Is bias correction necessary for CPRCM-driven flood simulation in mountainous region?

Convection-permitting regional climate models (CPRCMs) at kilometer scale can better represent intense precipitation, yet their added value for flood-risk applications is still limited and often inconsistent. A key reason is the presence of systematic biases in precipitation and temperature over complex terrain, which may strongly affect hydrological response. To address whether bias correction is necessary when using CPRCM forcing for flood modelling in complex terrain, we run WRF-Hydro with raw and bias-corrected 3-km HCLIM3 precipitation and temperature for two contrasting basins spanning coastal to mountainous terrain in western Norway: Røykenes (coastal, rainfall-driven floods) and Bulken (mountainous, snowmelt-influenced floods). We further compare two widely used bias-correction approaches, i.e., Quantile Mapping (QM) and Distribution Delta Mapping (DDM), applied to precipitation and temperature prior to the hydrological simulations.

The results show that bias correction reduces mean biases in both variables, but its effectiveness depends on basin type and metric. In Røykenes basin, QM does not adequately correct annual maximum 1-hour precipitation, whereas DDM provides a better adjustment of extreme precipitation. For temperature, the correction reduces absolute bias relative to raw HCLIM3 but also shifts the bias from cold to warm. In terms of hydrological performance, raw HCLIM3 forcing already yields a small flood-peak bias in Røykenes basin (~3% underestimation), while bias-corrected forcing can further worse this peak underestimation. In Bulken basin, temperature correction improves both flood peaks and flood seasonality, underscoring the strong sensitivity of snowmelt-influenced floods to temperature errors. By contrast, precipitation correction in this mountainous basin degrades flood-simulation skill. Overall, our results show that CPRCM forcing can be highly informative for flood simulations, but the benefits depend on process regime: temperature correction is critical for snowmelt-dominated basins, while precipitation correction over mountains requires particular caution.