PHYSICALLY-BASED CLASSIFICATION AND SATELLITE MAPPING OF BIOPHYSICAL CHARACTERISTICS IN THE SOUTHERN BOREAL FOREST

Fundamental problems inherent to the existing land cover and biophysic al characteristic algorithms are fourfold: (1) their failure to deal p hysically with global and interannual variations in surface reflectanc e arising from Sun and view angle variations, (2) decoupling of the la nd cover classification algorithm from the biophysical characteristic inference algorithm with no ability to control biophysical parameter e stimation error arising from misclassification, (3) invalid statistica l assumptions within classification algorithms used to model reflectan ce distribution functions, and (4) sole reliance on vegetation indices that can limit performance for several major land cover classes. To a ddress these problems, we develop an integrated, physically based clas sification and biophysical characteristics estimation algorithm that u tilizes canopy reflectance models to account directly for signature va riations from Sun angle, topographic, and other variations. Our approa ch fuses into a single algorithm both land cover classification and bi ophysical characteristics estimation, permitting one set of physically based canopy reflectance models to be used for both. The use of canop y reflectance models eliminates the need for unrealistic assumptions, such as multivariate-normal distributions, underlying many classificat ion algorithms. Using the algorithm, we have classified a 10,000 km(2) area of the BOREAS southern study area. Our classification shows that low-productivity wetland conifer is the dominant land cover and that nearly 7% of the area is occupied by boreal fens, a major source of me thane. In addition, nearly 23% of the area has been disturbed by eithe r fire or logging in the last 20 years, suggesting an important role f or disturbance to the regional carbon budget. A thorough evaluation of the physically based classifier within the southern study area shows accuracies superior to those obtained with conventional statistically based algorithms, implying even better performance when extended over multiple Landsat frames since the physically based approach can accoun t directly for regional variations in reflectance resulting from varyi ng illumination and viewing conditions (topography, Sun angle). The co nifer biomass density estimation algorithm is based on our discovery o f a convenient natural relationship between crown height and volumetri c density that renders the biomass density for black spruce stands ind ependent of tree height, and a function only of sunlit canopy fraction . This permits us to calculate directly the relationship between refle ctance and biomass density. An evaluation of the algorithm using groun d sites shows our algorithm can estimate black spruce biomass density with a root-mean-square error of 2.73 kg/m(2) for correctly classified sites. Our evaluation also demonstrates the importance of correct cla ssification. Root-mean-square errors for misclassified sites were 3.96 kg/m(2). Using this approach we have estimated the biomass density in the BOREAS southern study area for the dominant land. cover type in t he circumpolar boreal ecosystem, wetland black spruce. These results s how a bimodality to the biomass density regional distribution, control led perhaps by underlying topographic and edaphic factors.