Eurasian Cooling Patterns in the CMIP5 Climate Models

TL;DR

This paper evaluates how well CMIP5 climate models simulate Eurasian winter cooling and its link to Arctic sea ice loss, finding that many models fail to reproduce observed patterns due to inaccuracies in sea ice simulation.

Contribution

It provides a comprehensive comparison of CMIP5 models' ability to replicate Eurasian cooling and its connection to Arctic sea ice variability, highlighting key model deficiencies.

Findings

Most models show a pattern of co-variability similar to reanalysis.

Many models fail to reproduce observed winter cooling.

Inaccurate simulation of sea ice in key regions may explain discrepancies.

Abstract

The Arctic has warmed dramatically compared to the global average over the last few decades. During this same period, there have been strong cooling trends observed in the wintertime, near-surface air temperature over central Eurasia, a phenomenon known as Eurasian cooling. Many studies have suggested that the loss of sea ice, especially in the Barents and Kara Seas, is related to the cooling over Eurasia, although this connection and its possible mechanism is still a source of heated debate. Observations and reanalyses show a clear pattern of co-variability between Arctic sea ice and Eurasian wintertime temperatures. However, there is an open question regarding how robustly this teleconnection pattern is reproduced in the current generation of climate models. This study has examined Eurasian cooling in twenty models from the Coupled Model Intercomparison Project Phase 5 (CMIP5), both in terms of temperature trends and by using singular value decomposition to identify patterns of co-variability between Arctic sea ice concentrations and near-surface temperatures over Eurasia. These are compared to trends and patterns of co-variability found in the ERA-Interim reanalysis. While most of the CMIP5 models have a robust pattern of co-variability that is similar to that found in the reanalysis, many fail to reproduce the wintertime cooling that has been observed. An examination of Arctic sea ice extent in the models suggests one possible explanation is the inability of the models to accurately simulate the wintertime sea ice changes over critical locations such as the Barents and Kara Seas.