analysis of groundwater level in kouhdasht plain of lorestan using metaheuristic models

Prediction of groundwater levels (gwl) using machine learning techniques has gained substantial attention over the past few decades. several researchers have reported the advances in this field and provided clear understanding of the state-of-the-art machine learning models implemented for gwl modeling. in this research, a new hybrid model based on artificial neural network approaches has been developed to estimate the groundwater level. for this purpose, three optimization algorithms, including wavelet, innovative gunner, and black widow spider, were employed for modeling the groundwater level. the study utilized statistical data from four piezometers in kouhdasht plain of lorestan province, iran, as a case study over five combined scenarios of input parameters from 2002 to 2022. to evaluate the performance of the models, correlation coefficient, root mean square error, mean absolute error, and nash-sutcliffe efficiency coefficient were used as assessment criteria. additionally, time series charts and box plots were employed to analyze the model results. the findings indicated that the combined scenarios in the models improved the model’s performance. moreover, the evaluation results showed that the wavelet-support vector regression model exhibited higher accuracy than the other models across all the examined piezometric wells. overall, the results demonstrated that the use of intelligent models based on the hybrid approach of artificial neural networks can be an effective factor in water resource management.