GDNAT: Global Daily Natural forcing only Temperature Dataset

GDNat is a global daily counterfactual surface air temperature dataset for climate change attribution research: daily average near-surface air temperature at 0.25-degree resolution from 1979 to 2020 under natural forcing only. Quantile Delta Mapping is used to remove the estimated anthropogenic signal, which we obtain as the quantile-based difference between CMIP6 historical simulations and natural-forcing-only historical simulations at each location and timestep. These differences are then applied to ERA5 at each location and timestep to remove the anthropogenic signal. Therefore, differencing ERA5 against GDNat gives the portion of an observed temperature metric attributable to human forcing.

GDNat was produced by the Climate Impact Lab. Full methodology, validation, and limitations are in the associated data paper, while the methodological pipeline is at https://github.com/ClimateImpactLab/gdnat.

GDNat is meant to serve as a counterfactual, not an observation. It estimates temperatures under natural forcing alone and must not be used as a record of observed climate. ERA5 is the corresponding record we used as the 'factual' world. GDNat is comprised of six members, which are associated with their corresponding CMIP6 counterpart and are meant to be used together. The spread across them represents the uncertainty on the anthropogenic signal, so a result from a single member has no uncertainty attached to it.