Comparative Evaluation of Depth-Damage Estimation Approaches Using High-Resolution Building Inventory Data and Hydraulic Model Outputs

Accurate estimation of flood-induced economic losses remains a critical challenge in flood risk assessment due to uncertainties associated with vulnerability modeling and exposure representation. This study presents a comparative evaluation of depth-damage estimation approaches by integrating high-resolution building inventory data with hydraulic model outputs. A two-dimensional hydrodynamic model was developed to simulate spatial flood depth distributions, which were subsequently intersected with structure-level building datasets to enable object-based damage assessment. Three depth-damage estimation approaches were evaluated: an empirical curve model, a parametric relative damage function, and a multi-parameter vulnerability model incorporating structural attributes. Model performance was assessed using statistical metrics including Mean Absolute Error (MAE), Root Mean Square Error (RMSE), bias ratio, and sensitivity analysis under controlled hydraulic depth perturbations. Results indicate that high-resolution exposure data significantly influence loss estimation magnitude and spatial distribution, reducing aggregation bias commonly observed in traditional approaches. Depth-only formulations were found to systematically overestimate damages under shallow flooding conditions, while the multi-parameter model demonstrated improved robustness and reduced sensitivity to hydraulic uncertainty. Comparative analysis further revealed that vulnerability formulation contributes substantially to overall uncertainty in flood loss prediction, often exceeding hydraulic modeling variability. The findings highlight the importance of integrating asset-level exposure datasets and locally calibrated vulnerability relationships for reliable economic loss estimation. The proposed framework provides methodological guidance for flood risk practitioners, supports improved disaster loss accounting, and contributes to resilience-oriented urban planning. Future research directions include adaptive artificial intelligence–based damage functions and real-time digital twin systems for dynamic flood loss estimation.