Drought Potential Mapping using Remote Sensing Indices and Artificial Neural Networks (Case study: Kermanshah Province, Iran)

Drought is a continuous period of lack of rainfall that causes damage and reduced yields in crops and has a direct impact on the quality and quantity of water and agricultural resources in the region. Remote sensing technology is Geographic Information Systems (GIS) are useful tools for processing and interpretation of spatial data and can be used for drought monitoring. The purpose of this study is to determine the most appropriate remote sensing indicators and to present a composite drought index based on the intelligent artificial neural network method. Based on the results, the best remote sensing indicators to determine the risk of drought in the region are vegetation indices, rainfall, and land surface temperature. The results were evaluated based on standardized precipitation index (SPI) values ​​obtained from meteorological stations. Accordingly, the accuracy of the results of the multivariate regression method was R2 = 0.62 and the multilayer perceptron neural network was R2 = 0.91. Therefore, the multilayer perceptron neural network method is more robust than the multivariate regression method to create a more accurate hybrid drought index. According to the results, there is a potential for monthly drought in most areas of Kermanshah province. Moreover, the annual drought potential is observed in the eastern regions of the province. Islamabad, Songor, and Harsin are cities with a low risk of drought.