Uncertainties in ERA-5 and UERRA reanalyses detected by adjusted radar precipitation - a regional study for the Czech Republic

Regional reanalysis constitute of potentially attractive new data source for many applications. They can offer added value benefiting from their higher spatial and temporal resolution. On the other hand, similarly as other data sources in regular network, reanalysis comes with uncertainties especially in the case of extremes.

The monitoring capability of reanalysis is essential for their usage as the reference datasets for climate model validation as well as for hydrological models. In this contribution, we evaluate the agreement of precipitation between modern reanalysis products (Era5, Era5 Land, Harmonie and Mescan-Surfex; with resolution between 5.5 and 32 km) and observed data at different time scales, from annual to subdaily. Studied characteristics of precipitation are for instance annual cycle of precipitation amount and the number of wet days, diurnal cycle of precipitation, and extremes. The common period for all datasets is 2002 – 2018.

Observed data used in this study are represented by adjusted radar-derived precipitation totals in 1km raster over the Czech Republic. The adjusted radar-derived precipitation totals are gained as follows. First, the radar-derived rain field is spatially adjusted to measurements from the rain gauges as a whole. For each day, the ratio between the mean 1-day precipitation total calculated from all rain gauges and the mean 1-day precipitation total estimated from the corresponding radar pixels is determined and used for a multiplication of radar-derived precipitation in every pixel of the radar domain. Second, the spatially adjusted radar-derived rain rates are locally adjusted in individual pixels based on the distance from the closest rain gauge, whereas the weight of the observed precipitation is decreasing with increasing distance to the given pixel. Adjusted daily precipitation total is then divided according to 10 min radar-derived estimates, from which the precipitation accumulations of longer duration are calculated.