REMOTE-SENSING OF FOREST BIOPHYSICAL STRUCTURE USING MIXTURE DECOMPOSITION AND GEOMETRIC REFLECTANCE MODELS

Using geometric shadow and linear mixture models we develop and evalua te an algorithm to infer several important structural parameters of st ands of black spruce (Picea mariana), the most common boreal forest do minant. We show, first, that stand reflectances for this species can b e represented as linear combinations of the reflectances of more eleme ntal radiometric components: sunlit crowns, sunlit background, and sha dow. Secondly, using a geometric model, we calculate how the fractions of these radiometric elements covary with each other. Then, using han d-held measurements of the reflectances of the sunlit background, spha gnum moss (Sphagnum spp.), and assuming shadow reflectance to be that of deep, clear lakes, we infer the reflectance of sunlit crowns from t he geometric shadow model and low-altitude reflectance measurements ac quired by a helicopter-mounted radiometer. Next, we assume that the re flectance for all black spruce stands is simply a linear combination o f shadow, sunlit crown, and sunlit background reflectance, weighted in proportion to the relative areal fractions of these pixel elements. W e then solve a set of Linear equations for the areal fractions of thes e elements using as input helicopter observations of total stand refle ctance. Using this algorithm, we infer the values for the areal propor tions of sunlit canopy, sunlit background, and shadow for 31 black spr uce stands of varying biomass density, net primary productivity, etc. We show empirically and theoretically that the areal proportions of th ese radiometric elements are related to a number of stand biophysical characteristics. Specifically, the shadow fraction is increasing with increasing biomass density, average diameter at breast height, leaf ar ea index (LAI), and aboveground net primary productivity (NPP), while sunlit background fraction is decreasing. We show that the end member fractions can be used to estimate biomass with a standard error of app roximate to 2 kg/m(2), LAT with a standard error of 0.58, dbh with a s tandard error of approximate to 2 cm, and aboveground NPP with a stand ard error of 0.07 kg . m(-2). yr(-1). We. also show that the fraction of sunlit canopy is only weakly correlated with the biophysical variab les and are thus able to show why a popular vegetation index, Normaliz ed Difference Vegetation Index (NDVI), does not provide a useful measu re of these biophysical characteristics. We do show, however, that NDV I should be related to the fraction of photosynthetically active radia tion incident upon and absorbed by the canopy. This work has convinced us that a paradigm shift in the remote sensing of biophysical charact eristics is in order-a shift away from direct inference of biophysical characteristics from vegetation indices and toward a two-step process , in which (1) stand-level reflectance is approximated in terms of lin ear combinations of reflectance-invariant, spectrally distinct compone nts (spectral end members) and mixture decomposition used to infer the areal fractions of these components, e.g., shadow, sunlit crown, and sunlit background, followed by (2) the use of radiative transfer model s to compute biophysical characteristic values as a function of the en d member fractions.