2.1.2 CG lightning and fire weather simulation with WRF
CG lightning strikes and fire weather conditions are important factors
affecting fire behaviors and are highly dynamic across space and time.
Due to the lack of weather stations and very coarse resolution of
climatology data in the remote tundra region, we adopted the Weather
Research and Forecast (WRF) model as a downscaling tool to simulate CG
lightning probability and near-surface weather conditions at 5km
resolution. We used the National Centers for Environmental Prediction
Final Operational Global Analysis data (NCEP FNL; National Centers for
Environmental Prediction/National Weather Service/NOAA/U.S. Department
of Commerce, 2000) at 1-degree resolution and 6-hour interval for model
initialization. We ran two-way nested simulation for Alaska following
the parameterization settings from He and Loboda (2020).
Considering the computing complexity of WRF, we sampled years with
different fire season severities between 2001 and 2019 and ran WRF
simulations for all the detected fire events from these years for
further modeling efforts. We adopted the empirical-dynamical modeling
framework developed by He and Loboda (2020) to model the probability of
CG lightning strikes using WRF simulated variables and random forest
(RF) algorithm. CG lightning probability was then used as input data for
representing ignition sources of wildfires. To describe fire weather
conditions that affect burnings in the tundra, we extracted near-surface
weather conditions, including air temperature, relative humidity (RH),
wind speed, and 24-hr precipitation. We then calculated the Canadian
Forest Fire Weather Index System (CFFWIS; Van Wagner, 1987) using
WRF-simulated variables. The CFFWIS tracks the moisture content of
distinct fuel layers with three fuel moisture codes – Fine Fuel
Moisture Code (FFMC), Drought Moisture Code (DMC), and Drought Code
(DC). The three fire behavior indices – Initial Spread Index (ISI),
Buildup Index (BUI), and Fire Weather Index (FWI) – provide numeric
ratings of the fire spread process. Though not explicitly designed for
the tundra, this system is suitable for describing fire weather
conditions and quantifying fire danger in the ecosystems of the HNL
(French et al., 2015; Mölders, 2010).