The influence of Arctic sea-ice loss on mid-latitude weather and climate: exploring sensitivities and mechanisms
Over the past few decades, Arctic sea-ice extent has declined, while there has been an apparent increase in severe winter weather across some mid-latitude regions. This has led to much research into whether these trends are dynamically linked. It has been suggested that the link may involve the Arctic Oscillation (AO), which describes the observed oscillation in geopotential height anomalies between high and middle Northern Hemisphere latitudes. Sea-ice loss has been shown to excite the AO’s negative phase, which is linked to colder conditions in key regions of mid-latitudes, through various tropospheric and stratospheric mechanisms. However, the nature of the response to Arctic sea-ice loss and the mechanisms involved remain uncertain. This is because it is difficult to disentangle the complex web of potential processes involved, the modelled response to sea-ice loss is small relative to internal climate variability, and modelling studies find contrasting climatological mean responses to imposed sea-ice loss. Since all climate models project a continuation of Arctic sea-ice loss during the 21st century in response to anthropogenic greenhouse gas forcing, it is important that the potential influence of this on the highly populated mid-latitudes is better understood. In this thesis, the issues of complexity and statistical robustness are partly addressed by conducting idealised numerical modelling experiments using an intermediate complexity global circulation model, IGCM4. Such models are useful because they are complex enough to simulate a variety of important processes, but are relatively simple and computationally fast compared to full complexity state-of-the-art climate models. This helps to disentangle different processes from one another and allows for several-century-long simulations, making a statistically robust response more attainable. To understand the contrasting results of past modelling studies, it is firstly investigated whether the response to sea-ice loss is sensitive to the loss region (the Atlantic or Pacific sector of the Arctic). For different regions of loss, different effects on the
stratospheric circulation are found. While there are negative tropospheric AO responses in both cases, there are contrasting effects on mid-latitude surface temperatures. This is explained in this work using a method of decomposition into an ‘indirect’ part induced by the large-scale AO response, and a residual ‘direct’ part that is local to the ice loss region. A low signal-to-noise ratio makes it difficult to robustly determine the linearity of the response to different loss magnitudes. A stratospheric nudging method is then implemented in IGCM4 to isolate the roles played by tropospheric and stratospheric mechanisms in the remote response to sea-ice loss. For Atlantic sector loss, part of the negative tropospheric AO response is found to likely be caused by tropospheric mechanisms, and the other part likely involves changes in sudden stratospheric warmings (SSWs). For Pacific sector loss, there is likely a non-linear interaction between tropospheric and stratospheric mechanisms, where the stratospheric state alters vertical wave propagation such that the direct stationary Rossby wave response to the ice loss projects onto a negative tropospheric AO. Finally, motivated by the importance of SSWs in the Atlantic sector sea-ice loss experiment and their potentially large internal variability, this experiment is extendedby several centuries to examine the influence of atmospheric internal variability with regard to uncertainty in responses to sea-ice loss produced by model time-slice experiments of different lengths. This leads to a quantification of the minimum experiment length required to separate the signs of forced tropospheric and stratospheric changes due to sea-ice loss from internal variability. This has not been quantified to date for the latter, and is found to be large for both the stratospheric AO and SSW frequency(respectively around190and450yearsfortheDecember-Marchmean). This may explain contrasting stratospheric responses in past studies using an insufficient experiment length, with implications for the robustness (at least quantitatively) of the tropospheric responses in these studies. Here, the responses are qualitatively the same in the shorter and extended experiments, but there are some differences in magnitude and evolution. In summary, this thesis improves understanding of the influence of Arctic sea-ice loss on mid-latitude weather and climate, and the mechanisms involved. This is done by systematically examining various aspects that may explain contrasting model results – including different regions and magnitudes of loss, as well as atmospheric internal variability – and, hence, current uncertainty regarding the nature of a link. A better understanding of the mechanisms involved is obtained by decomposing the responses into parts due to tropospheric and stratospheric mechanisms.
Authors: McKenna, Christine Mary ORCID record for Christine Mary McKenna