Morphing sub-daily rainfall fields based on temperature shifts to project future changes in rainfall extremes

Understanding how the space-time properties of extreme rainfall shifts due to climate change is essential for assessing risks in water-related hazards. However, future sub-daily rainfall fields, which are the main trigger of pluvial and flash floods, are not readily available for most locations and many climate change scenarios, challenging the assessment of future hazards and risks. An alternative solution to running computationally expensive convection-permitting climate models to obtain future short-duration rainfall fields is morphing recorded rainfall fields considering temperature as a driving factor. Here, we suggest using a Gamma-based spatial quantile mapping (GSQM) method with temperature as a covariate to project an archive of plausible future rainfall fields that can be used to assess future changes in extreme rainfall frequency. Combined with a stochastic storm transposition (SST) method, which can estimate rainfall frequency for arbitrary spatial scales based on gridded rainfall, future changes in regional rainfall extremes can be efficiently projected. Using Beijing as a case study, we employ 22 years of 1 km² hourly rainfall and hourly air temperature data to demonstrate the validity of the GSQM-SST approach. First, the observed scalings governing changes in rainfall fields with temperature have been explored across various intensities of rainfall. Then, those scalings are used to morph the rainfall fields’ intensities, areas, and spatial coefficients of variation. Finally, future extremes of 2- to 100-year return levels under several warming scenarios are estimated by integrating the GSQM and SST methods.