Mapping Heterogeneous Fuel Characteristics and Fuel Consumption Using AVIRIS, LiDAR, and Field Data for Fire Emissions Modeling

Abstract:

As wildland fires become more frequent, fire emissions will have an increasing impact on climate and air quality at local to global scales. Fuels impact fire behavior and combustion, which affect the composition of source fire emissions. Yet, common approaches to mapping fuels using categorizations of “fuel models” do not accurately capture the heterogeneity of fuel characteristics that contributes to uncertainty in source emissions. In this study, we evaluate the relationship between Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and LiDAR data to field-measured fuel metrics to model and map heterogeneous fuel parameters for three fires: the Williams Flats Fire in Washington state and two prescribed burns in Fishlake National Forest. AVIRIS data were acquired from the Jet Propulsion Laboratory and LiDAR and field-measures were acquired from the US Forest Service. Topographic data were acquired from the Shuttle Radar Topography Mission. We derive maps of plant functional traits, fractional cover of green vegetation, non-photosynthetic vegetation, substrate, and charcoal as well as fuel moisture from AVIRIS, and use metrics derived from LiDAR such as canopy bulk density, percent understory and canopy cover, and 95^th percentile height. To fill gaps in pre-fire coverage, we developed models for AVIRIS fuel characteristics aligned with fuel model parameters (response) using random forests classification with Sentinel-2 and LiDAR (predictor). Once we had consistent maps of pre-fire AVIRIS data layers, we used partial least squares regression to derive fuel characteristics measured in the field. The models were then applied to predict and map fuel characteristics across the burned landscape. We present results of fuel characteristic heterogeneity by fuel model classification and burn severity to demonstrate the need to directly map fuel characteristics when estimating fuel consumption and subsequent biomass emissions.