Assessment of Severity of Droughts Using Geostatistics Method(Case Study: Southern Iran)

Abstract

Drought monitoring is a fundamental component of drought risk management. It is normally performed usingvarious drought indices that are effectively continuous functions of rainfall and other hydrometeorological variables.In many instances, drought indices are used for monitoring purposes. Geostatistical methods allow the interpolationof spatially referenced data and the prediction of values for arbitrary points in the area of interest. In this research,several geostatistical methods, including ordinary kriging (OK), indicator kriging (IK), residual kriging (RK),probability kriging (Pk), simple kriging (SK), universal kriging (UK), and inverse distance weighted (IDW) methodswere assessed for the derivation of maps of drought indices at 12 climatic stations in southern Iran. Data regardingmonthly rainfall, temperature, wind, relative humidity, and sunshine of three periods (1985, 1995, and 2005) weretaken from 12 meteorological synoptic stations and distributed areas. Based on the used error criteria, krigingmethods were used for spatial analysis of the drought indexes and were selected as the best method. Results alsoshowed that the lowest error (RMSE) is related to the kriging method. The results indicated that IK with treefrequency is more appropriate for the spatial analysis of the RDI index, and the Pk and SK methods are moreappropriate for the spatial analysis of the SPI index. The kriging methods mean errors (RMSE) selected years for RDIand SPI index respectively are 0.85 and 0.84. In several cases, the “moderately dry” class received a more criticalvalue by RDI. The results showed that by utilizing the ET0, the RDI can be very sensitive to climatic variability.