Seasonal to interannual variability and predictability of Arctic summertime sea ice associated with tropically forced planetary wave patterns

Increases in economic, environmental, and security interests in the Arctic demand improved prediction capabilities. The proposed project will explore a new path towards improved predictions of Arctic sea ice. We will investigate how teleconnections between tropical sea surface temperatures (SST) and high latitude circulation patterns can be exploited for sea ice predictions. Recent climate change in the Arctic is generally attributed to anthropogenic drivers and related feedbacks between sea ice, the ocean, and the atmosphere. However, work by Ding et al. (2014) and others (e.g. Trenberth et al. 2014) suggest that tropical Pacific SST variability is important in modulating recent Arctic climate variability by influencing the high-latitude atmospheric circulation. So far, these papers have examined the teleconnection between tropical SSTs and Arctic circulation and surface air temperatures. One unresolved question is how much does this tropical-Arctic teleconnection affect sea ice variability and predictability? This proposal aims to fill that gap. Indeed, preliminary results offered in this proposal suggest that these links with sea ice exist. The proposed work will focus on the implications of this link for seasonal predictability of sea ice and explore how a hierarchy of models captures this link and can be used and improved to enhance Arctic sea ice prediction.

The main source of sea ice predictability in current state-of-the-art models and statistical forecasts originates from the long-term trend of sea ice. On seasonal and interannual time scales, skillful prediction is related to initial sea ice thickness and the oceanic state. Less attention has been given to sea ice predictability arising from teleconnection in the atmosphere which might connect sea ice to ocean states elsewhere. In this project, we hypothesize that there are states of the high latitude atmospheric circulation that are predictable and that their impact on Arctic summer sea ice can be used to improve sea ice predictions. Specifically, our working hypothesis is that the sea ice state at the sea ice minimum (September) is predictable because it is tied to preceding summer Northern Hemisphere circulation patterns which in turn depend on tropical SSTs. If forecast models can replicate this tropical influence on sea ice, a gain in forecast skill can be expected. Whether or not this hypothesized link is robust and replicated in current models needs to be thoroughly examined and is the primary goal of the proposed work. We pursue these goals by executing the following tasks: We will (1) explore the statistical linkage between tropical SSTs-related, high-latitude circulation, and observed Arctic sea ice; (2) examine the dynamical and thermodynamical mechanisms that contribute to the observed connection between the tropically driven high-latitude circulation variability and sea ice by using the PIOMAS and the NCAR CESM1.2; (3) evaluate the performance of CFSv2 and NMME models in reproducing the tropical–sea-ice teleconnection over the last 30 years, which includes a diagnosis of successes and failures and an assessment of the associated sea ice prediction skill; (4) transfer our knowledge for incorporation into operational predictions within NOAA.

This proposal targets all three priorities of the CVP program solicitation by focusing on mechanisms of tropical-extratropical interaction affecting Arctic sea ice during summer and their relevance to the operational NCEP CFSv2 and NMME models. Moreover, all activities are strictly related to the objectives of the NOAA Next-Generation Strategic Plan of “improving scientific understanding of the changing climate system and its impacts.” The project combines the PIs’ expertise in large-scale climate dynamics (atmosphere-ocean-sea ice interaction) with experience using global climate models, regional sea ice-ocean models, and NOAA’s operational models and products that offer forecaster decision support and assist in real-time monitoring.