This dataset provides 50 years of seasonal synthetic hourly time series for five meteorological variables: incoming shortwave radiation, total precipitation, relative humidity, 2-meter air temperature and wind speed.

The data are generated from observational records (1990–present) from nine MeteoSwiss stations (ABO, BAS, BER, LUG, PUY, SCU, SIO, STG, WFJ). We use a sampling approach that preserves both the temporal structure and inter-variable dependencies. Specifically, the time series are constructed by drawing 7-day blocks from historical observations, ensuring realistic short-term variability and maintaining the physical coherence between variables.

The synthetic sequences are constrained to follow daily temperature change trajectories corresponding to three future climate scenarios for the period 2040–2070: RCP2.6, RCP8.5 and RCP0: a counterfactual scenario with no long-term warming trend, but with temperature variability consistent with the RCP2.6 one. The temperature constraints are derived from projections downscaled at the staiton level, available at: https://www.envidat.ch/#/metadata/climate-change-scenarios-at-hourly-resolution