Spectral BRF Data Analysis
Contact Persons
Keywords
Bidirectional Effects, Goniometry, Spectral Radiometry, BRDF, Linear Model Inversion, Airborne Campaign
Abstract
In remote sensing, goniometry means the radiometric measurement of angle-dependent reflectance of radiation of natural or artificial surfaces. The quantity governing the reflectance is the bidirectional reflectance distribution function (BRDF) and depends on the wavelength, the two angles (zenith and azimuth) of the incident light and the two angles (zenith and azimuth) of the reflected light. More about BRDF effects can be found here.
The main contributions to bidirectional effects come from shadow casting (wavelength independent) and mutual illumination and scattering (wavelength dependent in vegetation). Therefore the BRDF contains information about the surface structure of an object. In the past thirty years, a large number of physical and empirical BRDF models have been developed. By inversion of a model with a set of measurements, structural parameters can be determined.
Measurements can be made on the ground with the help of a goniometer (e.g., FIGOS), either in the laboratory with artificial pointlike illumination or in the field with natural illumination. A visualization tool - the GONIO package - has been developed for the IDL and ENVI environment and can be downloaded here (2 MB including test data). Bidirectional effects can also be measured by remote sensing image data due to the field of view of the instrument (FOV). In the case of a wide FOV line scanner (e.g., DAIS, HyMAP) the BRDF effect is a brightness gradient across the flight track. This gradient can be modeled with the abovementioned BRDF models.
In a first step, treating BRDF as noise, it is possible to correct the image for those effects, thus enabling a proper intracomparison of the scene. For this task empirical models are sufficient. In a second step, treating BRDF as a signal, structural quantities (e.g., LAI, chlorophyll content) can be determined more accurately. Here, elaborate physical models are needed, which contain those quantities as parameters.
Results
The image shows the bidirectional reflectance of a bare soil target visualized with the GONIO package.
Using a statistical method, the samples to invert the linear semiempirical AMBRALS (Algorithm for MODIS Bidirectional Reflectance Anisotropy of the Land Surface) model can be determined from a single image.
A comparison of goniometric measurements (Gonio, long lines) and modeled bidirectional data from the hyperspectral image (Bar1_15, short lines) for different wavelengths [nm].
The retrieved model parameters can be used to correct the image for BRDF effects. The images show a mosaic of two hyperspectral images acquired from two different flight directions. The checkerboard pattern indicates the reflectance differences due to bidirectional effects.
Left: Mosaic of original images. Right: Mosaic of the same images after BRDF correction with the Ambrals model.
