Hydrological drought early warning based on rainfall threshold

Hydrological drought early warning systems could greatly improve the efficiency of water resources allocation, particularly in semiarid areas where extensive storage systems are constructed. On the other hand, warning systems that rely on meteorological thresholds (triggers) are relatively easy to use in real-time conditions compared with complicated fully online forecast models. In this study, probabilistic rainfall thresholds triggering the hydrological droughts were determined on monthly basis. The case study region is Urmia Lake basin, holding one of the largest saline lakes in the world that is currently under the drying threat due to prolonged droughts and excessive surface water use. In the first step, the time series of total inflow to the Urmia Lake was transformed to a standardized hydrological drought index. A simple rainfall-drought index model was also developed at basin scale. Then, in an inverse modeling framework, rainfall threshold values causing different hydrological drought severity levels under several occurrence probabilities were determined through repetitive application of rainfall-drought index model. Derived rainfall threshold values were evaluated in a forecast mode against both historical and simulated data using the critical success index. Results revealed that the proposed approach was simple to use and sufficiently accurate in forecast of hydrological droughts, while mean probable threshold curve offered the best performance.