Hirotoshi Yoshioka

Optical characterization of the Brazilian Savanna physiognomies for improved land cover monitoring of the cerrado biome: preliminary assessments from an airborne campaign over an LBA core site

Abstract It is estimated that approximately 40% of the Cerrado, the second largest biome in South America, have been already converted. In this study, situated within the scope of the Large Scale Biosphere-Atmosphere Experiment in Amazonia project (LBA), we conducted a wet season ground and airborne campaign over the Brasilia National Park (BNP), the largest LBA core site in the Cerrado biome, to measure the optical and biophysical properties of the major Cerrado land cover types. We investigated land cover discrimination through the analyses of fine resolution spectra, convolved spectra (MODIS bandpasses), and vegetation indices—the normalized difference vegetation index (NDVI) and the enhanced vegetation index (EVI). At these three data levels, three major physiognomic domains (herbaceous, woody, and forested) could be readily identified, and the amount of data correctly classified into the five major land cover types found at BNP were 91% (full spectra), 78% (red and NIR), 75% (NDVI), and 71% (EVI). A synergism between the NDVI and EVI was also evident, and together, these two indices were capable of correctly classifying 82% of the total data set. Our results indicate the possibility of utilizing the MODIS NDVI and EVI images for operational land cover assessments in the Cerrado region.

An assessment of the MODIS vegetation index compositing algorithm using quality assurance flags and Sun/view angles

The normalized difference vegetation index (NDVI) products, as derived from the NOAA AVHRR sensor series, have been shown useful for the studies of the land biosphere characteristics and dynamics at regional to global scales. Standard pre-processing in generating these AVHRR NDVI products include the compositing process, in which the highest NDVI value from a series of multitemporal georeferenced images is retained for each pixel location in order to minimize cloud and atmosphere contamination. While this maximum value composite (MVC) procedure has been shown to produce NDVI images with a relatively high degree of radiometric consistency, adjacent composite pixels may have been acquired at widely varying viewing geometries due to the bidirectional reflectance distribution effects of the surface targets, of which magnitude varies with land cover types and atmospheric corrections, resulting in large inconsistencies. In response, the compositing algorithm of the Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation index (VI) products emphasizes a global, operational view angle standardization. It utilizes a bidirectional reflectance distribution function (BRDF) model to produce nadir looking equivalent reflectance values if enough cloud free observations are available during a 16-day compositing period. Otherwise, a backup, MVC criterion that includes a view zenith angle constraint (namely, the constraint view angle MVC criterion, CVMVC) is utilized to composite. In this study, the authors assess and characterize this new compositing algorithm using early MODIS data and compared it with the conventional MVC algorithm.

Preliminary analysis of MODIS vegetation indices over the LBA sites in the Cerrado region, Brazil

The Brazilian Cerrado, the second largest biome in South America, is marked by a distinct seasonal contrast and comprises an intricate mosaic of land cover types, vertically structured as grassland, shrubland, and woodland. Due to intensive agricultural and grazing occupation, the Cerrado is today the most severely threatened biome in Brazil. Therefore, the discrimination of the major Cerrado types, the monitoring of the vegetation conditions, as well as the identification of the current distribution of land-cover types and occurrences of land use intensification, play a major role in the Global Land-Cover and Land-Use Change Strategy within the scope of the NASA Earth Observing System (EOS) program. In fact, different studies have already demonstrated the potential of the optical vegetation indices to monitor the Cerrado vegetation and condition. These investigations, mainly based on temporal AVHRR data sets converted to the normalized difference vegetation index (NDVI), indicate that the dominant land cover types may be grouped into three spatial domains, encompassing the savanna formation and pasture sites, the forested areas, and the agricultural crops. In relation to the recently launched MODIS, the key instrument onboard the Terra platform, the expectation is that performing the NDVI, a continuity index, as well as the EVI (Enhanced Vegetation Index), will result in more precise and accurate measures of the vegetative cover. The extent, to which such sensor and VI concept improvements will directly impact biophysical parameter estimations and vegetation mapping in the Cerrado biome, is the main focus of this investigation.