On the use of fire radiative power, area, and temperature estimates to characterize biomass burning via moderate to coarse spatial resolution remote sensing data in the Brazilian Amazon

Spaceborne instruments provide a unique view of global vegetation fire activity many times a day. In this study, we assessed the fire characterization information provided by two major products: the Terra and Aqua MODIS Thermal Anomalies product (MOD14 and MYD14, respectively) and the Wildfire Automated Biomass Burning Algorithm (WF_ABBA) product derived from GOES East Imager. Using higher spatial resolution imagery data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and Landsat Enhanced Thematic Mapper Plus (ETM+) instruments, we analyzed the characterization of subpixel fires detected by MOD14, MYD14, and WF_ABBA over parts of Brazilian Amazonia. Our results suggest that MODIS and GOES fire radiative power (FRP) estimates derived for individual fire-pixel clusters are subject to errors due to the effects of the point spread function of those instruments (underestimation of up to 75%), improper fire background characterization (overestimation of up to 80% assuming a 10 K cold bias in background temperature), and omission of small fire lines. Detection limits were approximately 11 and 9 MW for MOD14 and MYD14, respectively, and were equivalent to 27 and 19 MW for WF_ABBA data acquired coincidently with MOD14 and MYD14, respectively. We found a positive correlation between FRP and percentage tree cover indicating that FRP is sensitive to biomass density. Fire area and temperature estimates derived from the application of Dozier's (1981) approach to GOES data did not agree with our reference data (i.e., ASTER and ETM+ active fire masks and in situ fire temperature data), suggesting that large and variable errors could affect the retrieval of those parameters.