| Description |
Although wild#12;re strength undergoes signi#12;cant interannual variability, much of the western North America is experiencing an increasing trend in wild#12;re activity. Fires release large amounts of CO2, CO, and other pollutants into the atmosphere, which can impact air quality and climate from regional to global scales. Since #12;res are strong emitters of CO, constraints on CO emissions can potentially lead to more accurate emissions of CO2 and other gases. Thus, in this study, we provided a \top-down evaluation" of CO emissions from two bottom-up #12;re inventories, against observed concentration data. Speci#12;cally, we simulated the wild#12;re-derived enhancements of CO at four observational sites over the western North America during summer 2012, using a receptor-oriented Lagrangian particle model (STILT), with the meteorological #12;elds driven by a mesoscale numerical weather prediction model (WRFv3.4.1). Two #12;re inventories |the newly developed U.S. Forest Service Wildland Fire Emission Inventory (WFEI) and the Global Fire Emissions Database (GFED Version 4.1s) |were evaluated separately. Simulations using both the WFEI and GFED inventories resulted in lower CO concentrations at the Niwot Ridge site in Colorado, compared to NOAAs ask observations, and higher values compared to continuous tower measurements near Fairbanks, Alaska as well as the East Trout Lake site in Saskatchewan, Canada. Regarding discrepancies between simulations and observations, three error sources |potential bias in WRF wind #12;elds, lack of a plume rise model in the STILT model, and uncertainties of CO emissions within #12;re inventories |are evaluated through individual case studies for each region. On average, model-data mismatches from the WFEI inventory are smaller than those from GFED. |
