Fire Weather in the Northeastern United States

Although easily overlooked, the Northeastern United States has a fairly large number of wildfires as well as a distinct fire season. The US Forest Service fire incident database (Short 2022) contains almost 67,000 records for the period of 1992-2020 for upstate New York State (NYS) alone. About half of these fires occurred in the spring season, between April 1 and May 30, with about half of those incidents coinciding with forested areas. While fires tend to be small (only 3% exceeded 5 acres in reported size), interannual variability is large, with 2005 and 2008 being recent very active spring seasons (about 3000 incidents each) and 2011 having been relatively quiet (about 500 records).

When fires do occur, they generally appear across NYS, suggesting they are responding to mesoscale to synoptic scale weather conditions. However, these fires do not appear to be aggressively warned. Between 2007-2023, the National Weather Service in Albany (ALY) issued only 23 separate Red Flag Warnings or Fire Weather Watches. 61% of those were in the month of April, and only one was not during the spring season.

To enhance situational awareness of wildfire conditions for public utilities in the Northeast, we are evaluating existing fire weather indices -- including the Canadian Fire Weather Index, the Santa Ana Wildfire Threat index (SAWTi), the Hot-Dry-Windy metric, etc. -- and formulating a new measure tailored to conditions in the New York and New England areas. Fuels components are being obtained from several sources, including NCAR's fuel moisture analyses. Weather components of these indices can be computed using operational forecasts on the short and medium range and also efficient artificial intelligence weather prediction models such as GraphCast. As described in Brewer, Fovell, and Capps (2024), we have developed a methodology for using the WRF model to downscale GraphCast forecasts, which can further enhance the value of these skillful but relatively coarse resolution (0.25 degree latitude-longitude) predictions.