Data supporting "Sharper constraints without overconfidence: a scoring rule for perfect-model tests and climate projection weighting"

Public data supporting V. Del Tatto, A. Gavrilov and G. Camps-Valls, Sharper constraints without overconfidence: a scoring rule for perfect-model tests and climate projection weighting. The archive contains the three inputs to the analysis: projections — global land-mean precipitation under historical + SSP5-8.5 for 95 CMIP6 (model, member) pairs, as annual values over 1995–2100. causal_graphs — time-series causal graphs inferred with PCMCI from historical sea-level pressure, per season, for ERA5 and 154 CMIP6 members (60 reduced modes, lags τ = 0…20 days). psl_historical_monthly — monthly-mean sea-level pressure, 1979–2014, for ERA5 and 152 CMIP6 members, regridded to 2.5°×2.5° between 87.5°S and 87.5°N and linearly detrended at each grid cell, together with the ESMValTool v2.14.0 recipe that produced them. The causal graphs and the precipitation projections were generated by Debeire et al. (2025, Earth Syst. Dynam. 16, 607–630) and are reused here with thanks. The sea-level pressure fields were produced for the present work from the same CMIP6 and ERA5 sources. Analysis code: https://github.com/vdeltatto/climwip-crps — place this unzipped directory (data/) at the repository root. See README.md in this record for the full file-level description.