Improved estimates of biomass burning emissions in the southeast United States
Abstract
Biomass burning is a major source of gas and particle emissions that affects air quality, human health, and climate. Prescribed burns in the southeastern United States consume more biomass and cover a larger area than fires in the rest of the United States combined. Although fires can be detected remotely from thermal infrared emission and changes to surface reflectance, there are multiple issues that make satellite detections difficult in the eastern United States. These include small fire sizes, short duration, low intensity, canopy coverage, and rapid vegetation regrowth. Some attempts have been made to compensate for this bias, for example the small fire product in the Global Fire Emission Database (GFED4.1s) product. The accuracy of GFED and other remotely sensed global fire emission inventories are largely unknown, outside of a few field studies, mainly because there are few independent datasets of fire extent. The Florida Forest Service (FFS) has extensive records on fire type, size, location, and time for both prescribed and wild fires, which have not previously been used to evaluate fire area and emissions. For our study period of 2004 to 2016, we compared FFS burn authorization data against GFED4.1s burned area. When averaged across the state of Florida, there is 4 times more land burned than detected from satellite sensors. When comparing FFS data against high quality records from Apalachicola National Forest, Avon Park Air Force Range, Eglin Air Force Base, Tall Timbers Research Station, and Tyndall Air Force base, the areal discrepancy between these records and FFS reports are +/- 15%, well below the 4 times detection discrepancy between satellites and FFS reports. We have developed a method to statistically correct this satellite bias in fire detections. Treating the FFS burn authorizations as accurate, we have found this bias ratio can be predicted from fire size, land cover type, leaf area, and month. The regression model incorporating these factors can predict greater than 80% of variance in bias ratio across Florida during the summer months with correlations around 0.6 on average. This improved estimate of burned area in Florida will be used in global circulation models to determine the true contribution of prescribed wild fires in the southeast United States to gas and particle emissions.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2017
- Bibcode:
- 2017AGUFM.A31C2194N
- Keywords:
-
- 0305 Aerosols and particles;
- ATMOSPHERIC COMPOSITION AND STRUCTURE;
- 0360 Radiation: transmission and scattering;
- ATMOSPHERIC COMPOSITION AND STRUCTURE;
- 0365 Troposphere: composition and chemistry;
- ATMOSPHERIC COMPOSITION AND STRUCTURE;
- 0368 Troposphere: constituent transport and chemistry;
- ATMOSPHERIC COMPOSITION AND STRUCTURE