Classification of flood-prone areas in Mexico: a model using Naive Bayes

Background Flooding is a critical problem in specific states of Mexico.Methods This study uses a Naive Bayes model for zone flood susceptibility identification based on an inventory compiled from official flood reports between 2000 and 2022. Four geomorphometric indices—TWI, TPI, SPI, and PFI (proposed in this research)—were used, along with slope and roughness.Results Analysis shows that higher values of TWI and PFI provide the strongest positive signal for occurrence of floods, while SPI and slope reduce this propensity in their high ranges; TPI and roughness give strong signal in intermediate ranges. These patterns confirm the differentiated contribution of each predictor and support the validity of the model. Model performance was evaluated using the ROC–AUC metric, showing a satisfactory discrimination capacity for flood-prone areas.Conclusions Here it is shown a framework that allows the generation of susceptibility maps useful for risk management in the region based on geomorphic indices.