Analysis of Reanalysis Data for Heavy Precipitation Grouping in Europe

Heavy precipitation clustering is important for flood risk in Europe, but its description in reanalysis datasets is still uncertain. This study examined how well ERA5, ERA5-Land, and JRA-55 reproduce the size and timing of extreme precipitation from 1981 to 2022. Observations from the E-OBS dataset were used as reference, with heavy events defined as daily totals above the 95th percentile. Consecutive wet days were grouped into clusters, and measures such as mean cluster length (MCL) and mean gap between clusters (MGC) were used. Correlations between reanalysis and observed MCL were 0.58–0.63 across seasons, with mean absolute errors of 0.9–1.2 days. The largest bias was found in convective areas, where MGC was underestimated by up to 0.6 days. Sensitivity tests showed that thresholds and linking rules had stronger influence on clustering than the dataset used. The results show that reanalyses reproduce large-scale patterns but tend to underestimate storm duration and event order, which affects flood modeling. Better use of data, improved physical methods, and denser observation networks are needed to reduce these limits and support climate adaptation.