1-day precipitation extremes over 2003-2020 in a transboundary Alpine area (Eastern Alps) identified by different methods and climate datasets

The repository contains the date lists, statistics and fields of the topmost extreme daily precipitation events detected over the period 2003-2020 in a transboundary Alpine area including South Tyrol in Italy and East Tyrol and Carinthia in Austria. The events are detected by applying different extreme statistical definitions and different gridded datasets. The extreme events and detection methods are described in detail and discussed in combination with geohydrological hazard records in Crespi et al. (2025, submitted). DATA The gridded precipitation datasets considered for the event detection are: SPARTACUS-TST: 1-km gridded dataset of daily precipitation from 1980 to present based on in-situ observations and derived by merging the gridded products available for Austria (SPARTACUS, Hiebl et al., 2017) and for Trentino-South Tyrol (TST, Crespi et al., 2021); INCA: 1-km gridded dataset of sub-daily precipitation from 2003 to present derived from nowcasting system integrating rain-gauge data, weather radar estimates and detailed topographical information (Ghaemi et al., 2021; Haiden et al., 2011); ERA5-Land: replay of the land component of the global reanalysis ERA5 with a finer spatial resolution of 9 km providing hourly fields of surface variables from 1950 to present (Muñoz-Sabater et al., 2021); CERRA-Land: land components of the CERRA regional reanalysis developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) for the pan-European region, featuring a horizontal resolution of 5.5 km and spanning the period September 1984 to May 2021 (Ridal et al., 2024). All datasets are aggregated daily and masked over the study area. The datasets are reprojected in MGI / Austria Lambert (EPSG: 31287), while grids are kept at the original spatial resolution. To ensure the temporal alignment CERRA-Land fields are shifted back one day, since in the other products the total precipitation is assigned to the start of the cumulation period. METHODS For each dataset and each day from 1st January 2003 to 31st December 2020 the following aggregated statistics are calculated: Areal mean: the average daily precipitation per grid cell (mm), derived by dividing the total precipitation over the grid by the number of grid cells Local 99th percentile: the 99th percentile of the daily precipitation values over all grid cells (mm) Magnitude: product of the mean daily precipitation anomaly over the grid cells and the portion of grid cells experiencing anomalous conditions. The daily anomaly at each grid cell is expressed as the difference from the corresponding daily normal over 2003-2020 divided by the standard deviation. Only grid cells whose daily anomaly is greater than two standard deviations are selected for the event magnitude calculation (Ramos et al., 2014). The three metrics are used individually to sort the daily time series for each dataset. Thus, three sorted lists are produced for each dataset. Based on the metric used, the detection methods are named “areal mean”, “local p99” and “anomaly”. The top 5% dates are then selected from each list (corresponding to 330 dates) to represent the most extreme precipitation events for each combination of metric and dataset. The selection is performed so that consecutive dates (i.e., in a 5-day window centred over the event under consideration) are associated to the same event and count only once. The STATISTICS folder includes one .csv file for each dataset providing the statistics calculated for each day from 1st January 2003 to 31st December 2020 and used to characterize and identify the topmost extreme events. Specifically, columns include the date, areal mean ("areal_mean", mm), local 99th percentile (“local_p99”, mm), local maximum ("maximum", mm), fraction of the area with daily anomaly greater than two standard deviations ("fraction_area"), the average of these upper-threshold anomalies ("mean_anomaly") and the magnitude, i.e., product of the mean anomaly and areal portion (“magnitude”). The EVENTS folder provides the netCDF files reporting the daily precipitation fields of the 330 selected events, for each dataset and detection method. Moreover, a summary table with the sorted lists of the 330 extreme dates for each combination of dataset and method is included ("top_330_precipitation_events_all_methods_datasets.csv").