Comparing heuristic and bivariate GIS-based methods for refining landslide susceptibility maps in Northern Mexico City

The northern portion of Mexico City's urban zone faces new geological hazards due to the large population growth. Specifically, at the Sierra de Guadalupe, between the Federal District and Mexico State, landslide activity has been present as rockfalls and wedge failures, especially during and after the rainy season. In 2004, the Metropolitan Geological Survey created the first landslide inventory map and geographic information systems-based landslide susceptibility analysis using an heuristic technique. Using the same database of triggering parameters and raster characteristics to ensure comparable results, we created a new landslide susceptibility zonation using the bivariate method. Accuracy of the predictions for both methods was tested through comparisons with the landslide inventory map and new landslide activity (during year 2005). The results using different methodologies represented a change of about 70 percent of the landslide susceptibility definition in the Sierra de Guadalupe. The study also demonstrated that the landslide susceptibility zonation can be improved by introducing quantitative analysis and choosing well-related causative parameters to the landslide phenomena. Specifically, the knowledge of experts on the effects and degree of influence of each triggering parameter used for the analysis is very important in order to define a highly representative zonation independent of the technique. Further improvements can be obtained by refining databases, by selecting triggering parameters, and by generating a historical landslide activity database for the Sierra de Guadalupe.