If you have any problems related to the accessibility of any content (or if you want to request that a specific publication be accessible), please contact us at email@example.com.
Evaluating Fire Emissions Inventories to Model Smoke Exposure in the Western United States
AdvisorHolmes, Heather A
AltmetricsView Usage Statistics
Understanding the impacts of wildland fire smoke on nearby communities requires knowing the amount and composition of the smoke emitted into the atmosphere. This may seem easily measurable using the existing network of ground-based air quality monitoring sensors, but these instruments capture pollution from all sources, making it difficult to isolate pollution exclusively from smoke plumes. Fire emissions invento- ries can be used to estimate emissions from an individual wildfire by using empirical data to model fire behavior and emissions. Outputs from the fire emissions inven- tories can also be used in chemical transport models and dispersion models. The emissions from a wildland fire are impacted by many variables, including fuel char- acteristics (i.e., amount, type, moisture content), and combustion type (i.e., flaming vs. smoldering). Variability in any of these factors creates differences in the chemical speciation and amount of emissions released into the atmosphere by the fire. Each fire emissions inventory models these variables as inputs to quantify emissions, but each inventory adopts a different method or dataset to determine input variables, leading to different emissions estimates. Evaluating the differences between inventory methods are crucial to understanding the effect on modeling wildfire emissions, and how these differences are propagated through smoke plume dispersion modeling used to determine downwind smoke plume concentrations.As there is no definitive dataset available for comparison, determining the differ- ences between fire emissions inventories requires a direct comparison between each inventory and additional data sources to determine the differences in the reported results of each inventory. Four fire emissions inventories are compared and discussed: the Fire INventory from NCAR (FINN) version 1.5, Global Fire Emissions Database (GFED) version 4s, the Missoula Fire Lab Emission Inventory (MFLEI) (both the 250m product and the spatially aggregated 10km product), and the Wildland Fire Emissions Information System (WFEIS) (two versions using different burned area products). Emissions results from these fire emissions inventories are compared for 2013 and a single, multi-day fire that impacted the Reno, Nevada area (Yosemite Rim Fire). Factors such as burned area, fire location, and the primary emissions of PM2.5 and other constituents were compared to highlight differences amongst the fire emissions inventory results. To aid in the comparison of these emissions inventories, a Bayesian single level model was developed for each fire emissions inventory using data from the fire emissions inventories. These models were compared to understand what each fire emissions captures as a distribution curve. Results show that each fire emissions inventory had a similar amount of burned area for both 2013 and the Yosemite Rim Fire, but primary emissions and the average daily emissions rate differed greatly between inventories. The Bayesian single-level models created for each emissions inventory were used to determine which points in the inventory have the most influence on the Bayesian model. These influential points are investigated using additional data sources to quantitatively assess how well each fire emissions inventory is representing the fire emissions for these influential points. These comparisons are used to inform the selection of the fire emissions inventory used for a broader project related to estimating the human health impacts of wildfire smoke exposure in the Reno area.