Machine Learning and Hyperparameters Algorithms for Identifying Groundwater Aflaj Potential Mapping in Semi-Arid Ecosystems Using LiDAR, Sentinel-2, GIS Data, and Analysis

Research identifies Bayesian optimization as the most efficient hyperparameter algorithm for predicting groundwater aflaj potential in semi-arid ecosystems, achieving 99% accuracy. The study, conducted in Nizwa, Oman, used machine learning models (SVM and XGB) combined with Sentinel-2, LiDAR, and GIS data to map potential zones into five classes: deficient, low, moderate, high, and very high. Bayesian-XGB outperformed other models (GS-XGB, RS-XGB, SVM) in validation, highlighting its superior efficiency for groundwater aflaj mapping. The findings emphasize the importance of integrating geospatial and machine learning techniques to support integrated water source management (IWSM) programs and protect historical aflaj irrigation systems from threats.

Al-Kindi, K. M., & Janizadeh, S. (2022). Machine Learning and Hyperparameters Algorithms for Identifying Groundwater Aflaj Potential Mapping in Semi-Arid Ecosystems Using LiDAR, Sentinel-2, GIS Data, and Analysis. Remote Sensing, 14(21), 5425. https://doi.org/10.3390/rs14215425