This is a data set comprising 29,248 flood events which were sampled from time-series of daily streamflow of 374 catchments across Germany over the period 1950-2010. The flood events were sampled by applying a Peak-Over-Threshold}approach on each catchment, separately. For each of the events, a set of preconditions was collected: First, preconditions in catchment averages of daily precipitation sum [mm/d], daily mean soil moisture [\%] and daily mean temperature of each flood event were sampled: Going back in time from the day of the flood event, the means of predictor values over multiple time periods, "delta t" = [0, 1, 3, 5, 7], were extracted. Note that P is the cumulative sum of rainfall and snowmelt. Both P and SM are model output from the "Mesoscale Hydrological Model" (see references below). Second, data on 10 catchment attributes of various types, i.e. static predictors, are included for each of the catchments. These include data on the average climatic conditions (aridity index, mean annual precipitation), topography (slope, elevation), geomorphology (channel slope, drainage density).

Further Information to the paper that originates from this dataset will be linked in a future version of this dataset. The code for the above paper can be found at https://gitlab.com/lennartschmidt/floodmagnitude/.

{"references": ["Zink, Matthias, et al. \"A high-resolution dataset of water fluxes and states for Germany accounting for parametric uncertainty.\" (2017)."]}