Intercomparison of precipitation frequency estimates derived from high-resolution weather radar and rain gauge measurements in Zealand, Denmark

Study region:: Zealand, Denmark. Study focus:: Precipitation frequency estimates (PFEs) guide the design of urban drainage systems by specifying return periods for pluvial flooding. PFEs are typically derived from point rainfall measurements, whereas weather radar provides spatially distributed precipitation measurements. This study compares PFEs and PFE model parameters derived from weather radar and rain gauge data at timescales from 1 min to 48 h, with a focus on spatial variability. PFEs are estimated empirically with the median plotting position and parametrically using a Peaks-Over-Threshold approach, generating partial duration series modeled by the Generalized Pareto distribution. New hydrological insights for the region:: Return levels from weather radar and rain gauges often exhibit increasing agreement at durations above approximately 200–300 min, depending on the estimation method and return period considered. At short timescales, rain gauges show higher mean exceedances μ and thresholds z0 than radar pixels, while yearly exceedance rates λ and shapes κ are more consistent between the data sets. Results at the shortest timescales are, however, associated with greater uncertainty. Spatial gradients are found for λ. Spatial autocorrelation of return levels, κ, and μ increases with increasing rainfall duration, while spatial autocorrelation for λ remains high for all timescales considered. Differences in PFEs and model parameters likely stem from inherent differences in measurement techniques and spatial scales between the weather radar and rain gauge data.