# Data to accompany "Sea-surface temperature pattern effects have slowed global warming and biased warming-based constraints on climate sensitivity" Data for "Sea-surface temperature pattern effects have slowed global warming and biased warming-based constraints on climate sensitivity", PNAS () ## Description of the data and file structure Included are a netcdf data file needed to make the figures. This includes the following variables (with descriptions). CMIP56_ECS: CMIP5/6 ECS values; units K CMIP56_subset_ECS: CMIP5/6 8 model subset ECS values (CanESM5, CNRM CM6, GISS E2, HadGEM3, IPSL CM6A, MIROC6, NorESM2, CESM1); units K CMIP56_subset_EffCS_amip: CMIP5/6 8-model subset EffCS values for AMIP models over 1981-2014 (CanESM5, CNRM CM6, GISS E2, HadGEM3, IPSL CM6A, MIROC6, NorESM2, CESM1); units K CMIP56_subset_tas_trend_amip_method1: CMIP5/6 8 model subset global near-surface air temperature trend over 1981-2014 estimated from EffCS in AMIP models and regression (CanESM5, CNRM CM6, GISS E2, HadGEM3, IPSL CM6A, MIROC6, NorESM2, CESM1); units K/dec CMIP56_subset_tas_trend_amip_method2: CMIP5/6 8 model subset global near-surface air temperature trend over 1981-2014 estimated from EffCS in AMIP models and equation 3 (CanESM5, CNRM CM6, GISS E2, HadGEM3, IPSL CM6A, MIROC6, NorESM2, CESM1); units K/dec CMIP56_tas_trend: CMIP56 global near-surface air temperature trend over 1981-2014; units K/dec CanESM5_EffCS: CanESM5 EffCS over 1981-2014; units K CanESM5_tas_trend: CanESM5 global near-surface air temperature trend over 1981-2014; units K/dec CESM1_EffCS: CESM1 EffCS over 1981-2014; units K CESM1_tas_trend: CESM1 global near-surface air temperature trend over 1981-2014; units K/dec CNRM_CM6_EffCS: CNRM_CM6 EffCS over 1981-2014; units K CNRM_CM6_tas_trend: CNRM_CM6 global near-surface air temperature trend over 1981-2014; units K/dec GISS_E2_EffCS: GISS_E2 EffCS over 1981-2014; units K GISS_E2_tas_trend: GISS_E2 global near-surface air temperature trend over 1981-2014; units K/dec HadGEM3_EffCS: HadGEM3 EffCS over 1981-2014; units K HadGEM3_tas_trend: HadGEM3 global near-surface air temperature trend over 1981-2014; units K/dec IPSL_CM6A_EffCS: IPSL_CM6A EffCS over 1981-2014; units K IPSL_CM6A_tas_trend: IPSL_CM6A global near-surface air temperature trend over 1981-2014; units K/dec MIROC6_EffCS: MIROC6 EffCS over 1981-2014; units K MIROC6_tas_trend: MIROC6 global near-surface air temperature trend over 1981-2014; units K/dec NorESM2_EffCS: NorESM2 EffCS over 1981-2014; units K NorESM2_tas_trend: NorESM2 global near-surface air temperature trend over 1981-2014; units K/dec

The observed rate of global warming since the 1970s has been proposed as a strong constraint on equilibrium climate sensitivity (ECS) and transient climate response (TCR) – key metrics of the global climate response to greenhouse-gas forcing. Using CMIP5/6 models, we show that the inter-model relationship between warming and these climate sensitivity metrics (the basis for the constraint) arises from a similarity in transient and equilibrium warming patterns within the models, producing an effective climate sensitivity (EffCS) governing recent warming that is comparable to the value of ECS governing long-term warming under CO2 forcing. However, CMIP5/6 historical simulations do not reproduce observed warming patterns. When driven by observed patterns, even high ECS models produce low EffCS values consistent with the observed global warming rate. The inability of CMIP5/6 models to reproduce observed warming patterns thus results in a bias in the modeled relationship between recent global warming and climate sensitivity. Correcting for this bias means that observed warming is consistent with wide ranges of ECS and TCR extending to higher values than previously recognized. These findings are corroborated by energy balance model simulations and coupled model (CESM1-CAM5) simulations that better replicate observed patterns via tropospheric wind nudging or Antarctic meltwater fluxes. Because CMIP5/6 models fail to simulate observed warming patterns, proposed warming-based constraints on ECS, TCR, and projected global warming are biased low. The results reinforce recent findings that the unique pattern of observed warming has slowed global-mean warming over recent decades, and that how the pattern will evolve in the future represents a major source of uncertainty in climate projections.