Innovative use of reanalysis data in Slovenia: Enhancing precipitation time series with a multiplicative cascade model

Slovenia Precipitation reanalysis products (PRP) with hourly resolution as ERA5-Land and COSMO-REA6 are a promising solution for hydrologic applications in data scarce regions as Slovenia. Both data sets are validated against observed hourly time series from 5 rain gauges and areal precipitation in 20 catchments in Slovenia regarding continuous and event-based precipitation characteristics as well as extreme values. Both, station and catchment precipitation are not well-represented by PRP, with a worse representation by ERA5-Land than COSMO-REA6. The aim of this study is to explore how if the PRP time series can, instead of being directly appliedtime series, from precipitation reanalysis products (PRP) can be used for the generation of high-resolution precipitation time series in unobserved catchments. A new approach is proposed to estimate parameters of a micro-canonical cascade model from ERA5-Land and COSMO-REA6 time series to generate hourly time series from daily observations. The proposed parameter estimation approach leads to a general better representation of precipitation characteristics. For some stations the disaggregation based on COSMO-REA6 parameters even outperforms disaggregation results with parameters estimated from observed time series at these stations. Significant spatial patterns of PRP errors are identified which can help improving future PRP. The parameterization approach is not limited to the study region and can be used for precipitation generation in unobserved catchments, particularly in catchments with complex terrain which are usually not well represented by PRP. For many hydrological applications precipitation data with hourly resolution are required. Unfortunately, this kind of data exists often only for a few stations or only for short periods in the region of interest. Precipitation reanalysis data (PRP), which are produced by simulating the earth’s atmosphere over the last decades, are a promising solution. However, the quality of PRP depends on multiple factors and has to be tested for each region before application. In this study widely used ERA5-Land and COSMO-REA6 are tested for five rain gauges and 20 catchments in Slovenia. The validation of PRP with available observations for the study area indicated strong deviations. So instead of using the PRP directly for hydrological applications, a precipitation generator was parameterized with information from the PRP. Its validation shows a better representation of the observations, which suggests their usage for subsequent hydrological applications. • New parametrization strategy for a precipitation generator and validation. • Generation of hourly precipitation data sets for unobserved catchments. • Generated time series outperform precipitation reanalysis products (PRP). • Identification of spatial patterns in PRP errors for Slovenia.