Downscaling of convection-permitting simulated precipitation for future projection

Convection-permitting regional climate models (CPRCMs) have emerged as a promising approach to produce high-resolution climate simulations as they are able to better represent sub-daily temperature and precipitation patterns and extreme values compared with coarser- resolution convection parameterizing simulations.  However, to bridge the gap between the grid scale (1-4 km) of CPRCMs and local spatial scale (ground station), bias-correction is still needed. The objective of our work is to test a statistical downscaling technique on CPRCMs to produce bias-corrected precipitation for a future period at the station level. Quantile mapping (QM) was chosen because this empirical method is among the most reliable and straightforward bias correction techniques. The quantile mapping process was implemented between CPRCM outputs for the historical period (1998 – 2009) and the point-scale observations and then applied to sub-hourly precipitation for the future period (2078 – 2089) in Zurich, Switzerland. Kloten station in Zurich was chosen to implement the experiment. At the same time, a simple nonparametric quantile mapping approach was used for other hourly climate variables such as temperature, relative humidity, air pressure, global radiation, and wind speed. The climate extremes indices for temperature and precipitation were also calculated to analyze the long-term climate trends in the study site. According to our results, the climate change signals of both precipitation and temperature were not altered during bias adjustment and the inter-variable dependencies were also preserved. Nevertheless, limitations of our approach still remain and future work is needed to determine whether more advanced techniques can improve sub-daily predictions for multiple climate variables.