Iranian Journal of Irrigation & Drainage

Iranian Journal of Irrigation & Drainage

Comparison of LS-SVR, ANFIS, MLP, and RBF Model Performance in Groundwater Level Modeling (Case Study: Central Part of Mashhad Plain)

Articles in Press

Document Type : Original Article

Authors
Ommeh hani mousavi khaledi 1
Ali Firoozzare 2
Arash Dourandish 3
Fereshteh Modaresi 4
1 Department of Agricultural Economics, Faculty of Agriculture, Ferdowsi University of Mashhad,Mashhad
2 Department of Agricultural Economics, Faculty of Agriculture, Ferdowsi University of Mashhad, Mashhad
3 Agricultural Economics, Faculty of Agriculture, University of Tehran, Karaj
4 Department of Water Science and Engineering, Faculty of Agriculture, Ferdowsi University of Mashhad
10.22034/idj.2026.546388.2628
Abstract
The decline in groundwater levels in Mashhad Plain, primarily driven by anthropogenic factors, presents a significant challenge to sustainable water resource management. The heavy reliance on these resources necessitates accurate prediction of changes for effective managerial decision-making. This study employs machine learning methods to simulate and predict groundwater level variations in the central part of the plain. Initially, key input variables and optimal time lags were identified using the Partial Autocorrelation Function (PACF) and Frequency Lasso Regression (FLR). Subsequently, the performance of Least Squares Support Vector Regression (LS-SVR), Adaptive Neuro-Fuzzy Inference System (ANFIS), Multi-Layer Perceptron Neural Network (MLP), and Radial Basis Function Neural Network (RBF) was evaluated for monthly prediction of groundwater level fluctuations over 30 years (1991-2021). Results demonstrated that all four models simulate groundwater levels with acceptable accuracy. The RBF model exhibited superior performance with R², MSE, NSE, and RMSE values of 1.00, 0.00, 1.00, and 0.009, respectively. The findings affirm the high potential of data-driven models in simulating hydrological processes and can serve as a foundation for developing intelligent tools for groundwater resource management in Mashhad Plain and similar regions, including resource allocation and warning system design.
Keywords
Machine Learning Models
Supervised Learning Methods
Data-Driven Models
Groundwater Level Fluctuations
Iranian Journal of Irrigation & Drainage

Articles in Press, Accepted Manuscript
Available Online from 15 May 2026