Prediction of reservoir storage anomalies in India

Abstract

Reservoirs provide water for irrigation and hydropower and protect downstream regions from flooding. Reservoir storage is affected by drought, which hampers the ability to provide water for irrigation and hydropower. Despite the need for reservoir storage prediction for planning and decision‐making, a reservoir storage forecast system at 1‐ to 3‐month lead has been lacking for major reservoirs in India. Here we evaluate the potential of observed accumulated precipitation, standardized precipitation index (SPI), standardized precipitation evapotranspiration index (SPEI), standardized streamflow index (SSI), and observed reservoir storage to provide reservoir storage anomaly forecast at 1‐ to 3‐month lead for the dry season (February to May) using a statistical approach. We find that accumulated precipitation for 3–11 months is strongly associated with the monthly reservoir storage anomalies in India. Moreover, accumulated precipitation and observed reservoir storage provide a reasonable (R2 = 0.7) forecast skill for reservoir storage anomalies at 1‐ to 3‐month lead. Similarly, SPI and SPEI can be used to predict reservoir storage anomalies at 1‐ to 3‐month lead in India. Since the prediction skill from SPI and SPEI is similar, reservoir storage at monthly timescale is largely affected by accumulated precipitation instead of variability in air temperature. We find that the forecast skill for the 1‐month lead from SPI, SPEI, and SSI is similar for Indira Sagar (Narmada) and Minimata (Mahanadi) reservoirs; however, prediction skill for reservoir storage anomalies improves substantially using SSI for 2‐ to 3‐month lead. Reservoir storage anomalies forecast at 1‐to 3‐month lead can be valuable for water management‐related decision‐making and planning during the dry season in India.