X-band polarimetric radar QPE for urban hydrology: The increased contribution of high-resolution rainfall capturing

Abstract

Increased population is exposed to urban flooding as a result of climate change. Flood early warning can largely reduce the loss of life and asset, while how detailed rainfall information is enough has always been a critical issue in achieving robust flood forecasting. Previous analysis has been limited by lacking high-resolution rainfall observation or a spatially-matched urban hydrological model. This study couples mature modeling technologies with comprehensive hydro-meteorological measurement, that is comprised of X-band polarimetric radar (75 m, 3 min), weather stations, and hydrometric stations, and this study further attempts to understand the rainfall spatial resolution effects and critical resolutions from a multi-scale perspective (e.g. block, district, and catchment scales). The results show that coarser resolutions always lead to underestimation of peak flows, even exceeding 40 % in the case of adopting 10 km rainfall data at the block scale. But such damping effects on discharge simulation can be largely attenuated with increased hydrological scale. In general, spatial resolutions of higher than 500 m, and 10 km are needed to obtain accurate simulation results (i.e. keep the relative bias within 20 %) at district and catchment scales, while at least 300 m for the block scale. In conclusion, refined rainfall forcing contributes to better simulation of peak flows in urban hydrological modeling, but the required rainfall spatial resolution changes with the scale of hydrological application. These findings help to better achieve the balance between rainfall observation cost and flood simulation accuracy.