Bahouq T., Moumane A., Elmotawakkil A., Touhami N., Bhagat S.K. (2026). Monitoring agricultural transformation and its impacts on water resources in the Gharb plain, Morocco, using machine learning and remote sensing. IEEE Access, 01/01/2026, vol. 14, p. 87852-87867.
https://doi.org/10.1109/ACCESS.2026.3699252
https://doi.org/10.1109/ACCESS.2026.3699252
| Titre : | Monitoring agricultural transformation and its impacts on water resources in the Gharb plain, Morocco, using machine learning and remote sensing (2026) |
| Auteurs : | T. Bahouq ; A. Moumane ; A. Elmotawakkil ; N. Touhami ; S.K. Bhagat |
| Type de document : | Article |
| Dans : | IEEE Access (vol. 14, 2026) |
| Article en page(s) : | p. 87852-87867 |
| Langues : | Anglais |
| Langues du résumé : | Anglais |
| Catégories : |
Catégories principales 07 - ENVIRONNEMENT ; 7.3 - Eau. Gestion de l'EauThésaurus IAMM GESTION DES EAUX ; SYSTEME DE PRODUCTION ; CULTURE PLUVIALE ; CULTURE IRRIGUEE ; AQUIFERE ; RESSOURCE EN EAU ; MAROC |
| Résumé : | Driven by national initiatives like the Green Morocco Plan, Morocco's Gharb Plain has shifted from traditional rain-fed farming to intensive, high-value commercial cultivation. Previous regional research has documented land-use changes and groundwater stress separately. A critical research gap remains in capturing the direct, seasonal interactions between long-term surface greening and subsurface aquifer degradation, where apparent vegetation expansion often masks underlying resource depletion. This study aims to bridge this gap by quantifying the direct relationship between long-term agricultural intensification and localized hydrological stress, within a representative study area of the Gharb Plain. We adopted an integrated, multi-scale approach combining four decades (1984-2024) of Landsat satellite imagery processed via Google Earth Engine (GEE) with seasonal field observations. A Random Forest machine-learning algorithm and spectral indices tracked land use/land cover (LULC) changes. These findings were cross-validated with in situ seasonal hydrochemical monitoring across 68 wells spanning four distinct soil-agricultural zones. Long-term data confirmed robust vegetation densification, with EVI increasing by 1.24 km(2)/year due to perennial avocado expansion. Seasonal LULC mapping showed the agricultural footprint peaks in winter (166.5 km(2)) and contracts to its minimum in summer (113.7 km(2)). This summer contraction marks peak hydrological stress: intensive irrigation pumping caused electrical conductivity (EC) to spike to 2, 909 mu S/cm in the sandy loam zone and 2, 888 mu S/cm in the sandy-clay zone, while water table depths dropped to seasonal lows of 6.75 m and 7 m, respectively. This work establishes a vital baseline reference. Practically, these insights enable regional water management authorities and policy planners to deploy targeted multi-year monitoring, implement water-efficient irrigation technologies, and restrict water-intensive crops to protect the plain's long-term agricultural viability. |
| Cote : | En ligne |
| URL / DOI : | https://doi.org/10.1109/ACCESS.2026.3699252 |


