¿Peligro o exposición? Evaluación caso-control de puntos críticos hídricos del Perú

Administrative hazard inventories guide investment in disaster prevention, but they are rarely tested against independent physical evidence. We assessed Peru's national inventory of critical river and stream reaches compiled by the National Water Authority (7 268 sites, 2016-2026) using a case-control design on the MERIT Hydro drainage network. Each site was compared with local controls, both random and matched by drainage area, using geomorphic descriptors (height above nearest drainage, slope, relative elevation), exposure (GHSL) and a paired event study with ten Sentinel-1 seasons. Conditional logistic regression and gradient boosting with basin-blocked cross-validation were applied. Sites occupy more flood-prone positions, with lower height above drainage (OR = 0.59 per standard deviation) and a larger low-floodplain fraction (OR = 1.42). However, for rivers of equal size, exposure discriminates better than terrain (out-of-sample AUC 0.74 versus 0.65; combined, 0.77). Radar showed no inundation signal tied to the registration year, and apparent signals obtained with random controls vanished once controls were matched by river size. A preliminary single-threshold score calibrated with 2023 events and mixed orbits (AUC = 0.79) fell to 0.58 when recomputed by orbit pass. The inventory behaves more as a risk and prioritization register than as a physical hazard map. We propose a Hydrogeomorphic Compatibility Index that flags 391 sites (5.4%) as priorities for field verification.