Articles | Volume 17, issue 6
Hydrol. Earth Syst. Sci., 17, 2131–2146, 2013
Hydrol. Earth Syst. Sci., 17, 2131–2146, 2013

Research article 05 Jun 2013

Research article | 05 Jun 2013

Predictability of Western Himalayan river flow: melt seasonal inflow into Bhakra Reservoir in northern India

I. Pal1,2, U. Lall3, A. W. Robertson2, M. A. Cane4, and R. Bansal5 I. Pal et al.
  • 1Department of Civil Engineering, University of Colorado, Denver, CO 80204, USA
  • 2International Research Institute for Climate and Society, The Earth Institute at Columbia University, Palisades, NY 10964, USA
  • 3Department of Earth and Environmental Engineering, Columbia University, New York, NY 10027, USA
  • 4Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY 10964, USA
  • 5Bhakra Beas Management Board, Punjab, India

Abstract. Snowmelt-dominated streamflow of the Western Himalayan rivers is an important water resource during the dry pre-monsoon spring months to meet the irrigation and hydropower needs in northern India. Here we study the seasonal prediction of melt-dominated total inflow into the Bhakra Dam in northern India based on statistical relationships with meteorological variables during the preceding winter. Total inflow into the Bhakra Dam includes the Satluj River flow together with a flow diversion from its tributary, the Beas River. Both are tributaries of the Indus River that originate from the Western Himalayas, which is an under-studied region. Average measured winter snow volume at the upper-elevation stations and corresponding lower-elevation rainfall and temperature of the Satluj River basin were considered as empirical predictors. Akaike information criteria (AIC) and Bayesian information criteria (BIC) were used to select the best subset of inputs from all the possible combinations of predictors for a multiple linear regression framework. To test for potential issues arising due to multicollinearity of the predictor variables, cross-validated prediction skills of the best subset were also compared with the prediction skills of principal component regression (PCR) and partial least squares regression (PLSR) techniques, which yielded broadly similar results. As a whole, the forecasts of the melt season at the end of winter and as the melt season commences were shown to have potential skill for guiding the development of stochastic optimization models to manage the trade-off between irrigation and hydropower releases versus flood control during the annual fill cycle of the Bhakra Reservoir, a major energy and irrigation source in the region.