This dataset comprises synthetic weather data generated for historical (“control” present, 1985-2014) and two future periods (near future: 2031-2060 (period1) and far future: 2071-2100 (period2)) across a domain encompassing Germany and its neighboring riparian countries. The dataset was produced through the following key steps:

(1) Classifying Weather Circulation Patterns for the Observed/Present Period: Weather circulation patterns (CPs) were classified for a European domain (35°N – 70°N, 15°W – 30°E), and regional average temperatures at 2 m height (t2m) were calculated for the German domain (45.125°N – 55.125°N, 5.125°E – 19.125°E). This classification used mean sea level pressure (psl) and mean temperature (tas) data from the ERA5 dataset provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) (Hersbach et al., 2020).

(2) Training Non-Stationary Climate-Informed Weather Generator (nsRWG): The nsRWG (Nguyen et al., 2024), conditioned on the classified CPs and using tas as a covariate, was set up and trained for the German domain using the E-OBS dataset, version 25.0e (Cornes et al., 2018). This training dataset includes 540 grid cells of mean daily temperature and precipitation totals for the period 1950–2021, with a spatial resolution of 0.5° x 0.5°.

(3) Generating Data for the Present Period: Long-term synthetic data for the present period is generated using the trained nsRWG.

(4) Assigning Circulation Patterns for Future Periods: The classified CPs from the present period were assumed to remain stable in the future. These CPs were assigned to future periods based on mean sea level pressure data from nine selected general circulation models (GCMs) from CMIP6 (Eyring et al., 2020) for the two future periods and two shared socio-economic pathways: SSP245 and SSP585 (IPCC, 2023). In total, CPs were derived for 36 scenarios, and regional average temperatures were also computed.

(5) Downscaling Data for Future Scenarios: The nsRWG was used to statistically downscale long-term synthetic weather data for all 36 future scenarios.

(6) Final dataset: The dataset includes synthetic weather data generated for the present period (Step 3) and future scenarios (Step 5).
This dataset is expected to offer a key benefit for hydrological impact studies by providing long-term (thousands of years) consistent synthetic weather data, which is indispensable for the robust estimation of probability changes of hydrologic extremes such as floods.