Performance-based assessment of tropical cyclone parametric precipitation models for flood simulation applications

Tropical cyclone (TC) rainfall is one of the primary drivers of flooding in coastal watersheds; therefore, accurate rainfall estimation is essential for reliable hydrologic and hydraulic (H&H) modeling. In TC-prone regions like the Caribbean Islands, which are also topographically complex, the wide range of models available for estimating TC rainfall presents a challenge in identifying the most suitable Parametric Precipitation Models (PPMs) for flood modeling applications. Thus, this study proposes a dual-component framework for benchmarking PPM-derived rainfall: (i) a rainfall-level comparison of PPMs and (ii) a performance-based assessment of the rainfall fields produced by PPMs for flood simulations applications. The proposed framework was applied to a watershed in Puerto Rico and tested under three historical hurricanes using five PPMs and Stage IV radar-based rainfall as a benchmark. The rainfall-level analysis assessed precipitation accumulation distribution, patterns, and cumulative totals, while the performance-based assessment examined the impacts on riverine flow timing, peak discharge, and overland flood levels in the selected watershed. Results show that the Parametric Hurricane Rainfall Model (PHRaM) and its shear-excluded variant performed most consistently across all metrics. However, PPM-derived rainfall fields are unreliable for real-time forecasting in their current state. Importantly, simulated peak flows were more affected by the rainfall uncertainty than flood extents. Thus, this study highlights the need to select an appropriate tool for assessing flood risk, particularly in steep, complex terrain. Although PPMs are efficient in data-scarce regions, their inherent structural limitations must be carefully evaluated to ensure reliable flood hazard predictions.