A new approach for drought forecasting using wavelet-ANN model and satellite images

Forecasting drought is a challenging endeavor due to various underlying factors and mechanisms. Thus, the need for robust and precise forecasting models is paramount. In this study, a method that utilizes the wavelet neural network and spatial proximity data derived from satellite images to enhance the accuracy of drought forecasts is presented. This technique applies satellite-based precipitation and evapotranspiration data to calculate drought indices. It then uses the wavelet neural network approach to forecast drought intensity in different months of the subsequent year. To better discern random fluctuations from actual drought signals and enhance forecast accuracy, we utilize spatial proximity data from satellite images to forecast drought at the East Isfahan climate station. Our findings validate the capability of the wavelet neural network approach to forecast drought with a reasonable degree of accuracy. Also, leveraging neighboring data can potentially improve forecasting precision, as evidenced by a correlation of 0.675 between the target and predicted values.