Understanding the predictability of the Arctic climate

Folmer Krikken

Research output: Thesisinternal PhD, WU


The barren and inhospitable Arctic region has over recent decades seen large changes in its natural environment. Observations have shown that the Arctic is warming twice as fast compared to the rest of the world, mostly noticed by the strong decrease of sea ice. These changes present large threats to the unique Arctic ecosystem and indigenous communities, but also provides opportunities such as improved navigation of the fabled northerly passages and exploration of natural resources. Hence, from a scientific, ecological and an economic viewpoint there is a strong need for accurate knowledge on future development of the Arctic climate, and specifically its sea ice cover. This thesis therefore focuses on the predictability of the Arctic climate on time scales ranging from seasonal to centennial, with an emphasis on the physical processes that give rise to, or inhibit, this predictability. This is achieved by studying the physical mechanisms related to Arctic climate variability and climate change, both in climate models and observations.

Over recent years there has been an increase in using fully coupled climate models for seasonal to decadal predictions. Hence, it is important to understand the physical processes that provide predictability beyond persistence of sea ice anomalies in these climate models. In chapter 2 we analyze the natural variability of Arctic sea ice from an energy budget perspective in multiple climate models and compare these results to observations. The Arctic energy balance components primarily indicate the important role of the ice–albedo feedback, through which sea ice anomalies in the melt season reemerge in the growth season. The role of the ocean lies mainly in storing heat content anomalies in spring and releasing them in autumn. Confirming a previous (observational) study, we demonstrate that there is delayed atmospheric response of clouds in autumn to spring sea ice anomalies. Hence, there is no cloud–ice feedback in late spring and summer, but there is a cloud–ice feedback in autumn, which strengthens the ice–albedo feedback. Anomalies in insolation are, counter-intuitively, positively correlated with sea ice variability. This is primarily a result of reduced multiple reflection of insolation due to an albedo decrease. This effect counteracts the ice-albedo effect up to 50%. Reanalysis products confirm the main findings from the climate models.

Observed and projected climate warming is strongest in the Arctic regions, peaking in autumn/winter. Attempts to explain this feature have focused primarily on identifying the associated climate feedbacks, particularly the ice-albedo and lapse-rate feedbacks. In chapter 3, we use a global climate model in idealized seasonal forcing simulations to show that Arctic warming (especially in winter) and sea ice decline are particularly sensitive to radiative forcing in spring, during which the energy is effectively 'absorbed' by the ocean (through sea ice melt and ocean warming, amplified by the ice-albedo feedback) and consequently released to the lower atmosphere in autumn and winter, mainly along the sea ice periphery. In contrast, winter radiative forcing causes a more uniform response centered over the Arctic Ocean. This finding suggests that intermodel differences in simulated Arctic (winter) warming can to a considerable degree be attributed to model uncertainties in Arctic radiative fluxes, which peak in summer.

The intermodel differences in projected Arctic warming are very large, owing to considerable differences between climate models. A clear understanding of this large uncertainty is currently lacking. In chapter 4 we use global climate models to show that springtime interannual variability in downwelling longwave radiation in the pre-industrial climate explains about two-thirds of the intermodel spread in projected Arctic warming under a high greenhouse gas emission scenario. This variability, which peaks on the land masses adjacent to the Arctic ocean, is related to the combined effects of the cloud radiative forcing and the albedo response to snowfall, which vary strongly among models in these regions. These processes govern interannual variability of downwelling longwave radiation in the pre-industrial climate, but also largely modulate the Arctic climate response. This finding elucidates the crucial interaction between clouds and surface radiation within the Arctic climate system. As such it provides important insights into possible reductions in the uncertainty in future Arctic climate projections that are required to constrain regional mitigation and adaptation strategies to Arctic climate change.

In chapter 5 we explore the error and improve the skill of the outcome from dynamical seasonal Arctic sea ice reforecasts using different bias correction and ensemble calibration methods. These reforecasts consist of a five-member ensemble from 1979 to 2012 using the general circulation model EC-Earth. The raw model forecasts show large biases in Arctic sea ice area, mainly due to a differently simulated seasonal cycle and long-term trend compared to observations. This translates very quickly (1–3 months) into large biases. We find that (heteroscedastic) extended logistic regressions are viable ensemble calibration methods, as the forecast skill is improved compared to standard bias correction methods. Analysis of regional skill of Arctic sea ice shows that the Northeast Passage and the Kara and Barents Sea are most predictable. These results demonstrate the importance of reducing model error and the potential for ensemble calibration in improving skill of seasonal forecasts of Arctic sea ice.

Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Wageningen University
  • Hazeleger, W., Promotor
  • Bintanja, R., Co-promotor, External person
Award date5 Sep 2018
Place of PublicationWageningen
Print ISBNs9789463434850
Publication statusPublished - 2018

Fingerprint Dive into the research topics of 'Understanding the predictability of the Arctic climate'. Together they form a unique fingerprint.

Cite this