HR: 1340h
AN: B43A-0134 [Abstracts]
TI: Uncertainties in Satellite Based Fire Emission Inventories in the Amazon
AU: * Nandi, S
EM: sreela@ucar.edu
AF: National Center for Atmospheric Research, 1850 Table Mesa, Boulder, CO 80305
United States
AU: Vanchindorj, U
EM: vuliisa@ucar.edu
AF: National Center for Atmospheric Research, 1850 Table Mesa, Boulder, CO 80305
United States
AU: Wiedinmyer, C
EM: christin@ucar.edu
AF: National Center for Atmospheric Research, 1850 Table Mesa, Boulder, CO 80305
United States
AU: Guenther, A
AF: National Center for Atmospheric Research, 1850 Table Mesa, Boulder, CO 80305
United States
AU: Prins, E
EM: elaine.prins@ssec.wisc.edu
AF: NOAA/NESDIS/ORA/ASPT, 17207 Alexandra Way
, Grass Valley, CA 95949
United States
AU: Setzer, A
AF: INPE, Av. dos Astronautas,1.758 - Jd, Sao Jos‚ dos Campos, SP 12227-010
Brazil
AU: Artaxo, P
AF: Universidade de Sao Paulo, Rua do Matao, Travessa R, 187
, Sao Paulo, SP 05508-900
Brazil
AU: Elvidge, C
EM: Chris.Elvidge@noaa.gov
AF: National Oceanic and Atmospheric Administration, 325 Broadway, Boulder, CO 80305
United States
AB:
The uncertainties of developing satellite geolocation based fire emissions inventories for air quality models are discussed
in this work. Various satellite hot spot detection and burn scar area products are routinely combined with emission factors
to develop monthly and daily gridded fire emission inventories for both air quality modeling applications and global models
Here, we compare the spatial autocorrelations between fire hot spots detected in the infrared by the Geostationary
Operational Environmental Satellites (GOES) Wildfire Automated Biomass Burning Algorithm (WF ABBA), the Moderate Resolution
Imaging Spectroradiometer (MODIS) 5 minute L2 thermal anomaly, and the NOAA-14 Advanced Very High Resolution Radiometer
(AVHRR), and the Defense Meteorological Satellite Program (DMSP) visible channel for one month from 20 September 2002 to 20
October 2002 for an approximately 1000 km x 1000 km domain in Amazonia. Because of the differing overpass times of the polar
orbiting satellites and the differing temporal and spatial resolutions of the sun-synchronous satellites and geosynchronous
satellites, there is no discernable spatial autocorrelation between the detected hot spots on a 1 to 2.5 kilometer scale.
Once these hot spots are counted and allocated to either 10 km2 or 20 km2 grid cells typically used for regional air quality
modeling applications, spatial autocorrelation increases from 0.55 to 0.69, indicating that all the satellites examined here
detect fires in the same general geographic locations. Further inventories of hot spots detected as a function of ecosystem
type (GLCC version 2.0) in the GOES WF ABBA data are consistent with recent fire spots as a function of ecosystem type in the
Global Wildland Fire Emission Model as reported by Hoelzemann et al in 2004.
Comparison of the number of hotspots in South America month period, respectively 227,159 for GOES WF ABBA, 28,359 for MODIS
L2 and 13,334 for AVHRR indicate that although these satellites observe similar spatial patterns, the number of hot spot
detections observed by the different satellites differs substantially and therefore emissions modelers must take this into
consideration. Examination of the fire area, maximum fire duration, and the diurnal pattern in the GOES WF ABBA dataset
further indicates that no one satellite product, is appropriate for detecting small short duration fires. The uncertainties
in emissions inventories can be reduced by using a
combination of satellite products.
DE: 0394 Instruments and techniques
DE: 0399 General or miscellaneous
SC: Biogeosciences [B]
MN: 2004 AGU Fall Meeting