Use of satellite remote sensing data for modeling carbon emissions from fires: A perspective in North America

Abstract

Biomass burning emits huge amount of gases and particles in various forms of carbon compounds and thus play a key role in global carbon cycling. To reach a closure in carbon balance, we need a full and accurate accounting of carbon emissions due to fire activities. Remote sensing is the only feasible means of monitoring fires around the globe. Maximum usage of satellite data is thus highly desired to achieve this goal. However, none of the emission related fire attributes are measured directly by any satellite sensors. Inversion algorithm and modeling are required to estimate fire emissions using satellite data. This chapter provides an extensive review of remote sensing data and methods that could be brought to bear on fire emission estimation in terms of their information content, extraction method, strengths and limitations. These parameters include burned area, burning fragmentation and spreading, fuel loading, fraction of fuel consumed by fire, and emission factors for different gases. In general, satellites can provide good information on burned area by combined use of hot spot data together with changes in vegetation indices. Fire fragmentation and/or severity depend critically on satellite sensor resolution. For moderately coarse resolution data like MODIS and AVHRR, unburned fire islands inside the fire polygons provided by forest agencies may be singled out, but little can be gained concerning inhomogeneous degree of burning. Limited information may be extracted on fuel loading in terms of forest regrowth age, fraction of tree coverage by using a combination of measurements from several passive channels, especially NIR and SWIR data. Vegetation height detected by space-borne lidar may be linked to biomass content. Satellite-based land cover classification on continental scale may help refine fuel type classification. Determination of the fraction of fuel consumption (crown and surface) usually requires modeling, except for a short-cut approach that links radiation emission with fuel consumption. © 2006 Tsinghua University Press.