A Data Machine Learning-driven Approach to Explore Resilience and Sustainability of Mexico’s Water Resource Management

Adrián Landaverde-Nava, Cristian Gonzaga-López, Michael Steven Delgado-Caicedo, Evelyn Geovanna Pérez-Gómez, Jesús Yair Ramírez-Islas, Luis Gerardo Lagunes-Nájera, Deborah Tirado-Hernández, Elisabetta Crescio, Miguel Gonzalez-Mendoza

Abstract


Nowadays, Mexico faces several water challenges. Among them, the most critical are waterscarcity, water quality, and flood management. These issues are exacerbated by climate change, population growth, and insufficient infrastructure, posing significan trisks to communities across the country. This projectleverages cutting-edge technologies, including artificial intelligence models and spatial inference in Python, to tackle these challenges and enhance the resilience of Mexican communities. A time series model, developed using Prophet, was trained to forecast drought levelsin every state of Mexico for the period 2024-2026, using available historical data. This predictive approach aims to support policymakers in preparing for and mitigating the effects of prolonged droughts. Inaddition, groundwater quality was analyzed using Ordinary Kriging Interpolation and clustering techniques, enabling the identification of areas most at risk of waterquality deterioration. Finally, an image segmentation deep learning model was implemented to analyze images, focusing on detecting large bodies of water and mapping flooded regions. Together, these tools offer a comprehensive strategy for managing Mexico’s pressingwater-related challenges.

Keywords


Water scarcity, water quality, flood segmentation, artificial intelligence

Full Text: PDF