Trends in Streamflow Magnitude and Timings in Satluj River Basin

Mohammed Sharif1 David Archer2 and Ayman Talib Hamid3

Professor, Department of Civil Engineering, Jamia Millia Islamia Central University, New Delhi, India; Email: msharif@jmi.ac.in JBA Consulting Engineers and Scientists, Skipton, North Yorks, England BD23 3AE; Email: davearcher@yahoo.com

Lecturer, Department of Civil Engineering, Mosul University, Mosul, Iraq; Email: aymantalib@st.jmi.ac.in

ABSTRACT Most operational decisions for water resources infrastructure are dependent on both the timing and magnitude of flows, and therefore climate change impact assessment must consider both these characteristics. This paper examines trends in the magnitude and timings of streamflow at several stations in Satluj River basin in India. The intention is to increase the reliability of water resources systems in the basin through consideration of the observed trends. The Mann-Kendall nonparametric trend test, incorporating a correction for serial correlation and presence of ties, was used for the detection of trends. Trends in seasonal proportions of annual flow as well as actual flows have been investigated. Timings of extreme flows at several locations in the basin have also been investigated. The results reveal more trends than would be expected to occur by chance for several of the magnitude and timing measures considered in this study. The contribution of glacier melt to annual runoff at Bhakra, the major reservoir in the basin, is approximately 59% and therefore any changes in magnitude and timings of water availability at the reservoir could have significant implications for its operation. Further, climate change induced changes in the timings and magnitude of water availability at Bhakra, the major reservoir in the basin, could put the drinking water supplies of millions of people at risk. The study will contribute towards climate change modeling efforts being undertaken in a climatologically sensitive basin such as Satluj. Keywords: Climate, Satluj, Trends, Mann Kendall INTRODUCTION Climate change is considered as the greatest challenge currently facing mankind. Evidence from climate models suggests that global climate is changing in an unprecedented manner. In a developing country like India, climate change represents an additional stress on ecological and socioeconomic systems that are already facing tremendous pressures due to population growth, rapid urbanization, industrialization and economic development. With its huge and growing population, a 7500-km long densely populated and low-lying coastline, and an economy that is closely tied to its natural resource base, India is considerably vulnerable to the impacts of climate change.

This paper examines trends in the magnitude and timings of streamflow at several stations in Satluj River basin, a Himalayan tributary of the river Indus in India. The intention is to improve the reliability of water resources systems in the basin through consideration of the observed trends. Satluj is the lifeline of northern India and highly vulnerable to the impacts of global warming mainly due to significant contribution to its flows from snow and glacier melt. The contribution of glacier melt to annual runoff at Bhakra, the major reservoir in the basin, is approximately 59% (Singh and Jain, 2002), and therefore any changes in magnitude and timing of water availability at the reservoir arising from changes in temperature or precipitation could have significant implications for its operation. Several studies have examined trends and patterns in magnitude and timing of streamflow in different parts of the world. Whittfield (2001) examined temperature, precipitation and streamflow data for sites in British Columbia and the Yukon. Trends in annual streamflow volume in northern British Columbia and the Yukon have been investigated by Fleming and Clarke (2003). Derry and Wood (2005) found a decreasing trend in streamflow in the Canadian Arctic and attributed it to various large scale atmospheric phenomenon. Zhang et al. (2001), Hodgkins et al. (2003), and Hodgkins and Dudley (2006) used the centre of volume date to define the timing of runoff. Burn (2008) examined several measures of the timing of runoff for catchments in the Mackenzie River Basin in northern Canada. The investigation of changes to the timings of streamflow has received considerable attention in the western United States. Stewart et al. (2005) used three measures to examine changes in streamflow timing in snowmelt dominated watersheds in western North America. Cayan et al. (2001) estimated the onset of spring runoff by defining a pulse day. Regonda et al. (2005) showed that the shifts in the timings of CoV vary with elevation. Trends in timings of CoV have also been investigated by Barnett et al (2008), Hidalgo et al (2009), and Dery et al. (2009) among others. Although changes to the timing of streamflow in Canada and western United States have been extensively studied, such studies have not been undertaken in Satluj River basin. However, timing trends in streamflow have received limited attention in the upper Indus basin. Khattak et al. (2011) assessed monthly trends in streamflow magnitude on the Upper Indus at eight stations and found predominantly increasing trends in winter and decreasing trends in summer. Archer et al. (2010) contended that any changes in the flow regime in the Indus River Basin could have serious impact on the livelihoods not only of those engaged in the agricultural sector but for the economy as a whole. Therefore, the focus of this paper is on the analysis of trends in magnitude and timings of streamflow in Satluj River Basin, a climatologically sensitive basin in India. STUDY AREA AND DATA The Satluj River originates from Mansarowar Lake in Tibet at an elevation of about 4572 m and is a major river of the Indus system. Satluj plays a key role in the economy of northern India where two out of three persons depend upon agriculture and allied activities for their livelihood. The entire basin lies between North latitudes 30 and 33 and east longitudes 76 and 83 (Figure 1). Over 90% of the catchment lies above

1525 metres above mean sea level (masl) where majority of the precipitation is snowfall. The permanent snow line is at an elevation of 5400 m. Snow line descends to an elevation of about 2000 m during winter and crosses 5000 m altitude by the end of the summer season. About 65% of the basin is covered by snow by the end of winter, and about 15% remains under perpetual snow/glacier.

Figure 1. Location of flow gauge stations in the Satluj basin The Satluj leaves Himachal Pradesh to enter the plains of Punjab at Bhakra, where the India's highest gravity dam has been constructed. The Bhakra Dam is Asia's second highest at 225.55 m high, next to the 261 m Tehri Dam in India. The catchment area of river Satluj up to Bhakra dam is about 56,876 km2 out of which 36,900 km2 lies in Tibet and 19,975 km2 in India. The total installed hydropower capacity at Bhakra is 1325 MW - 5 x 108 MW + 5 x 157 MW Francis turbines. Figure 2 shows the monthly distribution of runoff at Bhakra where it can be seen that the contribution to annual runoff is maximum between June and September. This is both the period of peak monsoon rainfall and peak glacier and snowmelt runoff in the plains of northern India. About 83% of the total runoff occurs between April to September. 140 120 100 80 60 40 20 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Figure 2. Average monthly runoff (mm) at Bhakra

The Satluj finally drains into the Indus in Pakistan but under the Indus Waters Agreement 1961, the flow of the Satluj is allocated entirely to India. The availability of observed data in the basin is, however, far from satisfactory mainly due to difficult climatic conditions, ruggedness of the terrain and poor accessibility to the area thereby limiting comprehensive analysis of climate data in the region. Sreamflow data in the basin are available at five stations namely, Bhakra, Kahu, Kasol, Suni, and Rampur. Many parametric and non-parametric methods have been applied for detection of trends (Kundzewicz & Robson, 2004; Zhang et al., 2006). Parametric tests are more powerful than the non-parametric ones, but the assumption regarding the normality of data must be satisfied. Hydro-meteorological time series are often characterized by data that is not normally distributed, and therefore nonparametric tests are considered more robust compared to their parametric counterparts (Hess et al. 2001). One of the most widely used non-parametric tests for detecting a trend in the hydro-meteorological time series is the Mann-Kendall test (Mann, 1945; Kendall, 1975). A major advantage of Mann-Kendall test is that it allows missing data and can tolerate outliers. Several researchers have employed Mann-Kendall test to identify trends in the hydro-meteorological variables (Burn, 1994; Burn et al. 2010). Trend evaluation using Mann-Kendall test relies on two important statistical metrics: the trend significance level or the p-value, and the trend slope . The p-value is an indicator of the trend strength whereas provides the rate of change in the variable allowing determination of the total change during the analysis period. The presence of a serial correlation in a data set can affect the outcome of the Mann-Kendall test; the version of the trend test used herein incorporates a correction, developed by Yue et al. (2002). METHODOLOGY A total of seven variables have been computed using the available streamflow data at five gauging stations. The first variable is the monthly and annual flows whereas the second variable is the monthly proportions of annual flows computed for each of the twelve months. Trends in proportional monthly flows have been investigated mainly because this was the preferred method used by researchers in the western United States. Four measures of extreme flows were extracted from the daily streamflow data for each of the five gauging stations. The first two extreme flow measures are the annual maximum daily flow event magnitude (HFM) and the corresponding date of occurrence (HFT). The next two extreme flow measures are the annual minimum flow event magnitude (LFM) and the corresponding date of occurrence (LFT). The final variable is the centre of volume date (CoV) which may be defined as the date by which 50% of the annual flow has passed through the gauging station using January 1 as the starting date (Regonda et al. 2005). The trend analysis has been carried out on the all seven variables using the MannKendall nonparametric test. The version of the Mann-Kendall test used here incorporates corrections for serial correlation and for the presence of ties in the data (Salas 1993). It is critical to apply a correction for ties, particularly for the timing data as it is expected to some tied data points. The results of trend analysis for the monthly flows and annual flows are presented in Table 1 whereas the results of trend analysis for

monthly proportions of annual flows are presented in Table 2. The results of trends in high and low flow magnitude and timing measures are presented in Table 3. In all cases, trends with p values less than 0.1 have been considered statistically significant. RESULTS The results of trends analysis for monthly and annual flows are shown in Table 1. It can be seen from Table 1 that the monthly flows at Bhakra exhibit increasing trends in all the twelve months out of which nine months showed statistically significant trends. The annual flow at Bhakra exhibited an increasing trend with a p-value of 0.011. No decreasing trend was observed at Bhakra. At Kahu, the decreasing trends outnumber the increasing trends but one of the trends was statistically significant. Table 1 Trend in annual and monthly flow
Bhakra Slope JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC ANN No. + No. 31.56 30.11 56.65 100.7 278.9 34.50 4.48 111.90 205.8 109.8 57.80 37.93 1093 13 0 p 0.006 0.028 0.017 0.016 0.001 0.742 0.965 0.324 0.006 0.001 0.001 0.008 0.011 Kahu Slope -0.56 -0.36 -0.76 -0.50 -0.64 -0.52 0.33 -0.41 2.80 -0.04 -0.31 -0.39 -0.98 2 11 0 1 p 0.313 0.587 0.127 0.165 0.046 0.524 0.952 0.952 0.605 0.924 0.552 0.458 0.95 Kasol Slope 0.06 -0.10 -0.10 -0.2 2.25 -5.82 -6.44 -4.7 -1.32 0.06 -0.04 0.19 -13.5 4 9 0 3 p 0.684 0.677 0.806 0.719 0.135 0.047 0.036 0.025 0.233 0.895 0.88 0.41 0.15 Rampur Slope -0.04 -0.16 -0.12 -0.84 0.90 -5.10 -5.97 -5.53 -1.86 -0.80 -0.18 0.01 -16.86 2 11 0 7 p 0.866 0.522 0.36 0.047 0.45 0.03 0.013 0.004 0.023 0.026 0.483 0.999 0.028 Suni Slope -0.05 -0.12 -0.51 -1.00 0.84 -6.59 -7.83 -5.91 -2.34 -0.58 -0.30 -0.04 -20.79 1 12 0 8 P 0.696 0.398 0.018 0.039 0.756 0.019 0.003 0.003 0.046 0.055 0.155 0.943 0.013

No. Sig+ 10 No. Sig- 0

Notes: 1. Bold values indicate statistically significant trends at 10% significance level; 2. Slope is in m3/s/year

At Kasol, there are nine months that show decreasing trends out of which three are statistically significant with p-values ranging from 0.025 to 0.047. There are four increasing trends at Kasol but they are not statistically significant. At Rampur, there are eleven decreasing trends out of which seven are statistically significant. There are twelve decreasing trends at Suni with eight being statistically significant.

Table 2 shows the results of trend analysis conducted on the monthly proportions of annual flows. At Bhakra, the monthly proportions exhibit increasing trends in nine months with five months showing statistically significant trends. The trends in monthly proportions of annual flows are in line with the trends in actual monthly flows. At Kahu, the trends in monthly proportions are similar to the trends in actual monthly flows. However, none of the trends at Kahu are statistically significant. At Kasol, there is reversal of trends from actual flows in monthly proportions of annual flows. The increasing trends in monthly proportions of annual flows outnumber the decreasing trends whereas in the case of monthly flows decreasing trends outnumber increasing trends as shown in Table 1. The remaining two stations, namely Suni and Rampur exhibit reversal of trends in annual proportions of monthly flows similar to that seen at Kasol. Table 2 Trend in monthly proportions of annual flow
Bhakra Slope JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC No. + No. 0.01 0.01 0.01 0.01 0.01 -0.01 -0.01 -0.01 0.01 0.01 0.01 0.01 9 3 p 0.434 0.328 0.207 0.046 0.001 0.06 0.01 0.1 0.194 0.001 0.004 0.091 Kahu Slope -0.01 -0.01 -0.01 -0.01 -0.01 -0.01 -0.01 0.01 0.01 -0.01 0.01 -0.01 3 9 0 0 p 0.398 0.453 0.428 0.27 0.138 0.313 0.816 0.993 0.224 0.776 0.73 0.541 Kasol Slope 0.01 0.01 0.01 0.01 0.01 -0.01 -0.01 -0.01 0.01 0.01 0.01 0.01 9 3 3 1 p 0.248 0.415 0.46 0.454 0.011 0.208 0.034 0.237 0.887 0.097 0.147 0.013 Rampur Slope 0.01 0.01 0.01 0.01 0.01 -0.01 -0.01 -0.01 0.01 0.01 0.01 0.01 9 3 3 0 p 0.116 0.124 0.163 0.689 0.09 0.135 0.197 0.307 0.979 0.929 0.083 0.003 Suni Slope 0.01 0.01 0.01 -0.01 0.01 -0.01 -0.01 -0.01 0.01 0.01 0.01 0.01 8 4 4 2 P 0.043 0.108 0.625 0.499 0.031 0.068 0.08 0.54 0.979 0.157 0.042 0.002

No. Sig+ 5 No. Sig - 2

Notes: 1. Bold values indicate statistically significant trends at 10% significance level

The results of trend analysis conducted on the high and low flow magnitude and timings measures, and CoV are presented in Table 3. The HFM shows increasing trend at Bhakra and Kahu, and a decreasing trend at Kasol, Rampur, and Suni. However, trends are not statistically significant except at Rampur where the p-value of trend significance is 0.026 thus indicating a strong trend. At all the five stations, HFT exhibits an increasing trend thus indicating a shift to a later date in the timings of occurrence of peak flows. The trend was found to be statistically significant at Suni only. At Bhakra,

LFM shows a statistically significant increasing trend with a p- value of 0.007. There is an increasing trend in LFM at Kasol and Rampur whereas a decreasing trend was observed at Kahu and Suni. At Bhakra, there is a decreasing trend in LFT whereas the other four stations exhibited increasing trends. None of the stations, however, exhibited statistically significant trends in LFT, either increasing or decreasing. Interestingly, four of the five stations including Bhakra exhibited an increasing trend in CoV timings. At Kasol, CoV exhibited a decreasing trend. None of the trends in CoV were statistically significant. Table 3 Trend in flow magnitude and timing measures
Bhakra Kahu Kasol Rampur Suni

Slope
HFM HFT LFM LFT CoV 28.89 0.238 29.67 -0.33 0.01

p
0.880 0.346 0.007 0.445 0.721

Slope
107.1 0.10 -0.15 0.04 0.10

p
0.211 0.737 0.474 0.776 0.611

Slope
-3.46 0.30 0.09 0.01 -0.05

p
0.437 0.248 0.544 0.865 0.314

Slope
-10.28 0.01 0.08 0.01 0.01

p
0.026 1.000 0.600 0.979 0.537

Slope
-6.17 0.55 -0.08 0.32 0.01

p
0.133 0.036 0.582 0.223 0.796

Notes: 1. Bold values indicate statistically significant trends at 10% significance level

DISCUSSION AND CONCLUSION The intention behind the analysis conducted in this research was to assist, through interpretation of trends, in the development of adaptation strategies to counteract the adverse impacts of climate change in Satluj river Basin. Results of trend analysis of monthly and annual flow as well as annual proportions of monthly flows at five gauging stations in the basin clearly indicated there were a large number of statistically significant trends. It is important to note that the monthly proportions of annual flow have shown a reversal of trends when compared to the monthly flows at three out of five stations. Analysis of trends in flow and timing measures has clearly shown that the changes are occurring in the hydrological regime for the sites examined, both in terms of peak magnitudes albeit weak indication, and the timings of peak events (generally occurring later). Changes are occurring in the low flow regime as well with a strong indication of increase in LFM events at Bhakra. The trends in LFT have been found to be generally week at all the sites. The CoV timings exhibited no definite trends in either direction. Most of the analyses have led to the conclusion that there were a significant number of significant trends in the variables. It is concluded that given the potential changes in the flow regime, it is critical to revise reservoir operating policy at Bhakra to be able to cope with the changing climatic conditions in the basin. The direction of the trends is inconsistent between the stations, notably between the increasing trend in annual flow at Bhakra and the decreasing trend upstream. The possible reasons for this could be a big input from the lower part of the catchment to Bhakra or a poor quality of the flow data. The version of the Mann-Kendall test applied herein alleviates some of the problems of trend analysis of historical data with the intent of establishing a research agenda that

can further address issues of non-stationarity in the flow records. A well known limitation of trend analysis of historical data is that the analysis is, by its nature, retrospective. A purely retrospective analysis of the available data may not be a good indicator of future conditions. Any discernible trends can be attributed only partially to the impacts of changed climate in the basin. To obtain a more comprehensive view of the present and plausible future conditions in the basin requires combining trend analysis of historical data with analysis of climate data obtained from modelling climate change projections using output from regional climate models REFERENCES Archer, D. R., Forsythe, N., Fowler, H.J. and Shah, S.M. (2010). Sustainability of water resources management in the Indus Basin under changing climatic and socio economic conditions, Hydrol. Earth Syst. Sci., 14, 113. Barnett, T. P., Pierce, D. W., Hidalgo, H. G. Bonfils, C., Santer, B. D., Das, T., Bala, G., Wood, A. W., Nozawa, T., Mirin, A. A., Cayan, D. R., and Dettinger M. D. (2008). Human-Induced Changes in the Hydrology of the Western United States, Science 319, 1080; DOI: 10.1126/science.1152538. Burn, D. H. (1994). Hydrological effects of climatic change in west-central Canada. Journal of Hydrology 160, 53-70. Burn, D. H. (2008). Climatic influences on streamflow timing in the headwaters of the Mackenzie River Basin. J Hydrol 352: 225-238. Burn, D. H., Sharif, M., and Zhang, K. (2010). Detection of trends in hydrological extremes for Canadian watersheds. Hydrological processes, Doi:10.1002/hyp.7625. Cayan, D. R, Kammerdiener, S.A, Dettinger, M.D, Caprio, J.M, and Peterson, D.H. (2001). Changes in the onset of spring in the western United States. Bulletin of the American Meteorological Society 82(3): 399415. Dry, S.J., Wood, E.F. (2005). Decreasing river discharge in northern Canada. Geophysical Research Letters 32: L10401. DOI:10.1029/2005GL0 22845. Dry, S.J., Stahl,K., Moore,, R. D., Whitfield, P. H., Menounos, B. and Burford, J.E. (2009). Detection of runoff timing changes in pluvial, nival, and glacial rivers of western Canada. Water Resources Research, 45, W04426, doi:10.1029/2008WR006975. Fleming, S. W., and Clarke, GKC. (2003). Glacial control of water resources and related environmental responses to climatic warming: empirical analysis using historical streamflow data from Northwestern Canada. Canadian Water Resources Journal 28(1): 6986. Hess, A., Lyer H., and Malm, W. (2001). Linear trend analysis: a comparison of methods. Atmos Environ 35: 5211-5222. Hidalgo, H. G., et al. (2009). Detection and attribution of streamflow timing changes to climate change in the western United States, J. Clim., 22, 38383855.

Hodgkins, G. A., and Dudley, R. W. (2006). Change in the timings of winter-spring streamflows in eastern North America, 1913-2002. Geophysical Research Letters 33: L06402. DOI:10.1029/2005GL025593. Hodgkins, G. A., Dudley, R. W., and Huntington, T.G. (2003). Changes in the timings of high river flows in New England over the 20th century. Journal of Hydrology 278: 244252. Kendall, M. G. (1975). Rank Correlation Measures, Charles Griffin, London, UK. Khattak, M. S, Babel, M. S., and Sharif, M. (2011). Hydrometeorological trends in the upper Indus River Basin in Pakistan, Climate Research, 46: 103-119. Kundzewicz, Z. W., and Robson, A. J. (2004). Change detection in hydrological records-a review of the methodology. Hydrological Sciences, 49 (1):7-19 Mann, H. B. (1945). Non-parametric tests against trend, Econometrica 13: 245-259. Regonda, S. K., Rajagopalan, B., Clark, M. and Pitlick, J. (2005). Seasonal cycle shifts in hydroclimatology over the western United States. J. Climate, 18, 372384. Salas, J. D. (1993). Analysis and modelling hydrologic time series, In Handbook of Hydrology, Maidment, D.R. (ed.), McGraw Hill, New York, 19.1 19.72. Singh, P., and Jain, S. K. (2002). Snow and glacier melt in the Satluj River at Bhakra Dam in the western Himalayan region. Hydrological Science Journal, 47(1), 93-106. Stewart, I. T., Cayan, D. R., and Dettinger, M. D. (2005). Changes toward earlier streamflow timing across western North America. Journal of Climate 18: 11361155. Whitfield, P. H. 2001. Linked hydrologic and climatic variations in British Columbia and Yukon. Environmental Monitoring and Assessment 67(12): 217238. Yue, S., Pilon, P. J., Phinney, B. and Cavadias, G. (2002). The influence of autocorrelation on the ability to detect trend in hydrological series. Hydrological Processes 16(9): 1807-1829. Zhang. Q, Liu. C, Xu C. Y., Xu Y. P., and Jiang, T. (2006). Observed trends of water level and streamflow during past 100 years in the Yangtze River basin, China. J. Hydrol 324 (1-4): 255-265. Zhang, X., Harvey, K. D., Hogg, W. D., and Yuzyk, T. R. (2001). Trends in Canadian streamflow. Water Resouces Research 37 (4), 987-998. .