Answers are grounded in this paper and cite their sources.
The reflectance of land and vegetation observed in satellite imagery depends on sun and viewing geometry. This bidirectional reflectance requires correction for monitoring changes in vegetation cover and condition. We used a digital camera mounted in a light aircraft, and fitted with a fisheye lens, to measure directional reflectance of a diverse range of landscapes along a long transect in Australia — between Brisbane and the Simpson desert. All, except one, of the measured directional reflectances were able to be characterized accurately (adjusted r2 > 0.95) by the product of two analytical functions. The first, , which represents volume scattering, is a function of illumination and viewing zenith angles, , and has one parameter : The second, , which represents geometric scattering from a surface with sunlit and shaded elements, is a function of phase angle, , and has three parameters, , , and : where is reference reflectance. For ease of parameterizing a bidirectional reflectance distribution function (BRDF) for standardizing satellite imagery, the four-parameter BRDF function may be simplified to two parameters, , and , with only a small reduction in accuracy. The successful characterization of directional reflectance of a diverse range of landscapes, with a product of two simple functions, suggests that multiplicative BRDF functions would be useful for radiometric correction of satellite imagery over large areas. However, overfitting should be avoided to achieve robustness by using few parameters.
No takes yet. Share an insight, caveat, or question.
Dymond et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: