Final Activity Report Summary - TREBEOD (Techniques for retrieval of biophysical parameters from EO Data)
The main objective of the research was to participate and gain advanced understanding of some of the issues in the NASA's proposed Vegetation Canopy (VCL) LIDAR satellite sensor; specifically the research being carried out at the Geography Department of the University of Maryland at College Park with regard to testing and validation of the technical aspects of the VCL mission and retrieval of vegetation canopy parameters using data from its airborne simulator 'Laser vegetation imaging sensor (LVIS)'. And because the VCL system was going to have limited acquisition on a global scale, it was thought that fussing its data with data from other sources, e.g. Bi-directional reflectance distribution function (BRDF) could help in the interpolation of the vegetation canopy parameters for the areas where the VCL have gaps. But due to the cancellation of the VCL mission, the research was shifted to using existing airborne VCL simulator (LVIS) data in combination with the elevation data from the Shuttle radar topographic mission (SRTM).
During the outgoing phase at the University of Maryland at College Park, I was able to gain insider information concerning algorithms used for the processing and extraction of vegetation canopy parameters from LVIS data. It was possible to code and modify some of the IDL software tools used within the group, e.g. the estimation of noise level for the derivation of canopy cover. We had to optimise the IDL code because the noise component is very critical for the identification of the bottom part of the vegetation canopy to avoid contribution of signal due to the under-story. Also, through our research into the characterisation of SRTM derived vegetation canopy height in relation to LIDAR derived metrics of vegetation canopy vertical structures, we were able to comprehensively analysis the SRTM derived canopy height using statistical measures, showed graphically the relationship between SRTM derived canopy height and LIDAR vegetation metrics, and explained the deviations in the SRTM derived parameter in relation to the C-band signal backscattering process with the vegetation canopy. Using vegetation canopy height derived from LVIS data, vegetation canopy mean heights generated from the SRTM elevation data for three US test sites, namely Sierra Nevada forest in California, Coweeta forest in North Carolina and Hubbard Brooks experimental forest in New Hampshire, were shown to be correctable using mathematical linear modelling.
Generally, data concurrency problem was encountered during the project. For example, although an LVIS data acquisition campaign was carried out in the Sierra Nevada forest site in the first week of July 2006, temporal and geometric decorrelation in the Polarimetric interferometric synthetic aperture radar data acquired over the same test site and during the same period of time by the Japanese ALOS PALSAR sensor hindered successful processing and consequently the validation analysis.
During the outgoing phase at the University of Maryland at College Park, I was able to gain insider information concerning algorithms used for the processing and extraction of vegetation canopy parameters from LVIS data. It was possible to code and modify some of the IDL software tools used within the group, e.g. the estimation of noise level for the derivation of canopy cover. We had to optimise the IDL code because the noise component is very critical for the identification of the bottom part of the vegetation canopy to avoid contribution of signal due to the under-story. Also, through our research into the characterisation of SRTM derived vegetation canopy height in relation to LIDAR derived metrics of vegetation canopy vertical structures, we were able to comprehensively analysis the SRTM derived canopy height using statistical measures, showed graphically the relationship between SRTM derived canopy height and LIDAR vegetation metrics, and explained the deviations in the SRTM derived parameter in relation to the C-band signal backscattering process with the vegetation canopy. Using vegetation canopy height derived from LVIS data, vegetation canopy mean heights generated from the SRTM elevation data for three US test sites, namely Sierra Nevada forest in California, Coweeta forest in North Carolina and Hubbard Brooks experimental forest in New Hampshire, were shown to be correctable using mathematical linear modelling.
Generally, data concurrency problem was encountered during the project. For example, although an LVIS data acquisition campaign was carried out in the Sierra Nevada forest site in the first week of July 2006, temporal and geometric decorrelation in the Polarimetric interferometric synthetic aperture radar data acquired over the same test site and during the same period of time by the Japanese ALOS PALSAR sensor hindered successful processing and consequently the validation analysis.