Fig. 2: Constructing the Climate Adaptation Feedback (CAF).

The CAF is constructed by combining high-resolution projections of climate change impacts on energy consumption from ref. ⁸ with the following emissions factors: (a) country-level carbon dioxide (CO₂) intensities from a gigajoule (GJ) of electricity use (in tons tCO₂ × GJ⁻¹); and (b) country-level CO₂ intensities for all other fuels combined (tCO₂ × GJ⁻¹). Maps (produced using Stata software with the spmap package using GADM shapefiles⁷³) show country-level average factor values over the 2010–2018 period. Together, these data allow us to in (c) compute the mean value (solid line) and 90 percent confidence intervals (shaded region) of cumulative adaptation-induced CO₂ emission changes (in gigatonnes of CO₂) for 2020–2099 under our baseline scenario, which pairs Shared Socioeconomic Pathway 2 (SSP2) with Representative Concentration Pathway 8.5 (RCP8.5). Finally, we estimate (d) the relationship between projected GMST change (in °C) and cumulative CO₂ emissions across RCPs and global climate models over 2020–2099 (see “Methods”). This plot shows a fitted linear model (solid line) with 90% confidence intervals (shaded area) and point estimate and p-value of the linear coefficient, as well as a local polynomial fit (dashed line) using an Epanechnikov kernel with a rule-of-thumb bandwidth⁷¹ (N = 5200).