Spectrodirectional ground-based remote sensing using dual-view goniometry: field BRF retrieval and assessment of the diffuse irradiance distribution in spectrodirectional field measurements

Abstract

Earth Observation (EO) data provide information of the surface characteristics with regard to the spatial, spectral, temporal and directional dimensions. In spectrodirectional Remote Sensing the Earth's surface reflectance characteristic is studied by means of its directional (angular) dimension. Almost all natural surfaces exhibit an individual anisotropic reflectance behaviour, which is described by the bidirectional reflectance distribution function (BRDF). BRDF effects in remotely sensed data occur in dependence on the specific observation-illumination geometry present for each pixel of the image during data acquisition. Various applications, such as BRDF correction of remote sensing data and
quantitative retrieval of vegetation or soil parameters therefore require accurate knowledge of spectrodirectional surface reflectance properties. However, the target specific bidirectional reflectance factors (BRF) cannot directly be measured but need to be retrieved from spectrodirectional measurements usually performed with goniometer systems either in the field or in a laboratory
environment. Field goniometry has the advantage that the target is left in its natural environment, including the natural illumination by the sun. The major disadvantage, however, is that atmospheric effects and undesired time variations of the illumination have to be taken into account. Laboratory goniometry on the other hand allows for a better control of environmental conditions but measurement
results are subject to conical illumination geometry and the inhomogeneity of the illuminated area. Therefore, the directly measured quantities for field and laboratory goniometry are only approximations to the real target BRF. The most exact BRF retrieval (correction of the diffuse
illumination due to the atmosphere) from field goniometer measurements can be achieved by following the procedure proposed by Martonchik et al. (1994) provided that the incoming diffuse radiation is observed as a function of the observation angles. However, most goniometer measurement
setups do not account for this. Others do so, but only over a limited spectral extent (multispectral) or are not yet operational. Consequently, prior to the study at hand, no operational hyperspectral goniometer system, which is able to characterize the angular distribution of the reflected and incoming radiation field, existed.

The presented dissertation primarily focuses on field goniometry and the assessment of the diffuse influence in spectrodirectional field measurements. The main research questions are 1) how can the required input quantities for the BRF retrieval be accurately measured, 2) to what extent are traditional field reflectance measurements influenced by the diffuse irradiance and 3) to what extent do directional effects influence vegetation parameters derived from ground-based and spaceborne multiangular measurements.

Results consist of the development and characterization of the first hyperspectral dual-view field goniometer system (dual-view FIGOS), which is able to simultaneously obtain the reflected and the incoming diffuse radiation at high angular resolution. A reliable characterization of the angular distribution of the incoming diffuse illumination is presented for several atmospheric conditions and
the performed field BRF retrieval for an artificial, as well as a natural target led to a reasonable assessment of the diffuse influence. Furthermore, this study clearly demonstrated the necessity of accounting for atmospheric changes during the measurement period. The direct comparison of multiangular spaceborne and ground-based measurements revealed that forward scattering reflectance is especially sensitive to target heterogeneity and associated canopy element scales, which both strongly affect the distribution of illuminated and shadowed surface areas.

The dual-view FIGOS showed a stable and reliable performance during several extensive measurement campaigns and strongly supports future surface BRF generation being used for e.g. model validation and inversion purposes as well as for albedo calculations. Additionally, its combined
use with multiangular spaceborne or airborne data acquisition provides the possibility of improved directional calibration instead of using nadir-view ground-truth measurements.