Wolfgang Lucht - New England, USA

BRDF and Remote Sensing
New England, USA



The results given here are from extensive work by d'Entremonet (1997) and d'Entremont et al. (1998). Daily NOAA-14 AVHRR red band data and hourly GOES-8 imager visible band data with spatial resolutions of one kilometer were combined for a period in September, 1995 (AVHRR, September 2-18; GOES, August 25 to September 6), over New England in the United States (40.6 to 44.2 degrees north, 69.7 to 74.6 degrees west). The data were carefully calibrated and mapped to pixels measuring 1.1 by 0.75 km using orbital models, resulting in a 400 by 402 grid (160800 pixels) with mapping residuals of around 0.4 km. The data were meticulously and conservatively cloud-cleared, using 8 spectral tests on all 5 bands of the AVHRR and temporal differencing on the hourly geostationary GOES-8 data. Cloud shadows and pixels immediately adjacent to clouds were also removed. Atmospheric corrections were performed using visibility data from ground stations in the area to characterize aerosols profiles and the MODTRAN and 6S atmospheric radiative transfer codes. The AVHRR data display an almost constant solar angle while the viewing angle varies from one orbit to another with observations located close to the principal plane. The GOES data are characterized by a constant viewing angle and varying solar zenith and relative azimuth angles. The data were used to study RossThick-LiSparse BRDF model inversions of each pixel and the spatial distribution of model parameters, from which 1-km albedo maps were derived.

Relevant papers:

The first image shows the bihemisperical albedo derived from the RossThick-LiSparse BRDF model for each pixel. Note how the the albedo correclty reflects landscape patterns such as the Connecticut river valley, the agricultural Mohawk river valley, the woody regions of the Adirondacks and the Berkshires, and the cities.

The second image shows a normalized difference index of the volume scattering and geometric scattering model coefficients. There are some residual effects of clouds, but nonetheless one can clear see how woody areas produce stronger volume scattering while urban areas and areas with probable gaps in the vegetation cover due to human activity show shadowing.

The third image is given for reference with respect to surface structure and the distribution of forests and gaps in the vegetation. It is a color composite of the AVHRR channel 1 and 2 reflectances obtained in January, 1996, after a snow storm blanketed the region. Forested areas are clearly visible. Compare to the other figures.