The Mezquital Valley (MV) hosts the largest unmanaged aquifer recharge scheme in the world. The metropolitan area of Mexico City discharges ~60 m³/s of raw wastewater into the valley, a substantial share of which infiltrates into the regional aquifer. In this work, we aim to develop a comprehensive approach, adapted from oil and gas reservoir modeling frameworks, to assess water supply sources located downgradient from unmanaged aquifer recharge zones. The methodology is demonstrated through its application to the Mezquital Valley region. Geological, geoelectrical, petrophysical and hydraulic information is combined into a 3D subsurface model and used to evaluate downgradient supply sources. Although hydrogeochemical variables are yet to be assessed, outcomes suggest that the newly-found groundwater sources may provide a long-term solution for water supply. Piezometric analyses based on 25-year records suggest that the MV is close to steady-state conditions. Thus, unmanaged recharge seems to have been regulating the groundwater balance for the last decades. The transition from unmanaged to managed recharge is expected to provide benefits to the MV inhabitants. It will also be likely to generate new uncertainties in relation to aquifer dynamics and downgradient systems.

La Inteligencia Artificial Generativa (IAGen) está remodelando la educación actual al permitir nuevas formas de acceder y generar conocimiento. En geociencias, y particularmente en hidrogeología, la adopción de herramientas de IAGen como ChatGPT permanece en gran medida inexplorada en comparación con otras disciplinas. Este estudio examina la aplicabilidad y confiabilidad de ChatGPT en la enseñanza de la hidrogeología a través de un análisis exploratorio de las interacciones pregunta-respuesta en diferentes niveles cognitivos: definiciones básicas, resolución de problemas cuantitativos, evaluación crítica, análisis de artículos científicos y creación de diagramas, utilizando como marco de referencia la taxonomía revisada de Bloom. Nuestros resultados indican que ChatGPT tuvo un desempeño satisfactorio en hidrogeología, mostrando fortalezas en definiciones y problemas matemáticos. Sin embargo, mostró limitaciones en análisis más profundos y referencias científicas. Su calificación promedio fue de 6.62/10.0. Aunque este modelo facilita las respuestas inmediatas y el autoaprendizaje, persisten las preocupaciones sobre la precisión, la profundidad y las consideraciones éticas. Este estudio contribuye al debate en curso sobre el aprendizaje asistido por IA en geociencias, haciendo hincapié en la necesidad de incorporar estrategias pedagógicas estructuradas que incorporen la IA y que al mismo tiempo mantengan el rigor científico.

Groundwater recharge (GWr) estimation is fundamental for assessing aquifer storage changes, ensuring groundwater sustainability, and supporting effective water management. However, regional groundwater recharge assessment remains challenging because it requires integrating large geospatial datasets, computational resources, and hydrogeological information that are often unavailable in data-limited regions. This study presents a cloud-based framework for estimating potential groundwater recharge (GWpr) using open satellite products and the Thornthwaite-Mather (T-M) water balance model implemented in Google Earth Engine (GEE) and executed through Google Colab. The framework integrates globally available datasets, including CHIRPS precipitation, MODIS potential evapotranspiration, and OpenLandMap soil properties, within a reproducible Python workflow that minimizes dependence on local computational infrastructure. The Basin of Mexico (BM), an endorheic volcanic basin affected by intensive groundwater exploitation, was selected as a case study to evaluate the framework against a previously calibrated Soil Water Balance (SWB) model. The proposed implementation estimated a mean annual ground water potential recharge (GWpr) of ~ 44.7 mm/year, closely matching the reference model at the basin scale. Differences between both approaches were ~ 4–5% in the Zona Metropolitana and Tecocomulco groundwater management units (GMUs), whereas larger discrepancies (~ 20–30%) occurred in eastern and northern GMUs, reflecting differences in model temporal representation and local hydroclimatic conditions. Overall, the proposed cloud-computing framework provides an accessible, reproducible, and transferable approach for estimating GWpr using globally available datasets, facilitating groundwater assessments in data-limited regions with diverse hydrogeological settings.