A173-02
Applying seasonal forecasts for summer fire weather outlooks in Alaska at a 3-month lead

Tuesday, 15 December 2020: 04:06
Virtual
Cecilia Borries Strigle1, Uma Suren Bhatt1, Peter Bieniek2, Heidi Strader3, Eric Stevens3, Alison S York4 and Robert Ziel5, (1)University of Alaska Fairbanks, Fairbanks, AK, United States, (2)International Arctic Research Center, Fairbanks, AK, United States, (3)Alaska Interagency Coordination Center, Fairbanks, United States, (4)University of Alaska Fairbanks, Alaska Fire Science Consortium, Fairbanks, AK, United States, (5)University of Alaska, Fairbanks, Fairbanks, United States
Recorded Presentation
Abstract:
Increasing temperatures in Alaska are expected to extend the wildfire season, enhance fire activity, and therefore increase the costs associated with fire suppression. While the Alaska fire season typically ends by late-July, the 2019 fire season saw two costly late-season fires in August which disrupted road travel, prompted evacuations, and destroyed permanent structures. Persistent drought conditions and high winds helped to fuel the 2019 late season fires, and the estimated suppression costs for both fires were approximately $59 million. Providing fire managers with skillful information regarding the upcoming fire season will inform management decisions which can allocate resources efficiently and reduce costs. This research uses newly available multi-model seasonal forecasts to develop summer fire weather outlooks in March when management information is needed.

As Alaska is part of the Boreal forest, Alaska fire managers use the Canadian Forest Fire Weather Index (FWI) System which was developed especially for boreal ecosystems. The FWI consists of four indices to rate fire danger: the Fine Fuel Moisture Code, the Duff Moisture Code (DMC), the Drought Code (DC), the Buildup Index (BUI). The BUI is calculated from the DMC and DC and represents the flammability of organic material in both the subsurface and deeper soil layers and is well suited for seasonal forecasts. In our study, BUI was calculated from data from three seasonal forecast models - the NCEP CFSv2, ECMWF SEAS5, and Météo France System 6 - and compared to observed BUI values from station data throughout regions in Alaska for the entire Alaska fire season (1 April - 30 September) as well as four individual fire subseasons within that time period. Forecast skill for each Alaska region and subseason was measured by the Relative Operating Characteristic (ROC) score which is a comparison of the hit rate versus false alarm rate.

We found that forecast skill mainly occurred for BUI values forecast in the upper tercile throughout all regions and subseasons and that regions with a greater number of station data tended to have a greater ROC score. Météo France System 6 typically had skill later in the fire season, whereas CFSv2 and ECMWF SEAS5 varied in skill depending on the region and subseason.

<< Previous Abstract | Next Abstract >>