Design areal precipitation in Czech catchments – radar data usage instead of station data reduction

Design precipitation intensity is an essential variable used in water management practice. In particular, it is essential for the hydrological modeling of proposed measures in ungauged catchments. However, point design precipitation totals cannot be used in larger catchments as they do not include information on the areal variability of precipitation. Therefore, in practice, point design precipitation totals are reduced to catchment areas using empirically derived Area Reduction Factors that are subject to a number of uncertainties. To reduce these uncertainties, we present here an evaluation of areal design precipitation derived from radar data using the Czech Republic as an example.

The input dataset is radar reflectivity data at an altitude of 2 km (pseudo-CAPPI 2km) for 20 years between 2002 and 2021 with a temporal resolution of 10 minutes and a spatial resolution of 1 km. These data are then adjusted with daily precipitation totals from stations. From the adjusted precipitation intensities, precipitation totals are determined for durations from 30 minutes to 3 days. From these rainfall totals, their areal averages are calculated for individual catchments at four hierarchical levels. From these, the L-moment method is used to derive Generalized Extreme Value (GEV) distribution parameters, which are used to determine design areal precipitation totals in the considered catchments for all considered durations.

Naturally, as the catchment area increases, the areal design precipitation decreases. The average  1-hour design precipitation total with a return period of 20 years is 36 mm for the smallest catchments, whereas the design precipitation is approximately half of that for a catchment of 1000 km². However, for longer periods of rainfall accumulation, the decrease is not as pronounced. Therefore, on larger catchments, larger ratios of the design totals are achieved between longer and shorter accumulation times. In terms of the spatial distribution of design areal precipitation totals, the design 1-hour totals are very randomly distributed, with an obvious dependence on the catchment size. On the contrary, in the case of the design 24-hour totals, the influence of georelief is a major factor, with the highest totals being strongly concentrated in the mountains and their foothills.

We have shown that the adjusted radar data are suitable for estimating design area precipitation over the catchment, although the longer return periods are particularly subject to the uncertainty caused by the short data series. The demonstrated influence of georelief on longer accumulation times then points to the inappropriateness of the approach of deriving design areal precipitation using reduction coefficients empirically derived from point values for large areas. Radar data, allowing direct calculation of design areal precipitation totals for a specific catchment, seem to be more appropriate.