Adaptive neuro-fuzzy inference system for drought modeling

Abstract

. Drought exerts a significant impact on both the environment and agricultural sectors, particularly in farming. Addressing the impact of drough is essential for all stackholders who depends on natural water sources.  Adaptive Neuro-Fuzzy Inference System (ANFIS), one of the hybrid artificial neural networks, is primarily used in this study to model the drought. The study utilizes the monthly precipitation data spanning for last 39 years for the Coimbatore district. Initial steps involve the estimating of Standardized Precipitation Index (SPI) values at a 3-month scale using monthly precipitation values, considering the impact of North-East Monsoon over the district. Several ANFIS forecasting models are developed using the mean precipitation value of North-East Monsoon season and the computed values of SPI. The evaluation of these models incorporates several error metrics like Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R²), allowing for a comprehensive comparison between the projected ANFIS model and observed values. The model which exhibits the lowest RMSE and MAE, coupled with a high R², are considered as robust fit to the data.