Morphing 2-D rainfall fields based on temperature shifts as a basis for future urban flood assessment

Abstract

2-D rainfall fields play a critical role in assessing urban flood impacts and planning drainage systems. High-resolution rainfall fields, obtained from remote sensing devices such as weather radar and satellites, are not largely available and are even more limited for rainfall and flood frequency applications. One method that can be used to estimate extreme rainfall frequency—even with limited data—is Stochastic Storm Transposition (SST), which transposes observed rainfall fields within a region. In the context of climate change, there is a need to alter the observed rainfall fields to account for nonstationary changes in storm intensity and structure. Here, we suggest using Spatial Quantile Mapping (SQM) to modify the intensities and structures of rainfall fields with temperature as a covariate to generate an archive of plausible rainfall fields, which can then be used within SST as an input to assess changes in rainfall and floods. We take Beijing city as a case study, employing 22 years of 1 km hourly downscaled rainfall from CMORPH and near-surface air temperature data from ERA5, to demonstrate the effectiveness of this approach. Initially, SST is run under the current climate and validated for the 2- to 100-year rainfall return levels compared with those of 21 stations within Beijing city. Subsequently, according to the observed relationships between hourly rainfall and temperature, the rainfall fields are modified by the SQM method to fit future temperature conditions. Ultimately, the future extreme rainfall intensities, ranging from 2- to 100-year return levels, are obtained through the integration of the SST and SQM methods. The results indicate that the combined SST-SQM approach can be efficiently used to estimate future rainfall extremes in a changing climate.