Drought waring model based on combination of current and future drought indexes-A case study for shihmen reservoir

This study aims to develop a drought early warning model based on the standardized drought indices (SDIs) for the water resources system of Shihmen Reservoir in northern Taiwan. Different hydrological variables (e.g., rainfall, reservoir inflow and storage) were used as the basis of SDIs to characterize and monitor drought condition for water resources system. The SDIs, including Standardized Precipitation Index (SPI), Standardized Streamflow Index (SSI), and Standardized Reservoir Storage Index (SRSI) calculated by the historical rainfalls, reservoir inflows and storages, respectively, were analyzed. It is found that the value of SDIs can clearly reflect the situations of historical drought events. Further, the discriminant analysis was adopted to establish the drought early warning model based on the relationship among current SDIs, future SDIs and future regime lights. In the drought early warning model, the best combination of drought warning indices include current SDIs (i.e., SPI1, SRSI1) and future SDIs (i.e., SPI1 and SPI3) for predicting the future regime lights. The results show that adding future SDIs can improve the prediction accuracy of future regime lights. For both the calibration period (1958-1997) and the verification period (1998-2017), the drought early warning model reveals satisfactory performance for predicting future regime lights. The prediction performances during the dry season are better than those during the wet season.