LBA area, Brazil

BRDF and Remote Sensing
The LBA Area, Brazil



In this study, NOAA-9 AVHRR GAC data with a spatial resolution of 4 kilometer was inverted for large parts of central South America (Hu, dissertation, 1997; Hu et al., RSE, 1998) on a per-pixel basis after bulk atmospheric correction using the RossThick-LiSparse BRDF model. A small subset of these data fall onto the region of the Large-Scale Biosphere-Atmosphere Experiment (LBA) in Brazil.

The following first image shows BRDF-corrected NDVI and the traditional maximum-value NDVI for a 30-day period in July, 1987. For comparison, a land cover classification of the region is also shown, as well as a map of vegetation heights (these latter two images were taken from Calvet et al., Bulletin AMS 76, 1997, copyright American Meteorological Society). For non-scientists: the NDVI measures relative brightness differences between the red and the near-infrared bands. Since vegetation is bright in the near-infrared and dark in the red, large values of the NDVI indicate large abundance of vegetation, low values lack of vegetation. The gaps in the image are locations where cloudiness made the inversions unreliable. This image demonstrates that the BRDF-corrected NDVI picks up the difference between the clearcut agricultural areas with low vegetation and the surrounding high rainforest.

The second image shows bihemispherical land surface albedo in the visible band as derived from the BRDF model, and as derived from a Lambertian assumption using the same reflectances as in maximum-value NDVI compositing. Panel (c) shows the BRDF model volume scattering coefficient, indicating where the BRDF is predicted to have the strongest bowl-shape (least shadowing). Again the land cover classification (Calvet et al., Bulletin AMS 76, 1997) is given as a baseline.

Relevant papers: