Comparison of Land Cover Characterization Using EOS MISR and MODIS Data and a Decision Tree Classifier

Land cover characterization at a regional scale using spaceborne multi-angle remote sensing data is in its infancy. A data mining technique was employed to evaluate the degree to which the accuracy of land cover classification can be increased using multi-spectral, multi-temporal and multi-angle remote sensing data. The study area is around the Jornada Rangeland in New Mexico, USA with shrubland, woodland, grassland, desert barren land and irrigated cropland. Data used for this study included EOS MISR surface BRF and MODIS 16-day NDVI composite acquired from 2002-2003. Eight land cover types were classified using a decision tree algorithm with multiple classifications obtained to evaluate classification accuracy using different input data (MODIS data only, MISR data only, and MODIS plus MISR data). The overall classification accuracy from a five-fold cross-validation using MODIS data alone was 46% (standard error of 2.6%) as compared with 51% (standard error of 2.3%) using MISR data. The class-specific accuracy obtained from using MISR data was equal to or higher than those using MODIS data (0-28%) for all land cover types. The largest increase in accuracy was observed for open oak woodland, coniferous woodland, and woody wetland using MISR and MODIS data, and for irrigated cropland and barren land using MISR data only. The increased classification accuracy using MISR data was attributed to difference in directional spectral signature and its temporal behavior among open woodland, woody wetland, irrigated cropland, and barren land.