Evaluation of Near Real-Time IMERG Precipitation Estimates for Fire Weather Applications in Alaska

In Alaska, wildfire outbreaks are common during the summer months when the days are long and convection is frequent. Daily fire danger metrics in Alaska are calculated using the Canadian Forest Fire Danger Rating System, which primarily uses meteorological data (temperature, wind speed, rainfall, relative humidity) as inputs. Currently, the precipitation estimate input fields are produced by the Alaska Pacific River Forecast Center (APRFC), which depend heavily on in-situ observations and radar. Given the sparse nature of in-situ observations and lack of radar coverage, spatially heterogeneous variables like rainfall are potentially misrepresented. Inaccuracies in the daily precipitation estimates can create problems in forecasting of fire potential, resource allocation, and safety of residents and first-responders. This presentation will evaluate a newly available remotely-sensed dataset as a potential solution Alaska’s data scarcity problem.

In 2014, the Global Precipitation Measurement (GPM) mission began with the launch of the GPM Core Observatory. Using retrievals from passive microwave and infrared instruments onboard satellites in the GPM constellation, quasi-global estimates of precipitation are produced by the Integrated Multi-satellitE Retrievals for GPM (IMERG) algorithm. Algorithm upgrades contained in the new Version 06B now enable computing these data for Alaska during the fire season (1 June – August 31). For each timestep, the IMERG algorithm is executed three times, producing three independent datasets to be used for both time-sensitive and research applications. For fire weather applications, precipitation estimates from the near real-time IMERG-Early (IMERG-E; 4-hour latency) run are the most pertinent.

As an initial analysis of this dataset, 24-hour accumulations of IMERG-E precipitation estimates are evaluated across Alaska during five fire seasons (2014-18). These data are evaluated using both in-situ observations and gridded APRFC quantitative precipitation estimates. Using a regional quantile mapping approach, the IMERG-E precipitation estimates from the 2019 fire season are then bias-corrected and compared to the respective APRFC estimates. The verification of this dataset will determine algorithm strengths and limitations for use in fire weather applications, and will serve as a baseline for its performance in other high-latitude regions.