EventsThe 10th International Electronic Conference on Water Sciences
Published
This submission belongs to the session S10. Groundwater Systems Modelling, Analysis, and Vulnerability—Quantity and Quality Issues of the event The 10th International Electronic Conference on Water Sciences
Published date
19 Nov, 2026
Academic Editor
author-avatarPantelis Sidiropoulos
Contribution&Funding
Conceptualization,
[Omid Ghasemi Fare ]
Methodology,
[Hossein Nabi] ; [Farshad Jahangiri]
Software,
[Farshad Jahangiri]
Validation,
[Hossein Nabi] ; [Omid Ghasemi Fare ] ; [Farshad Jahangiri]
Formal analysis,
[Farshad Jahangiri] ; [Hossein Nabi]
Investigation,
[Farshad Jahangiri] ; [Hossein Nabi]
Resources,
[Hossein Nabi] ; [Omid Ghasemi Fare ]
Data curation,
[Farshad Jahangiri]
Writing - original draft,
[Farshad Jahangiri]
Writing - review,
[editing] ; [Hossein Nabi] ; [Omid Ghasemi Fare ]
Visualization,
[Hossein Nabi] ; [Farshad Jahangiri]
Supervision,
[Omid Ghasemi Fare ]
Funding: --
Citation
Hossein Nabi, Farshad Jahangiri, Omid Ghasemi Fare , Seasonal Assessment of Groundwater Level Decline Susceptibility Using Remote Sensing and Interpretable Machine Learning in the Kaweah Subbasin, California, in Proceedings of The 10th International Electronic Conference on Water Sciences, 24 November–27 November 2026, MDPI: Basel, Switzerland
Share
Email
Facebook
Twitter
LinkedIn

Seasonal Assessment of Groundwater Level Decline Susceptibility Using Remote Sensing and Interpretable Machine Learning in the Kaweah Subbasin, California

Omid Ghasemi Fare 2
1. Department of Civil Engineering, Sharif University of Technology, Tehran, Iran
2. Department of Civil and Environmental Engineering, University of Louisville, Louisville, USA
Abstract

Groundwater depletion threatens water security, agricultural productivity, and aquifer integrity in intensively irrigated alluvial aquifers, where groundwater extraction, climate variability, and agricultural water demand interact over space and time. This study assessed seasonal groundwater-level decline susceptibility (GLDS) at monitored wells in the critically overdrafted Kaweah Subbasin, California, using remote sensing, gridded environmental data, and machine learning (ML). A quality-controlled database of 1,195 well-year records from 383 wells during 2010–2024 was developed. Seasonal decline was calculated from maximum January–May and minimum September–November groundwater elevations; declines greater than 5 ft were classified as substantial. Seventy-six hydroclimatic, evapotranspiration, vegetation, land-cover, topographic, and soil predictors were evaluated using logistic regression, random forest (RF), and LightGBM within nested well-group cross-validation. Substantial decline occurred in 54.7% of records, with an overall median of 7.0 ft and an annual event frequency of 82.4% in 2021. RF provided the strongest overall performance, with a mean precision–recall area under the curve of 0.773 ± 0.032 and a Brier score of 0.193 ± 0.011; LightGBM achieved the highest balanced accuracy (0.721 ± 0.025). High susceptibility was concentrated mainly in the southeastern monitoring cluster. Hydroclimatic variables contributed 73.0% of total held-out SHAP importance, led by wet-season solar radiation, water-year precipitation, and wet-day frequency. Model rankings were stable after excluding low-coverage soil records. The framework provides a transparent screening tool for prioritizing monitoring, drought preparedness, and groundwater-demand management while avoiding unsupported wall-to-wall mapping claims.

Keywords
Groundwater level decline susceptibility; Remote sensing; Machine learning; Grouped cross-validation; Kaweah Subbasin
Precision Surface Irrigation with Geospatial Monitoring: Water and Energy Efficiency Improvements in Okara, Punjab, Pakistan
Spatio-Temporal Dynamics of Groundwater Potential Zones under Climate Variability and Land-Use Change in Agago District, Uganda: An Integrated GIS–AHP and CA-ANN Approach