GDNat: a global, daily, high-resolution, natural-forcing only temperature data set for attribution research

Abstract. Attribution studies quantify anthropogenic contributions to meteorological changes and commonly rely on approaches that compare historical observations with global climate model (GCM) simulations that are driven solely by natural forcing (Bindoff et al., 2013). However, GCM simulations have limited spatial resolution and are biased by parameterized climate processes and uninitialized conditions that lead to the lack of representation of historical meteorological events (Cannon et al., 2015; Eyring et al., 2016; Almazroui, 2021; Zhang et al., 2023). In addition, attributing climate extremes to human activity requires assessing these events in an unperturbed climate system. We address these gaps by developing a novel high-resolution dataset that provides daily average global temperatures over the past four decades without the influence of anthropogenic climate forcing. We use quantile delta mapping (QDM), a quantile trend-preserving bias adjustment method, to remove anthropogenic warming from the fifth generation of the European Centre for Medium-Range Weather Forecasts Reanalysis (ERA5) using historical and natural-forcing-only simulations from the Coupled Model Intercomparison Project Phase 6 (CMIP6). The resulting dataset consists of historical Global Daily Natural temperature (henceforth, GDNat) records at 0.25° x 0.25° spatial resolution from 1979–2020, providing a valuable resource for attributing extremes and their impacts to anthropogenic warming. We demonstrate a regional application of GDNat by assessing changes in the number of days in the year where daily average temperatures reach or exceed 35°C over North Africa, the Middle East, South Asia, and Australia. We find an average of an additional 15–30 days per year attributable to anthropogenic to warming over the 2000 to 2020 period.