UNAM Unveils Landslide Risk and Inequality Map in Mexico City
Mexico City, April 3 – A comprehensive study by researchers at the National Autonomous University of Mexico (UNAM) has revealed a detailed map of landslide susceptibility in Mexico City and its metropolitan area. The groundbreaking research, recently published in the journal PlosOne, suggests the possibility of developing differentiated policies to benefit the region’s inhabitants.
Dr. Mario Alejandro Mercado Mendoza, lead researcher from UNAM’s Institute of Economic Research (IIEc), explained that the study, co-authored with Armando Sánchez Vargas and Pierre Mokondoko, aims not to cause alarm but to address the stark inequalities observed in areas like Santa Fe. These regions often feature affluent communities alongside those experiencing extreme poverty, with the latter being more vulnerable to the impacts of potential landslides.
Mapping Vulnerable Zones
The research focused on a large area encompassing ravines within the municipalities of Huixquilucan and Naucalpan de Juárez in the State of Mexico, and the Cuajimalpa de Morelos and Álvaro Obregón boroughs in Mexico City. These specific zones were chosen due to the extensive development of commercial centers, high-end residences, and luxury offices by real estate companies. This development, Mercado Mendoza noted, has led to gentrification and exacerbated economic inequality for the original inhabitants.
The study emphasizes that landslides, particularly during rainy seasons or due to water infiltration causing slope collapse, can significantly impede sustainable development.
Advanced AI for Risk Assessment
To create the susceptibility map, the researchers utilized 13 machine learning models, incorporating topographical and hydrological indicators to visualize landslide risks from earth flows. Among these, the ‘Extreme Gradient Boosted Trees’ model was selected for its superior precision in differentiating terrain variations.
This model allowed for the integration of social and economic data with risk assessment. By combining data on social lag and calculating a potential relationship between these variables and landslide susceptibility, using techniques like Copulas and SHapley Additive exPlanation (SHAP values) via beeswarm plots, the study provided nuanced insights.
Mercado Mendoza described the findings using an analogy:
