Improved Implicit Stochastic Optimization technique for multireservoir water systems under drought conditions

Drought is a creeping phenomenon, making its onset and end difficult to determine. Damages from droughts can exceed those resulting from any other natural hazard, although it is difficult to assign a monetary value to them. In the Mediterranean area a severe drought period occurred over the years 2000-2002 and economic losses from that drought exceeded 250 million euros in Sardinia (Italy) (source: ENAS Regional Water Authority). Currently, technological developments and environmental modelling tools are improving our ability to more effectively manage water supply systems. Models can provide decision makers with better and more timely data and information. In this paper a methodology is proposed to support water decisions by selecting and evaluating reservoir operating rules based on hydrological scenarios generated from historical records. The methodology includes optimization and simulation tools. In particular, the paper presents some improvements to the traditional Implicit Stochastic Optimization (ISO) approach that overcome some severe limitations affecting previous works. Thanks to the collaboration with Regional Water Authorities in Southern Italy, the proposed methodology has been tested in the Agri-Sinni water system. Specifically this investigation focuses on: 1) Defining the reservoir operating rules based on correlations between releases, storages and inflows in a risk management approach of drought events; 2) Analyzing the significant reduction of computational time the user can get by applying the ISO technique in a GRID computing approach.