Investigating the behavior of statistical indices for performance assessment of a reservoir

Commonly employed indices to assess the performance of a reservoir include reliability, resilience, and vulnerability. Depending upon the way in which resilience and vulnerability are computed, different values and behavior of these are obtained. This study has investigated the behavior of these indices using the data of an Indian reservoir. Synthetic data were generated for 1500 years employing long-memory and short-memory models. Monte Carlo simulation was carried out and the behavior of a number of performance measures was investigated. Keeping in view the sensitivity of measures and their monotonic behavior, time and volume reliabilities along with vulnerability were found to give suitable information about the performance of the reservoir. A reservoir sustainability index computed using time reliability and maximum vulnerability is proposed. Further, in semi-arid regions, evaporation losses are high and if these are ignored, one may overestimate the reliability of the reservoir by up to 10%.