The Arctic amplification of global warming is a well-known feature of climate
projections, which is particularly prominent from November to March. It is partly driven by the Arctic sea-ice response to global warming, but other potential mechanisms may also contribute. Their net effect and the sea-ice feedback both remain highly model-dependent, even in the latest generation of climate models. Given the increasing evidence of the human influence on global and regional near-surface temperatures, it is now possible to constrain the climate projections with the available instrumental record.
The use of statistical methods to narrow the range of climate projections is a growing but still empirical field. Recent studies have show for instance that linear-regression methods based on emergent relationships between observable and future climate variables may lead to over-confident projections. Here we use a Bayesian statistical method which is called Kriging for Climate Change (KCC). It allows us to derive a posterior distribution of the projected additional Arctic warming compared to its global counterpart based on a prior distribution derived from climate model outputs and conditionned on the HadCRUT5 temperature observations. In doing so, we first isolate the forced model response to the anthropogenic versus natural forcings and we account for both model and instrumental uncertainties to derive the posterior distribution.
Results show that the recently observed four-fold ratio between the Arctic and global warming rates is mostly but not entirely due to a human influence, and will decrease with increasing radiative forcings. Global versus regional temperature observations lead to complementary constraints on the projections. When Arctic amplification is defined as the additional polar warming relative to global warming, model uncertainties are narrowed by 30% after constraint. Similar results are obtained for projected changes in the Arctic sea ice extent (40%) and when using sea ice concentration and polar temperature observations to constrain the projected polar warming (37%), thereby confirming the key role of sea ice as a positive but model-dependent surface feedback.
With the growing emergence of anthropogenic climate change in instrumental and satellite records, it becomes possible not only to detect and attribute the human-induced component of observed changes in regional climates, but also to use the available observations to constrain climate projections. The KCC method used in the present study takes advantage of the entire observational record to constrain past and future responses to various external forcings in a consistent way. It has been tested successfully against synthetic observations and has been already applied in a number of global and regional studies, including this one on Arctic amplification. It will provide even more tightly constrained projections as soon as more reliable and/or longer observational timeseries are available.
- Ribes, A., Qasmi, S., Gillett, N. Making climate projections conditional on historical observations. Science Advances, 7, eabc0671 (2021).
- Qasmi, S. & Ribes, A. Reducing uncertainty in local temperature projections. Science Advances, 8, eabo6872 (2022).
- Douville H., S. Qasmi, A. Ribes, O. Bock (2022) Global warming at near-constant tropospheric relative humidity is supported by observations. Communications Earth & Environment, 3, 237, https://doi.org/10.1038/s43247-022-00561-z
- Ribes A., J. Boé, S. Qasmi, B. Dubuisson, H. Douville, L. Terray (2022) An updated assessment of past and future warming over France based on a regional observational constraint. Earth System Dynamics, esd-2022-7, https://doi.org/10.5194/esd-2022-7
- Douville H. and K. Willett (2023) A drier than expected future, supported by near-surface relative humidity observations. Science Advances, https://doi.org/10.1126/sciadv.ade6253
- Dutot E. and H. Douville (2023) Revisiting the potential to narrow uncertainty in annual runoff projections of Arctic rivers. Geophys. Res. Lett., 50, e2023GL104039, https://dx.doi.org/10.1029/2023GL104039