Document Type : Original Article

Authors

• Sina Khoshnevisan ¹
• Samad Emamgholizadeh ²
• Mohammadreza Asli Charandabi ¹
• Seyedeh Fatemeh Khakzad ²

¹ Department of Water Resources and Environmental Engineering, Faculty of Civil Engineering, Shahrood University of Technology, Shahrood, Iran

² Department of Water Resources and Environmental Engineering, Faculty of Civil Engineering, Shahrood University of Technology, Shahrood, Iran.

Abstract

The continuous decline of groundwater levels in arid and semi-arid regions such as the Shahrood and Bastam plains driven by excessive extraction, agricultural and industrial expansion, and climate variability, has become one of the major challenges in water resources management. This persistent decline leads to consequences such as reduced aquifer storage, land subsidence, deterioration of water quality, and threats to agricultural sustainability. Therefore, accurate prediction of groundwater levels is essential for sustainable management, planning, and policy-making. This research aims to evaluate and compare the performance of five machine learning algorithms, including XGBoost, CatBoost, Decision Tree, Support Vector Regression, and K-Nearest Neighbors for predicting groundwater levels in the Shahrood and Bastam aquifer during the period 2000–2014. The input dataset consisted of climatic variables. groundwater extraction from wells and qanats, and agricultural return flow. The models were trained using 80 percent of the data and tested on the remaining 20 percent, and their performance was assessed using MAE, RMSE, and correlation coefficient (r). The results showed that gradient boosting models outperformed the classical algorithms; CatBoost achieved the lowest error with an MAE of 1.403 m and an RMSE of 1.948 m, while XGBoost produced the highest correlation (r = 0.818). In contrast, the DT, SVR, and KNN models exhibited lower accuracy due to their limited capability in capturing nonlinear relationships. Overall, the findings suggest that boosting algorithms can serve as powerful tools for groundwater level prediction and can support decision-making related to extraction control, climate impact assessment, and sustainable aquifer management.

Keywords

• Shahrood Aquifer
• Boosting Models
• Sustainable Water Resources Management
• Climatic and Hydrological Factors
• Groundwater Level Fluctuations

Main Subjects

• ًQanat and aquifer