A methodology to validate coarse resolution biophysical products from multi-resolution MERIS data
F. Javier García-Haro(1), Aleixandre Verger(1) and Beatriz Martínez(1)
(1) University of Valencia, Dr. Moliner 50, 46100 Burjassot, Spain
The validation of biophysical satellite-derived vegetation products is a challenging task particularly regarding the extent of the products, the spatial resolution, as well as the dynamics of the vegetation. In this work a novel methodology, which relies on a multi-step “bottom-up” scaling approach from local field-level measurement to global comparison with satellite products is proposed. It exploits the unprecedented potential offered by the MERIS/ENVISAT sensor, which provides fine resolution optical data at two different spatial resolutions -300m and 1200m- with a revisit time of approximately three days. One feature that makes ENVISAT a unique system is the possibility of combining MERIS full resolution (FR), which is detailed enough to link field-measured biophysical variables obtained over a relatively small (3-10 km) validation sites with remote sensing reflectance, and MERIS reduced resolution (RR), which is meaningful to extend the obtained functional relationship over large areas.
The ground measurements were collected in the framework of different international validation field campaigns, and included land cover, leaf area index (LAI) and fractional vegetation cover (FVC). Following the methodology of the VALERI project, a transfer function is established to relate the ground measurements of the biophysical variables to the high spatial resolution (TM, SPOT) satellite imagery. Then FVC and LAI fields are aggregated to fit the spatial characteristics of concurrent MERIS acquired at full resolution. The SMAC algorithm provided by the BEAM software –which makes use of several MERIS scene atmosphere parameters- has been applied to calculate the surface top-of-canopy MERIS reflectance in FR mode (300 m) and RR mode (1200m), in 15 spectral bands, of which 13 bands are used for land surface characterization (excluding bands 11 and 15). Different statistical models for relating field-measured biophysical variables to MERIS FR reflectance have been considered. They include multiple robust regression, canonical correlation analysis and non-linear models. One advantage of these methods is that they exploit the full spectral dimensionality of MERIS system. Further, land cover information has been used to examine different stratified models and accounting for upscaling effects. The cross-validation method has been applied to evaluate the performance of the different models. Results reveal that unbiased continuous estimates of biophysical variables such as FVC and LAI can be obtained. Finally, the predictive model has been readily extended to a temporally composited mosaic of MERIS RR covering a large area in Europe -by combining all MERIS RR paths whose orbits fall within this area. This may be justified by the observed radiometric consistency of MERIS in the RR and FR modes. All MERIS data were obtained from an ESA cat-1 project. The upscaled ground-based estimates of biophysical parameters can be a valuable reference for validating satellite-derived vegetation products. Retrieved fields have shown an overall good agreement with regard to existing coarse resolution satellite products. Results suggest that the proposed approach is robust and performs well over a range of vegetation communities.
Keywords: ESA European
Space Agency - Agence spatiale europeenne,
observation de la terre, earth observation,
satellite remote sensing,
teledetection, geophysique, altimetrie, radar,
chimique atmospherique, geophysics, altimetry, radar,