Incorporating the use of MODIS data into INPE deforestation mapping
Yosio
Shimabukuro, INPE, yosio@ltid.inpe.br
(Presenting)
Ruth
DeFries, University of Maryland, rdefries@geog.umd.edu
Douglas
Morton, University of Maryland, morton@geog.umd.edu
Liana
Anderson, INPE, liana@ltid.inpe.br
Marcelo
Lopes
Latorre, INPE, latav@ltid.inpe.br
Matthew
Hansen, University of Maryland, mhansen@glue.umd.edu
Ellen
Jasinski, University of Maryland, ejasinsk@pop900.gsfc.nasa.gov
Moderate resolution satellite data, such as those from the EO-1 Terra and Aqua MODIS sensors, permit more rapid analysis of deforestation in the Brazilian Amazon than high resolution Landsat data due to their regional coverage with small data volumes. We evaluated the use of MODIS data to prioritize high resolution analysis in the INPE PRODES program by comparing MODIS deforestation maps generated by four different algorithms to 2002 INPE PRODES deforestation maps from Landsat ETM data. For 6 Landsat scene test areas, NDVI differencing and linear mixture model techniques were performed on both daily surface reflectance and 16-day composite surface reflectance MODIS products at 250 m resolution. Vegetation Cover Change (VCC) and Vegetation Continuous Fields (VCF) algorithms were also run on 250 m resolution MODIS data. The accuracy of each MODIS deforestation map with respect to the 2002 PRODES analysis was calculated as the percent of true positives, percent of PRODES deforestation area, and percent of PRODES deforestation polygons detected by polygon size. Results from all four algorithms show a high degree of potential to reliably identify new deforestation. The relative importance of minimizing false positives, maximizing deforestation area identified, or maximizing the percent of small (0.25-3 ha) deforestation areas detected will determine the most appropriate algorithm and data type. The accuracy and speed of MODIS data analysis techniques show a high degree of potential to assist priority setting for high-resolution analysis as part of operational deforestation monitoring at INPE.
Submetido por Douglas Morton em 17-MAR-2004
Tema Científico do LBA: LC (Mudanças dos Usos da Terra e da Vegetação)