Climate Models Underestimate the Sensitivity of Arctic Sea Ice to Carbon Emissions

TL;DR

This paper shows that climate models underestimate how sensitive Arctic sea ice is to carbon emissions, implying the Arctic may become ice-free sooner than predicted by current models.

Contribution

It provides empirical evidence that observed Arctic sea ice decline is more strongly linked to carbon emissions than climate models suggest, highlighting a gap in model accuracy.

Findings

Observed data shows a stronger ice-carbon relationship than models.

Climate models underestimate Arctic sea ice sensitivity to emissions.

No significant improvement from CMIP5 to CMIP6 in modeling this sensitivity.

Abstract

Arctic sea ice has steadily diminished as atmospheric greenhouse gas concentrations have increased. Using observed data from 1979 to 2019, we estimate a close contemporaneous linear relationship between Arctic sea ice area and cumulative carbon dioxide emissions. For comparison, we provide analogous regression estimates using simulated data from global climate models (drawn from the CMIP5 and CMIP6 model comparison exercises). The carbon sensitivity of Arctic sea ice area is considerably stronger in the observed data than in the climate models. Thus, for a given future emissions path, an ice-free Arctic is likely to occur much earlier than the climate models project. Furthermore, little progress has been made in recent global climate modeling (from CMIP5 to CMIP6) to more accurately match the observed carbon-climate response of Arctic sea ice.