Satellite-based drought monitoring using optimal indices for diverse climates and land types

Drought is considered one of the most destructive natural disasters, and many areas are experiencing water scarcity. Expanding knowledge of this phenomenon is a prerequisite for developing drought monitoring and forecasting tools. To this end, various indices are available for studying drought in different environments using field and remote sensing data. This study applies satellite-based indices for monitoring drought in different land cover, landforms, and climate classes. The in-situ standardized precipitation index (SPI) with a three-month time scale was applied to evaluate the performance of 13 remote sensing indices and parameters. The results indicated that the indices based on actual evapotranspiration, precipitation, and soil moisture, respectively, performed best in different parts of the basin. After additional analysis, the evapotranspiration condition index (ETCI), derived from actual evapotranspiration data, was deemed the optimal metric. The accuracy assessment results indicated that the correlation between the ETCI and the three-month SPI was 0.655, which was slightly higher than the actual evapotranspiration (0.637), and that the root-mean-squared error (RMSE) decreased from 0.71 to 0.65, indicating the best performance among the indices evaluated in the study area. Moreover, the drought map of the region was developed using the optimal indices, including the ETCI, the precipitation condition index (PCI), and the random forest (RF) algorithm. According to the results of the accuracy evaluation, the correlation between the estimated model and the observed three-month SPI values in 2017 was 0.72, with an RMSE of 0.60.