Skip to main content
ARS Home » Research » Publications at this Location » Publication #106521
Title: TESTING LINEAR SEMI-EMPIRICAL KERNEL-DRIVEN BRDF MODELS OVER A SEMI-ARID GRASSLAND REGION WITH VISIBLE AND NEAR-INFRARED ATSR-2 AND AVHRR DATA

Author
item CHOPPING, MARK

Submitted to: International Journal of Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/28/2000
Publication Date: N/A
Citation: N/A

Interpretive Summary: The aim of this study was to test the hypothesis that linear semi-empirical kernel-driven (LiSK) bidirectional reflectance distribution function (BRDF) models are able to predict the reflectance of semi-arid grassland targets at geometries other than those used in model inversion. This ability is important in remote sensing for rangeland mapping and monitoring applications. Data from the NOAA s Advanced Very High Resolution Radiometers (AVHRRs) on the TIROS series satellites and Rutherford Appleton Laboratory s Along-track Scanning Radiometer (ATSR-2) flown on the European Space Agency's ERS-2 satellite are used to test this hypothesis. Results show that the models are able to predict bidirectional reflectance over semi-arid grasslands at the sun-target-sensor geometries of the Along-track Scanning Radiometer (ATSR-2) with reasonable accuracy. Results also show that the Advanced Very High Resolution Radiometers (AVHRRs) on the NOAA TIROS series satellites can provide an adequate sampling of the BRDF for inversion of the models over semi-arid grasslands, provided that both AM and PM sensors are used. Note that this provides only a partial validation; nevertheless, it represents one of the few attempts to validate such models with shortwave visible and near-infrared data acquired by an operational spaceborne sensor. While there are currently few alternatives to the AVHRR and the ATSR-2 for this kind of validation exercise, the near future will see the launch of both the Moderate Resolution Imaging SpectroRadiometer (MODIS) and Multi-Angle Imaging SpectroRadiometer (MISR) sensors on NASA's EOS Terra satellite and the second French POLDER instrument on NASDA's ADEOS-II satellite. These sensors will provide valuable opportunities for more extensive BRDF model validations.

Technical Abstract: This study provides an assessment of the capability of two linear semi-empirical kernel-driven (LiSK) bidirectional reflectance distribution function (BRDF) models to predict bidirectional reflectance at geometries other than those of the observations used in inversion when the models are inverted against the sparse set of angular samples available from the operational Advanced Very High Resolution Radiometers (AVHRRs) on NOAA TIROS series satellites. Shortwave red (0.65 micrometer) and near-infrared (0.87 micrometer) spectral reflectance estimates derived from the Along-Track Scanning Radiometer-2 (ATSR-2) sensor flown on the European Space Agency's ERS-2 satellite are used as reference. The models evaluated are the 3-parameter Roujean and isotropic-LiSparseMODIS-RossThin models (Roujean et al. 1992 and Wanner et al. 1995, respectively). The results show that in spite of the difficulties posed by such a task, LiSK BRDF models can be rapidly inverted against multi-angular AVHRR observations to predict bidirectional reflectance at the acquisition geometries of the ATSR-2 with reasonable accuracy. Correlations between the modeled and observed atmospherically-corrected ATSR-2 surface reflectance estimates were increased over those obtained with uncorrected AVHRR data for the 0.65 micrometer and 0.87 micrometer nadir views and the 0.87 micrometer forward view, with a very small decrease for the 0.65 micrometer forward view. Root-mean-square deviations from the observed values were about 0.04 reflectance for the nadir views and about 0.06 reflectance for the forward views.