hide
Free keywords:
-
Abstract:
The observed near-surface warming in the Arctic has been four times faster than its global counterpart. This Arctic amplification (AA) of global warming is also found in CMIP5 and CMIP6 simulations, but at a rate that is both forcing- and model-dependent. The present study aims at using global and regional temperature observations and reanalyses, as well as reconstructions of the Northern Hemisphere sea-ice and snow cover extents to constrain the projections of these relevant climate change indices. The method is based on a Bayesian statistical R package called KCC (Kriging for Climate Change). The focus is mostly on the October-to-March season with a maximum AA, but similar qualitative results have been obtained with annual mean timeseries. Results show that global versus Arctic temperature observations lead to contrasted but complementary constraints on the projected Arctic warming. When AA is estimated as the additional polar warming north of 60°N (compared to global warming), AA uncertainties are narrowed by about 30% (after versus before constraint) at the end of the 21st century. This constraint has however a weak impact on the projected response of the northern midlatitude circulation given the limited contribution of AA to the inter-model spread in the projected response of large-scale dynamics.