The impact of long-term memory on the climate response to greenhouse gas emissions

Abstract

Global warming exerts a strong impact on the Earth system. Despite recent progress, Earth System Models still project a large range of possible warming levels. Here we employ a generalized stochastic climate model to derive a response operator which computes the global mean surface temperature given specific forcing scenarios to quantify the impact of past emissions on current warming. This approach enables us to systematically separate between the "forcing-induced direct" and the "memory-induced indirect" trends. Based on historical records, we find that the direct-forcing-response is weak, while we attribute the major portion of the observed global warming trend to the indirect-memory responses that are accumulated from past emissions. Compared to CMIP6 simulations, our data-driven approach projects lower global warming levels over the next few decades. Our results suggest that CMIP6 models may have a higher transient climate sensitivity than warranted from the observational record, due to them having larger long-term memory than observed.