An Observational Estimate of the Pattern Effect on Climate Sensitivity: The Importance of the Eastern Tropical Pacific and Land Areas

Abstract The patterns associated with the top-of-the-atmosphere radiative response R to surface temperature T are typically explored through two relationships: 1) the spatially varying radiative response to spatially varying changes in temperature (Δ R i /Δ T i ) and 2) the spatially varying radiative response to global-mean changes in temperature (Δ R i /Δ T ). Here, we explore the insights provided by an alternative parameter: the global-mean radiative response to changes in spatially varying temperature (Δ R /Δ T i ). The pattern Δ R /Δ T i indicates regions where surface temperature covaries with R and thus provides a statistical analog to the causal response functions derived from atmospheric Green’s function experiments. The pattern can be transformed so that it can be globally averaged and thus indicates the local contribution to the global feedback parameter. The transformed version of Δ R /Δ T i corresponds to the pattern in surface temperature whose expansion coefficient time series explains the maximum fraction of the covariance between R and T i . It explains roughly the same fraction of internal variability in R as that explained by the Green’s function approach. We focus on the physical insights provided by Δ R /Δ T i when it is estimated from regression analyses of monthly mean observations. Consistent with the results of Green’s function experiments, the observational analyses indicate negative contributions to the global internal feedback parameter over the western Pacific and positive contributions over the southeastern tropical Pacific. Unlike the results of such experiments, the analyses indicate notable negative contributions to the global feedback parameter over land areas. When estimated from observations, temperature variability over the land areas accounts for ∼70% of the negative global internal feedback, whereas variability over the southeastern tropical Pacific reduces the global-mean negative internal feedback by ∼10%.