Drought prediction over the East Asian monsoon region using the adaptive neuro‐fuzzy inference system and the global sea surface temperature anomalies

Abstract

ABSTRACTPrediction of droughts has a great importance in the management and planning of water resources. This study developed an adaptive neuro‐fuzzy inference system (ANFIS) based model for prediction of droughts, and evaluated its applicability in the seven homogeneous rainfall zones of the East Asian monsoon region (20°–50°N, 103°–149°E). Standardized Precipitation Index (SPI) was used to characterize the drought events. SPI series were computed for each zone using a 30‐year (1978–2007) gridded rainfall dataset (0.5° grid resolution) at the corresponding grid points. The influence of sea surface temperature anomalies (SSTA) on droughts was assessed using a lagged‐correlation between global gridded SSTA (0.2° grid resolution) and the SPI of each zone. SSTA were used as a potential predictor variable based on the premise that the land‐sea thermal contrast is a major driver of the monsoon. The model was trained and validated using a 25‐year (1978–2002) dataset, with different configurations to obtain the optimum model structure and a set of suitable predictors. The performance of the model was demonstrated by comparing the model simulated results with the observed drought index and drought categories using a 5‐year (2003–2007) independent checking dataset. The model predicted the drought categories accurately for 50 to 70% cases in checking period for different zones. The results showed the viability of the proposed model for drought prediction with substantial enhancement in accuracy when past SSTA were used as a predictor compared with the use of only past SPI data.