A Comparative Study of Artificial Neural Networks and Multiple Linear Regression for Predicting Average Monthly Rainfall in Northern Cyprus

Accurately forecasting rainfall is crucial in the hydrological cycle yet remains challenging due to its unpredictable and uncertain nature. Despite efforts, an ideal model for rainfall prediction is still elusive, necessitating ongoing research to improve our understanding and predictive capabilities. Therefore, Nonlinear Autoregressive with exogenous inputs (NARX), Layer Recurrent neural network (LRNN), and Elman recurrent neural network (ERNN) are used for predicting the monthly rainfall in Northern Cyprus. Besides, Multiple Linear Regression (MLR) is proposed to predict rainfall to evaluate the accuracy of developed models. To this aim, meteorological data from seven stations across Northern Cyprus were gathered between 2011 and 2017. Based on the results, it has been observed that the NARX model demonstrates superior statistical performance in terms of predicting accuracy compared to all models considered. The artificial neural network models show potential for application in similar studies focused on predicting monthly rainfall.