Seasonal predictability of sea ice and boreal fire in Alaska

Abstract

Attribution studies have shown that warming Arctic temperatures have led to a substantial loss of sea ice and an increase in large wildfire seasons. With wildfire and sea ice decline impacting everyday life by disrupting transportation or damaging infrastructure, for example, a clear need for skillful forecasts at seasonal timescales has emerged. This dissertation addresses critical challenges in seasonal prediction for wildfire and sea ice in Alaska. The first study evaluates outlooks of Buildup Index derived from three seasonal forecasts to assess their skill in predicting fire conditions. Forecast skill is greatest during the wind (April 1-June 10) and drought (July 21-August 9) fire subseasons and in the Western Boreal subregion of Alaska. Combining the forecasts into a multimodel ensemble substantially improves skill. Next, the same seasonal forecasts undergo a comprehensive evaluation of temperature and precipitation in Alaska. While forecasts exhibit similar biases in average precipitation, biases in average temperature vary among the models. However, wrong forecasts for all models tend to forecast temperature anomalies that are too warm. Case studies of good and bad forecast years suggest that forecast skill is influenced by the strength of teleconnection indices relevant to Alaska climate. The final study quantifies impacts of anomalous ocean heat content (OHC) in the Bering Sea and ocean heat transport through the Bering Strait on September Arctic sea ice concentrations (SIC). The addition of one-time OHC anomalies in the Bering Sea results in enhanced SIC decreases along the shallow continental shelves near Alaska but increases on the order of ~10% in the Central Arctic. Increasing SIC is modeled by suppressed upward heat flux and atmospheric forcing. This research was conducted in collaboration with community partners, and their sustained engagement ensured that the resulting tools and insights were scientifically robust and directly relevant to operational decision making.