Towards retrieval of forest cover density over snow from the Multi‐angle Imaging SpectroRadiometer (MISR)

Abstract

AbstractVegetation structure and density affect the dynamics of snow accumulation and ablation. Vegetation also affects our ability to estimate snow‐covered area accurately from satellite‐based sensors. The objective of this case study is to demonstrate how the angular pattern of reflectance from vegetation over snow can provide information on forest cover density. Imagery from the Multi‐angle Imaging SpectroRadiometer was acquired over north‐central Colorado on 15 February 2002. Angular reflectance data were extracted from a variety of image locations and were analysed in conjunction with a digital elevation model and maps of forest cover density and forest cover type. The Rahman–Pinty–Verstraete semi‐empirical parametric model was successfully used to simulate the angular patterns of reflectance. The model's k parameter, a measure of reflectance anisotropy, was used to characterize the angular signatures of selected pixels. Results show distinct patterns of anisotropic reflectance that depend on density and cover type. Non‐forested areas exhibit a bowl‐shaped pattern (k < 1·0) of reflectance versus viewing angle. Low‐density deciduous forests also have this bowl‐shaped reflectance pattern, but this changes as the density increases. Other forest cover types show transitional patterns between bowl and bell shapes and distinct bell‐shaped patterns (k > 1·0) for higher densities. However, the relationship between k and density does not hold for forest cover densities that approach 100%. For a density of 99%, the fir–spruce forest cover type has a distinct bowl shape and a k value of only 0·69. This is in agreement with previous work indicating that sub‐pixel homogeneity (whether because of sparse vegetation cover or extremely dense vegetation cover) will result in k < 1·0. This preliminary study indicates from a qualitative standpoint that multi‐angle reflectance data captures sub‐pixel‐scale information on forest cover density. Copyright © 2004 John Wiley & Sons, Ltd.