New Publication | Earth System Science Data
What gets lost when a whole day is just one number?
Please check out the video abstract on https://av.tib.eu/media/73819.
A heatwave does not hold the same intensity all day, a thunderstorm does not drizzle evenly from midnight to midnight, and a wind turbine does not care about the daily mean wind speed. Many of the processes that climate impact models simulate, from human heat stress and surface runoff to solar and wind power generation, respond non-linearly to what happens within a day. The global climate projections that feed these models, however, are archived at daily resolution for reasons of storage, so that no globally consistent, bias-corrected hourly climate projections have so far been available to the impact modelling community. Writing in Earth System Science Data, Michel Bechtold (KU Leuven), Benjamin Poschlod (TUHH, IGWS) and colleagues at the Potsdam Institute for Climate Impact Research and the University of Basel now close that gap.
The team disaggregated the entire daily climate archive of the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP3) to hourly resolution using Teddy, the Temporal Disaggregation Tool co-developed by Poschlod. Instead of interpolating, Teddy searches a bias-corrected hourly reanalysis for the historical day that most closely resembles each simulated day at that location, and transfers its observed diurnal profile. Because the profiles of all variables are taken from the same analogue day, the timing between temperature, humidity, radiation, wind and rainfall stays physically coherent, while the daily mass and energy totals of the climate model are reproduced exactly. The result covers seven variables on a global 0.5° land grid, for five CMIP6 climate models, three emission scenarios and the years 1850 to 2100. Producing it took roughly 7 million CPU hours and amounts to 16 TB of openly available data.
Four example applications show what the added resolution buys. Hourly data reveal rainfall intermittency that daily wet-day counts obscure, and they allow the diurnal cycles of temperature and humidity to be combined into a heat index: in the tropics, hours with dangerous heat already average more than 1500 per year, more than double by the end of the century under a high-emission scenario, and spread polewards into regions that are not affected today. For wind and photovoltaic power, the authors computed the same metrics once from daily and once from hourly forcing. The temporal resolution changed not only the magnitude of the result, but also the projected climate change signal, in a way that depends on region and metric and cannot be anticipated from the daily data alone. Dataset and code are freely available.
Read the full open-access article: Bechtold, M., Poschlod, B., Otto, C., Volkholz, J., Büchner, M., and Zabel, F. (2026): A global hourly ISIMIP3 climate forcing dataset for impact modeling. Earth System Science Data 18, 6637–6648. https://doi.org/10.5194/essd-18-6637-2026
Dataset (ISIMIP Repository): https://doi.org/10.48364/ISIMIP.736682 (ISIMIP3a) and https://doi.org/10.48364/ISIMIP.170328 (ISIMIP3b)
