Development of a PSO-Optimized Pythagorean Hesitant Fuzzy ANFIS Model for Drought Prediction in Istanbul

Abstract

Drought is a significant natural disaster that has serious impacts on agriculture, water resource management, and ecosystem health. Therefore, the early prediction of drought is of great importance for implementing timely and effective mitigation measures. In this study, for short-term meteorological drought prediction, an Adaptive Neuro-Pythagorean Hesitant Fuzzy Inference System optimized by Particle Swarm Optimization (ANPHFIS-PSO) method is proposed. The proposed model aims to provide high accuracy in Standardized Precipitation Index (SPI)-1 predictions by effectively handling the nonlinear relationships and uncertainties present in meteorological data. The performance of the model was compared with that of the Multilayer Perceptron Artificial Neural Network (MLP-ANN), Adaptive Neuro-Fuzzy Inference System (ANFIS) optimized via Grid Search (ANFIS-GS), ANFIS optimized by Particle Swarm Optimization (ANFIS-PSO), Long Short-Term Memory network (LSTM) and ANPHFIS optimized by Grid Search (ANPHFIS-GS). For model evaluation, Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Coefficient of Determination (R2) Kling Gupta Efficiency (KGE) and Bias Factor (BF) were used as performance metrics. The obtained results demonstrate that the ANPHFIS-PSO model yields the lowest MSE, RMSE, and MAE values, along with the highest R2 and competitive KGE and BF scores. These findings confirm that the ANPHFIS-PSO model achieves superior predictive performance compared to the other methods.