Evaluating Performance and Applicability of Several Drought Indices in Arid Regions

Abstract

Drought is a complex phenomenon that strong indices should be used to quantifying it. Reconnaissance drought index (RDI) is very strong index that extensively used in drought investigating researches. Major problem to calculate this index especially in arid and semi-arid regions at undeveloped countries is lack of data for calculating evapotranspiration. This study investigated this problem’s solution via simulating monthly RDI using other indices (PN, DI, SPI, CZI, MCZI and Z-Score) that in order to calculate these indices, only precipitation data is used. To simulate RDI, Generalized Estimating Equations (GEE) was used. For validation of estimating equations, different indices of goodness of fit were used (NSE, RMSE, MAE, R2, QIC and QICC). Results of this study indicated that the SPI, CZI, MCZI and DI indices had the most appropriateness for simulating RDI. When the SPI index (the best index for simulating RDI) was used to simulate the RDI, according to the results of T-Test, the observed and simulated data series hadn’t significantly difference (P value >0.05) in all stations. The average values of NSE, R2, RMSE, MAE, QIC and QICC obtained 0.976, 0.139, 0.088, 0.976, 21.24 and 17.82 sequential.