCO 1J3
8 600 36 .....
'' 43 47 52 57 02 61; 73 79 56 93 101 ::~9
8 800 J(, 40 43 47 52 57 62 63 73 79 93 !Cl :00
9 000 :;.3 4C 43 47 52 57 62 G:l 71, :;.) E6 S4 102 ilC
9 200 3.J 40 43 43 5:! 57 62 63 74 $() $7 9~ 10 2 12.C
9 400 36 40 44 43 52 57 62 74 £J i> 7 9.!t 102 ::.c
9 600 36 40 44 43 52 57 63 6J 74 30 137 9!1 l J'.:! !.l!.
9 BOO 3L> 40 44 4.3 52 57 63 6~ 74 GO 57 95 103 lll
10 000 37 40 44 4~ 52 57 G3 6S 74 80 J7 95 lJJ 1:2
11 000 37 L.C 44 4:3 52 57 .;3 u:s 7:, 31 5:3 96 :c!. ::J
12 000 37 40 44 43 52 57 63 63 74 Sl o..;. . 96 lCS E~
13 000 52 57 63 63 74 Sl 97 lOS
14 000 52 57 63 6S 74 Sl 97 105 ::s
15 000 Sl 97 lJ~ ::"
16 000 Sl 97 l:.l :LlS
17 000 97 105 115
ANNE X 2
GREATEST KNOWN RAINFALLS
World-wide record and near-record rainfalls are listed in Tables A.2.1 and
A.2 .2 respectively . The values of Table A.2.1 are shown plotted against duration in
Figure A.2.1, which also gives the equation of the straight envelope, with R being the
rainfall in inches, and D, the duration in hours.
The extreme rainfall values of Tables A.2.1 and A.2.2 may be used in judging
the general level of PMP for some locations. Such values are associated with a small
number of storm types and geographic locations, and their applicability is limited.
The record values of Table A.2.1 for 9 hours to 8 days are from two different tropical
storms on the Island of La Reunion in the Indian Ocean. There, typhoons, or cyclones
as they are called in that part of the world, collide with steep mountains reaching up
to over 3 000 metres under circumstances so favou r able for rain that the resulting de-
luge is not readily transposable to other r egions lacking equally steep and high moun-
tains so close to the sea. The near-record rainfall values listed for China in Table
A. 2. 2 suggest that its PMP may be of the same order of magnitude as that for La Reunion .
For locations of less rugged topography, lower values of PMP might be expected, and
there is then justification for excluding the values listed for La Reunion and China
in Tables A.2.1 and A.2 . 2 as guides for estimating PMP.
Since the values listed in Tables A.2.1 and A.2.2 for durations from 4 hours
to 8 days are mostly from tropical storms, they should not be used as indicators of PMP
magnitude in regions not frequented by such storms . Obviously, small-area PMP in cold
climates or over basins well protected by orographic barriers and located far enough
from their crests so as not to be affected by spillover will fall considerably below
the values listed in these two tables.
The point values of Tables A.2.1 and A.2.2 may be reduced to areas up to
l 000 km2 by means of Figure 4.6. This reduction for size of area is far from a re-
fined procedure since such area-reduction curves vary both regionally and with storm
type. These curves are generally too unreliable to permit the point values of these
two tables from being used as guides to PMP estimates for large basins. World - record
and near-record rainfall values on a volumetric basis are unavailable . Table A.2.3
gives maximum depth-area-duration data obtained from some 700 analyzed storms in the
United States. The large majority of these listed data are from tropical storms, and
caution should be used in developing ratios from this table for use in other regi ons.
.......
°'"
200 0 ff -
100 0
BOO
-~~]- · - --- .---~-· - ·-=
.~1-::rc----
1--- ·T- ~ 1 ~~1- i =r=l~.71·/.<r=-/-
I I//'
600
L ~· ,~ -
r -~- -_--
•CO f----1--f--+-~
..L~
r"1
-- --· - -· ·· -·---- -·- .._J -··; - - _c:_ (./)
Ul 200
Ql
,.q
--- _L_-:::1=rrr= --- -- - -- ,_, __ ,__: I
·'' f-- -
.
-I
H
3:
g 100
)>
-I
•.-! BO ---H/ i"L - - ~_, __ ]__ - H
. ~~- .- -y;: - -Cilao1 L0 -- _ _ _ 0
- ~. -- ,.. ~· --1--l z
60
r-. , - -·- - - -· Reun~on
·
Belauve,'- -Lo· Re union -
';;j •O - 1 - - -1
...,
0
'+-i • Smethport, Po.
----~~ 1D'Hanis,
~
-~ 20 Texas .!- .l.- -- --1-- I -0
:::0
I
: ~ =-- _.·l~--;~u~~,-C;:i:- t-::,~. Holt,

Rockport, W. Vo. 0
P:l
±I OJ
1
10
~§-=a·-i-1/ Mo. ]
1
)>
OJ
d~0 ::igc•:~· ic -- =-~- .
r
Rou±mon
·
r"1
1~·-· Fosse~, Bova~io 3:
';_" '"':'~·-"IJj_~L _Lu_,_~

)>
2 ~l---1+ . - ·--- - ----1 - 1--l-+ I

x
H
3:
c
3:
-0
_u_ JJliLLJ_ ___ L L l I I 1111 I I JL_ L I :::0
I 2 4 6 8 10 20 •O 60 3 6 9 12 16 24 5 10 20 30 2 3 6 9 12 2•
r"1
",,---- '-----:y-----___, '------·· ----y- ._ _ _ _ ) (')
H
Minutes Hours Days Months -0
Duration H
);!
-I
H
0
z
Figure A. 2 . 1- World 1 s greatest obs erved point rainfalls

ANNEX 2 177
Table A.2.1 World's greatest observed point rainfalls
Depth
Duration Location Date
(in) (mm)
l min 1.50 38 Barot Guadeloupe 26 Nov, 1970

8 min 4.96 126 Fussen, Bavaria 25 May 1920
15 min 7.80 198 Plumb Point, Jamaica 12 May 1916
20 min 8.10 206 Curtea-de-Arges, 7 July 1889
Roumania
42 min 12.00 305 Holt, Mo. 22 June 1947
2 h 10 min 19.00 483 Rockport, W. Va. 18 July 1889
2 h 45 min 22.00 559 D'Hanis, Tex . 31 May 1935
(17 mi. NNW.)
4 h 30 min 30.8+ 782 Smethport, Pa. 18 July, 1942
9 h 42.79 l 087 Belouve, La Reunion 28 Feb. 1964
12 h 52.76 l 340 Belouve, La Reunion 28-29 Feb. 1964
18 h 30 min 66.49 l 689 Belouve, La Reunion 28-29 Feb. 1964
24 h 73.62 l 870 Cilaos, La Reunion 15-16 Mar. 1952
2 d 98.42 2 500 Cilaos, La Reunion 15-17 Mar. 1952
15 d 188.88 4 798 Cherrapunji, India 24 June - 8 July, 1931
31 d 366.14 9 300 Cherrapunji, India July 1861
2 mo 502.63 12 767 Cherrapunji, India June-July 1861
3 mo 644.44 16 369 Cherrapunji, India May-July 1861
4 mo 737.70 18 738 Cherrapunji, India Apr.-July 1861
5 mo 803.62 20 412 Cherrapunji, India Apr.-Aug 1861
6 mo 884.03 22 454 Cherrapunji, India Apr.-Sept 1861
11 mo 905.12 22 990 Cherrapunji, India Jan,...Nov. 1861
l yr l 041.78 26 461 Cherrapunji, India Aug.1860-July 1861
2 yr l 605.05 40 768 Cherrapunji, India 1860-1861
Table A.2.2 Near-record rainfalls
Depth
Duration Location Date
(in) (mm)
l min 0.65 17 Opid's Camp, Calif. 5 Apr. 1926

5 min 2.48 63 Porto Bello, Panama 29 Nov. 19ll
14 min 3.95 100 Galveston, Tex. 4 June 1871
40 min 9.25 235 Guinea, Va. 24 Aug. 1906
l h 10.00 254 Catskill, N.Y . 26 July 1819
l h 20 min ll.50 292 Campo, Calif . 21 Aug. 1891
3 h 16.00 406 Concord, Pa. 5 Aug.1843
4 h 23.00 584 Basseterre, St. Kitts, 12 Jan. 1880
W. Indies
12 h 30.72 780 Baguio, Philippines 17 Oct. 1967
15 h 34.50 876 Smethport, Pa. 17-18 July 1942
18 h 36.40 925 Thrall, Tex. 9 Sept. 1921
21 h 41.7 l 059 Kadena Air Force Base, 8 Sept. 1956
Okinawa
24 h 65.83 l 672 Hsin-liao, China 17 Oct. 1967
24 h 49.13 l 248 Paishih, China 10-ll Sept. 1963
24 h 47.86 l 216 Baguio, Philippines 17-18 Oct. 1967
24 h 40.80 l 036 Cherrapunji, India 14 June 1876
24 h 40.10 l 019 Jowai, india ll Sept. 1897
39 h 62.39 l 585 Baguio, Philippines 14-16 July 19ll
2 d 88.94 2 259 Hsin-liao, China 17-18 Oct.1967
2 d 82.ll 2 086 Bowden Pen, Jamaica 22-23 Jan. 1960
2 d 63.64 l 616 Cherrapunji, India 14-15 June 1876
2 d 15 h 79.12 2 010 Baguio, Philippines 14-17 July 19ll
3 d 108 . 21 2 749 Hsin-liao, China 17-19 Oct. 1967
3 d 99.52 2 .528 Bowden Pen, Jamaica 22-24 Jan. 1960
3 d 80.52 2 045 Cherrapunji, India 25-27 June 1931
3 d 15 h 87.01 2 210 Baguio, Philippines 14-18 July 1911
4 d 109.79 2 789 Bowden Pen, Jamaica 22-25 Jan. 1960
4 d 101.84 2 587 Cherrapunji, India 12- 15 June 1876
5 d ll4 . 50 2 908 Silver Hill Plantation, 5-9 Nov. 1909
Jamaica
5 d ll4.14 2 899 Cherrapunji, India 12-16 June 1876
6 d 122.50 3 ll2 Silver Hill Plantation, 5-10 Nov. 1909
Jamaica
6 d ll9.37 3 032 Cherrapunji, India ll- 16 June 1876
7 d 131.15 3 331 Cherrapunji, India 24- 30 June 1931
7 d 129.00 3 2n Silver Hill Plantation, 4-10 Nov. 1909
Jamaica
8 d 135.05 3 430 Cherrapunji, India 24 June-1 July 1931
8 d 135.00 3 429 Silver Hill Plantation, 4- ll Nov. 1909
Jamaica
ANNEX 2 179
Table A.2.3 Maximum observed depth-area-duration data for the United States
(Average rainfall in inches and (millimeters))
Area Duration (hours)

6 12 18 24 36 48 72
10 mile; 24.7a 29.8b 36.3c 38.7c 41. 8c 43. lc 45.2c

26 km (627) (757) (922) (983) (1062) (1095) (1148)
100 mile; 19. 6b 26.3c 32.5c 35.2c 37.9c J8.9c 40.6c

259 km (498) (668) (826) (894) (963) (988) (1031)
200 mile; 17.9b 25.6c 31.4c 34.2c 36.7c 37.7c 39. 2c

518 k111 (455) (650) (798) (869) (932) (958) (996)
500 mile; 15.4b 24.6c 29.7c 32.7c 35.0c 36.0c 37.3c

1 295 km (391) (625) (754) (831) (889) (914) (947)
1 000 mile; 13.4b 22.6c 27.4c 30.2c 32.9c 33.7c 34.9c

2 590 km (340) (574) (696) (767) (836) (856) (886)
2 000 mile; 11. 2b 17.7c 22.5c 24.8c 27.3c 28.4c 29 . 7c

5 180 km (284) (450) (572) (630) (b93) (721) (754)
5 000 mile; 8. lbj 11. lb 14. lb 15.5c 18. 7d 20.7d 24.4d

12 950 km (206) (282) (358) (394) (475) (526) (620)
10 000 mile; 5.7j 7.9k 10.le 12. le 15.ld 17.4d 21. 3d

25 900 km (145) (201) (257) (307) (384) . (442) (541)
20 000 mile; 4.0j 6.0k 7.9e 9.6e 11.6d 13.8d 17.6d

51 800 km (102) (152) (201) (244) (295) (351) (447)
50 000 mile; 2.5eh 4. 2g 5.3e 6.3e 7.9e 8.9e ll.5f

129 500 km (64) (107) (135) (160) (201) (226) (292)
100 000 mile; l. 7h 2.5ih 3.5e 4.3e 5.6e 6.6f 8.9f

259 000 km (43) (64) (89) (109) (142) ( 168) (226)
Storm Date Location of centre
a 17-18 July 1942 Smethport, Pa.

b 8-10 Sept 1921 Thrall, Tex.
c 3-7 Sept 1950 Yankeeto'Wll, Fla. Hurricane
d 27 June-l July 1899 Hearne, Tex.
e 13-15 Mar 1929 Elba, Ala.
f 5-10 July 1916 Bonifay, Fla. Hurricane
g 15-18 Apr 1900 Eutaw, Ala.
h 22-26 May 1908 Chattanooga, Okla.
i 19-22 Nov 1934 Millry, Ala.
j 27 June-4 July 1936 Bebe, Tex.
k 12-16 Apr 1927 Jefferson Parish, La.
SUBJECT INDEX
Area-reduction curves 4.2.5, 5.3.6 . 5
(See also Depth-area relations)
Areal distribution of PMP:
basin PMP 2.11.3.l
idealized storm patte r n 2.11 . 3, 3.4.2.5, 5.3.6.5
labelling idealized pattern isohyets 2.11.3.2
Mekong river basin 5.3.5 . 7
observed storm pattern 2.11.2
Tennessee river basin 3.4.2.5
within- basin depth- area curves 2.11.3, 2.13.5, 5.3.5.7
(See also Depth- area and Depth-area-duration relations)
Atmospheric moisture, estimation 2.2
assumption of saturated pseudo- adiabatic atmosphere 2.2.l
maximum persisting dew points 2. 2. 5
persisting 12-hour dew points 2.2.3
precipitable water 2.2.6, 2. 3.2, Annex l
representative storm dew points 2 . 2.4
surface dew points as index 2.2 . 2
Cautionary remarks on PMP estimates
adequacy of storm sample 2.13.l
areal distribution 2.13.5
basic data deficiencies 3.5.l

comparison with extreme rainfalls 2.13.2
consistency of estimates 2.13.3
orographic regions 5.4
orographic separation method 3.3.5, 3.5.2
seasonal variation 2.13.4
statistical estimates 4.5
Convergence PMP for combining with orographic PMP:
adjustment of index map values 3. 3.4.7
combining with orographic PMP 3.3.5
construction of index map 3.3.4.6
moisture envelopes 3.3.4.l
P/M ratios 3.3.4.2
reduction for elevation, upwind barriers, basin size 3.3.4.3-3.3.4.5
Depth-area relations:
area-reduction curves 4.2.5, 5.3.6.5
derivation 2.8.2
thunderstorm 5.3.6.5
within-basin 2.11.3, 5.3.5.7
Depth-area-duration relations:
derivation 2.8.2
Hawaiian Islands PMP 5.3.2.3
Mekong river basin PMP 5.3.5.3, 5.3.5.4, 5.3.5.7
Tennessee river basin PMP 5.3.3.8, 5.3.4.1
Depth-duration relations:
derivation 2.8.2, 4.2.6

SUBJECT INDEX 183
Tennessee river basin PMP 3.4.2.4
thunderstorm PMP 5.3.6.2

Dew points:
applicability of 12-hour for all storm durations 2.3.3.
maximum persisting 12-hour l 000 mb 2.2.5
moisture index 2.2.2
persisting 12-hour 2.2.3
reduction to l 000 mb 2.2.2
reference dew point for moisture adjustment 2.6. l. l
representative storm dew point 2.2.4
Envelopment in estimating PMP 2. 8.1, 2.8. 2
Generalized estimates:
base maps 5.1. l
consistency between maps 5.1.3.2
durational and areal consistency 5.1.2
general remarks 5.1.4
Hawaiian Islands PMP 5.3.2
Mekong river basin PMP 5.3.5
non-orographic regions 5.2.1-5.2.4
orographic model 3.3.2
orographic regions 5.3
regional smoothing of isohyets 5.1.3
statistical method 4.4
Tennessee river basin PMP 5.3.3-5.3.4
thunderstorm PMP, north-western U.S . 5.3.6

Hawaiian Islands PMP 5.3.2
generalized estimates 5.3.2.3
non-orographic PMP 5.3.2.l
slope intensification 5.3 . 2.2
Isohyets, drawing of 5.1.3
Manual, purpose and scope 1.3.l, 1.3.2
Maximization:
combined sequential and spatial 2.7.4
moisture (see Moisture maximization)
sequential 2.7.2
spatial 2.7.3
wind (see Wind maximation)
Maximum possible precipitation 1.1.3
Mekong river basin PMP for 5 000 to 25 000 km2 5.3.5
adjustment of U.S. tropical storms 5.3.5.3
adjustment of Vietnam tropical storms 5.3.5.4
areal distribution 5.3.5.7
generalized estimates 5.3.5.5
mean seasonal precipitation 5.3.5.l
PMP for specific basins 5.3.5.8
time distribution 5.3.5.6
typhoon as PMP prototype 5.3.5.2
Models:
convergence 2.1.l, 2.1.2
orographic (see Orographic model) 3.2

SUBJECT INDEX 185
Modification of non- orographic PMP for orog rap hy,
Tennessee river basin:
depth - duration relation 3.4 .2.4
derivation 3.4.2 .2
geographic distr i bution 3 4.2. 5
seasonal variation 3.4 . 2.3
time distribution 3.4 . 2. 6
topographic effects 3 . 4.2 . l
(see also examples of PMP studies in Chapter 5)
Moisture maximization:
adjustment for storm relocation 2.6 . l
barrier adjustment 2 . 3.4
elevation adjustment 2.6.2
maximizing storm in place 2 . 3.4
moisture envelopes 3.3.4.l
orographic model 3.3.l.2
persisting 12-hour dew points for all storm durations 2.3.3
P/M ratios 3.3.4.2
reference dew points 2.6 . l.l
seasonal limitations 2.3.l
Orographic model:
air streamlines 3.2.3.3
computation of orographic precipitation 3.2.3.6
description 3.2 . 2
freezing level 3.2.3.4

ground profile 3.2.3.1
inflow data 3.2.3.2
multiple layer model 3.2.2.2
precipitation trajectories 3.2.2.3, 3. 2.3.5
single layer model 3.2.2.1
sources of error 3.2.3.8
test on observed storm 3.2.3
Orographic PMP
computation by orographic model 3.3.1, 3.3.2
convergence PMP for combination with 3.3 . 4
orographic influences on PMP 5.3.4.2
variations (seasona l, durational and areal) 3.3.3
Orographic separation method :
cautionary remarks 3.3.5, 3.5.2
computation of orogra phic PMP 3 . 3.1
conve r gence PMP for combination with orographic PMP 3.3.4
definition 3.1.5.1, 3.2.1
generalized estimate s of orographic PMP 3.3.2
variations in orographic PMP 3.3.3
(see Orographic model)
P/M ratios 3.3.4.2
Precipitable water 2.2.6, 2. 3.2, Annex 1
Precipitation in orographic regions:
mean annual and seasonal 3.1.3
meteorological influences 3.1.2

SUBJECT INDEX 187
orographic influences 3.1.l, 3.4.2.l
trajectories for orographic model 3.2.2.3, 3.2.3.5
Probable maximum precipitation:
accuracy of estimates 1.2.l
areal distribution (see Areal distribution of PMP)
definitions l.l.l, 1.1.2
idealized PMP storm patterns 2.11.3, 3.4.2.5, 5.3.6.5
limits, lower and upper, confidence bands 1.2
maximum possible precipitation 1.1.3
orographic regions 3.1.5
probable maximum storm 1.1.4
seasonal variation (see Seasonal variation of PMP)
time distribution (see Time distribution)
(see also Generalized estimates and Orographic PMP)
Probable maximum storm l.1.4
idealized storm pattern 2.11.3, 3.4.2.5, 5.3.6.5
Record rainfalls Annex 2
Seasonal variation of PMP:
cautionary remarks 2.13.4
daily station precipitation 2. 10.5
maximum persisting dew points 2.10.3
moisture inflow 2.10.4
observed storms 2.10.2
reason for determining 2.10.l
Tennessee river basin 3.4.2.3

thunderstorm PMP, north-western U.S. 5.3.6 . 3
weekly precipitation data 2.10.6
Sequential maximization 2.7.2
Sequential and spatial maximization combined 2.7.4
Sliding technique 2.11. 2
Slope intensification 5.3.2.2
Spatial maximization 3.7.3
Statistical estimates:
adjustment for fixed observational time intervals 4.2.4
adjustment for maximum event 4.2.2
adjustment for sample size 4.2.3
application of procedure 4.3
area-reduction curves 4.2.5
cautionary remarks 4.5
development of procedure 4.2.1-4.2.5
generalized estimates 4.4
Storm:
indicators of convergence and vertical motion 2.1.2
models, convergence 2.1.l
orographic model (see Orographic model)
probable maximum 1.1.4
Storm transposition:
adjustment factors 2.6.4.2
barrier adjustment 2.6.3, 5.3.5.4
definitions 2.5.l
SUBJECT INDEX 189
elevation adjustments 2.6 .2
exam pl e 2. 6. 4
moistu r e adjustme nt fo r r eloca t ion 2. 6. l
orogr aphic r egions 3 . 1.4
steps in t r ansposition 2. 5. 2
storm types 2.5 . 2. 2
thunderstorms 2. 6 . 2. 2
topographic controls 2.5.2.3
transposition adjustments 2. 6
transposition limits 2 . 5. 1
Tennessee river basin PMP for 250 km2 or less 5.3 . 3
adjustment for moistu r e and latitudinal gradient 5. 3. 3. 5
broad-scale topographic effects 5 . 3. 3.3
depth- duration curves for 15 km2 5.3.3.4
local topographic classification 5.3 . 3. 2
outstanding rainfalls 5.3 . 3. 1
PMP for specific basins 5.3.3 . 8
Six- hour 15 km2 PMP index map 5. 3. 3.6
time distribution of rainfall 5 . 3. 3.7
Tennessee river basin PMP for 250 to 8 000 km2 5.3 . 4
areal and time distribution 5.3.4.3
derivation of non - orographic PMP 5 . 3. 4. 1
orographic influences on PMP 5.3.4.2
PMP for specific basins 5.3.4 . 4
Thunderstorm PMP for north-western U. S. 5.3.6

depth-area relation 5.3.6.5
depth-duration relati on 5.3.6.2
elevation adjustment 5.3.6.4
general remarks 5.3.6.l
PMP for specific basins 5.3.6.7
seasonal and regiona l variations 5.3.6.3
time distribution 5.3.6 . 6
Time distribution of PMP :
based on observed storm 2.12.2
Mekong river basin 5.3.5.6
order of presentation 2.12.l
Tennessee river basin 3.4.2.6, 5.3.3.7
Undercutting 2.8.3
Wind maximization:
maximi zation ratio 2.4.4
non-orographic regions 2.4.2, 2.9.2
orographic model 3.3.l.l
wind direction 2.4.3.l
winds representative of moisture inflow 2.4.3 .2
wind speed 2.4.3.2
Within-basin depth-area curves 2.11.3, 2.13.5, 5.3.5.7

Ohr No.1 - PMP Manual - Wmo No.332 - 1973

Uploaded by

Copyright:

Available Formats

Ohr No.1 - PMP Manual - Wmo No.332 - 1973

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ohr No.1 - PMP Manual - Wmo No.332 - 1973

Uploaded by

Copyright:

Available Formats

WMO

MANUAL FOR ESTIMATION OF

I WMO - No. 332 I

..- I { ,\ \ ·. ~., l ..1 l; ( , ,(,. (

SUMMARY (Engli s h, French, Russian, Spanish) XIII

Chapter l - INTRODUCTION •••••••• .......................................... l

1.1 Definitions of probable maximum precipitation (PMP) l

1.1.l Conceptual definition ......................................... l

1.1.2 Operational definition ........................................ l

1.1.3 Maximum possible precipitation ................................ l

1.1.4 Probable maximum stor m (PMS) 2

1.2 Lower and upper limits of PMP ................................ . 2

1.2. l Accuracy of PMP estimates ..................................... 2

1.2.2 Confidence bands 3

1.3 The manual •...•.••••••.•••.•.•..•••.••.•••..•••.•.••....•..•••. 3

l. 3. 3 Use of examples from actual studies 4

Chapter 2 - ESTIMATES FOR NON- OROGRAPHIC REGIONS 7

2.1. l Convergence model ••••.•••••••••...•••••••• e ••• • •••••••• o •••••• 7

2.1. 2 Observed storm rainfall as an indicator of convergence and

2.2 Estimation of atmospheric moisture 7

2.2.l Assumption of a saturated pseudo- adiabatic atmosphere 7

2.2.2 Surface dew poi nts as a moisture index ........................ 8

?..2.3 Persisting 12-hour dew points 8

2 . 2.4 Representative persisting 12-hour 1 000 mb storm dew points • • •• 10

2. 2.5 Maximum persisting 12- hour 1 000 mb dew points 11

2 . 2.6 Precipitable water 13

2.3 Moisture maximization 14

2.3.1 Seas on al limitations 14

2.3.2 Depth of precipitable water 14

2.3.3 Applicability of persisting 12-hour dew points for all storm

2 . 3. 4 Maximization of storm in place 15

2.4.2 Use in non-orographic region s 16

2.4 . 3 Winds repre sentati ve of moisture inflow in stor ms 16

2.4.3 . 1 Wind direction 17

2.4.4 Wind maximization ratio 17

2. 5 Storm transposition ........................................... 18

2. 5.1 Definitions ... ................................................ 18

2.5.2 Steps in transposition •·•·••••••••••••••••• • •••••••••••••••••• 18

2.5.2.1 The storm ••••••••••••••• 0 ••••••••••• 0 ••••••••••••• • ••••••••••• 18

2.5.2.2 Region of influence of sto r m type 18

2.5.2. 3 Topog r aphic controls I I I I I I 9 I I I I I I I I I I I I I I I I I I 9 I 9 I I I I I I I I I I I I 19

2.6 Traosposition adjustments 20

2.6.l Moisture adjustment for relocation 20

2.6 . l.l Reference dew point for moisture adjustment 20

2.6.2 Elevation adjustments ....................................... 20

2.6.2.l General storms 22.

2.6.2.2 Local thunderstorms 22

2.6.3 Barrier adjustment •••••••••••••••••••••••••••••••••••••••••• 22

2.6.4 Example of storm transposition and maximization 22

2.6.4.l Hypothetical situation •••••••••••••••••••••••••••••••••••••• 22

2.6.4.2 Computation of adjustment factor••••••••••••••••••• ••••••••• 23

2.7 Sequential and spatial maximization 24

2.7.2 Sequential maximization 24

2.7.3 Spatial maximization •••••••••••••••••••••••••••••••••••••••• 25

2.7.4 Combined sequential and spatial maximization 28

2.8 Envelopment •......••••••••••.••••••••••••••.•••.....•.•..... 28

2.8.l Introduction ................................................ 28

2.8.2 Envelopment••••••••••••• • ••••••••••••••••••••••••••••••••••• 29

2.8.3 Undercutting ••••.•••.••••••.•.•••••.•.•••••.•••.......•..... 32