Forecasting Different Types of Droughts Simultaneously Using Multivariate Standardized Precipitation Index (MSPI), MLP Neural Network, and Imperialistic Competitive Algorithm (ICA)

Precipitation deficit causes meteorological drought, and its continuation appears as other different types of droughts including hydrological, agricultural, economic, social droughts. Multivariate Standardized Index (MSPI) can show the drought status from perspective simultaneously. Forecasting multivariate provide good information about future a region will be applicable for planners water divisions. In this study, MLP model hybrid form with Imperialistic Competitive Algorithm (MLP-ICA) have been investigated first time in studies. For purpose, two semi-arid stations western Iran were selected, their precipitation data provided Iranian Meteorological Organization (IRIMO), during period 1988–2017. MSPI was calculated 5-time windows MSPI3–6 (drought perspectives soil moisture surface hydrology simultaneously), MSPI6–12 (hydrological agricultural MSPI3–12 (soil moisture, hydrology, MSPI12–24 agriculture groundwater MSPI24–48 (socio-economical droughts). The results showed acceptable performances forecasting both stations, larger (MSPI12–24 MSPI24–48) had better predictions than smaller ones (MSPI3–6, MSPI6–12, MSPI3–12). Generally, it reported that, by decreasing size window, gradual changes index give way to sudden jumps. This weaker autocorrelation consequently predictions, e.g., simultaneously (MSPI3–6). MLP-ICA shows stronger prediction simple all comparisons. ICA optimizer could averagely improve MLP’s accuracy 28.5%, which is significant improvement. According evaluations (RMSE = 0.20; MAE 0.15; R 0.95), are hopeful simultaneous tested similar areas.