Assessment of the real-time fire emissions (GFASv0) by MACC

TitleAssessment of the real-time fire emissions (GFASv0) by MACC
Publication TypeMiscellaneous
Year of Publication2010
AuthorsHeil, A, Kaiser, J, van der Werf, GR, Wooster, MJ, Schultz, MG, van der Gon, HD
Secondary TitleTechnical Memorandum
Date PublishedJuly
Type of WorkTechnical Memorandum

The Global Fire Assimilation System (GFAS) used in MACC provides real-time fire emissions for atmospheric forecasts using satellite-observed fire radiative power (FRP) information from MODIS and SEVERI. We evaluate the quality of FRP-based estimates of combustion rates by comparing them with corresponding data from the latest version of the GFED inventory (GFED3). FRP-based estimates and GFED3 agree well with each other in terms of the temporal variability and the regions of major burning. On a global scale around 80% of the monthly variation in the amount of dry matter burned (GFED3 inventory) can be explained by monthly variations in MODIS FRP. However, GFED3 burned dry matter is on average 2.2 times higher. The difference is most pronounced in areas with potential soil organic matter burning, pointing to limitations in the detection capability of FRP-based approaches in these biomes. We find good correlations of the FRP-based and GFED combustions rates within each of the eight investigated land cover types, though. Therefore, using land cover-specific conversion factors will generally allow reproducing the GFED combustions rates with FRP observations within GFED's estimated accuracy. The remaining major difference is that the MODIS FRP observations appear to detect more fires than are included in GFED. We also compare global total monthly CO emissions that have been produced by the MACC real time service (GFASv0) in 2009 to the GFED inventory. Both agree relatively well in terms of total amounts, but there are differences in the seasonal cycles in various regions. Several recommendations for the future development of the GFAS MACC system are derived from the comparisons. We conclude that their implementation will finally result in a fire estimation system that combines the advantages of the well established GFED inventory with those of the new FRP-based approach.