Geographisches Institut


EXAR Long-Term Weather Data

A collaborative project of the University of Bern (S. Brönnimann, C. Raible) and the Institute for Atmospheric and Climate Science, ETH Zurich (C. Schär)

The goal of the EXAR project [] is to review the current information for assessment of the hazard from extreme flood events in the Aare and Rhine river catchments. This entails quantifying flood event with recurrence periods longer than 1000 years. To test various approaches for determining the corresponding precipitation event, the University of Bern and ETH Zurich provide long data sets of temperature and precipitation derived for Switzerland based on model simulations and reanalyses (EXAR Long-Term Weather Data). The data are made publicly available here on this website.

The data can be used to evaluate stochastic (and other) methods within the EXAR project. The data were specifically designed to

  • test the performance for weather generator for high return period by calibration and validation within climate model simulations,
  • test the stability of the underlying distributions across ensemble members and different data sets, and
  • test the effect of low-frequency variability and climate change on the distributions (and thus on the behavior of weather generators).

The data cannot, under any circumstance, be used to directly calculate high return levels such as a 10’000 yr precipitation event. This is due to the fact that the underlying models have model-related biases due to the simplified representation of physical processes, orography, the land surface or other important factors.

The data sets consist of long model simulations on various scales and in various set-ups (simulations of the past and the future, coupled and uncoupled simulations, global and regional simulations, single member or ensemble studies). In addition, reanalyses and downscaled reanalyses are also provided. In total, over 30,000 simulation years are provided.

All data sets were downscaled to three target regions (the Aare catchment in Brugg, Reuss, Limmat, and the combination of the three catchments) using quantile mapping. The raw data as well as the downscaled data are provided.