Integrated flood vulnerability assessment in a data-scarce tropical mountain-to-piedmont watershed

Abstract Floods rank among the most destructive natural disasters, and assessing flood vulnerability is critical in climate-sensitive regions like Colombia’s Guatiquía River watershed, which supports fragile high-mountain ecosystems. This study integrates geomorphological, hydroclimatic, and anthropogenic variables to develop a flood vulnerability map using geospatial tools and the Frequency Ratio (FR) method. Results highlight the middle-to-lower basin as the most flood-prone area, with 31.83% of the watershed classified as moderate-high (21.19%) to high vulnerability (10.64%), and largest areas of high vulnerability found in Puerto López (38% of its area within the watershed), followed by Villavicencio (18%), Restrepo (16%), and Cumaral (16%), which together comprise the main urban and agricultural centers in the study basin. Flood vulnerability is driven by flat slopes, higher frequency, intensity, and accumulation of heavy rainfall, elevated runoff, and high Curve Number (CN) values under wet conditions (up to 95.94), which reflect extensive urbanization and land transformation, meaning wide extension of impervious surfaces that add to the flooding conditioning factors. Moreover, sustained population density growth, particularly in Villavicencio (a department’s capital of 600,000 inhabitants), underscores the need for ongoing risk monitoring. These findings were supported by the Receiver Operating Characteristic (ROC) analysis, a method that evaluates the performance of binary classification models. The results showed strong predictive performance (AUC = 0.82). This represents the first comprehensive flood vulnerability assessment of the watershed and underscores the need for integrated and region-specific watershed management to mitigate evolving flood risks. Beyond providing robust baseline data, this study offers a replicable methodology for future research in similar high-Andean watersheds.