A KNOWLEDGE-BASED EXPERT SYSTEM FOR INFERRING VEGETATION CHARACTERISTICS

The overall goal of the research is to develop a robust extraction technique for inferring physical and biological surface properties of vegetation using nadir and/or directional reflectance data as input. A prototype knowledge-based expert system VEG is described that concentrates on extracting spectral hemispherical reflectance using any combination of nadir and/or directional reflectance data as input. VEG is designed to facilitate expansion to handle other inferences regarding vegetation properties such as total hemispherical reflectance, per cent ground cover, leaf area index, biomass, and photosynthetic capacity. This approach is more accurate and robust than conventional extraction techniques developed by the investigator and others. VEG combines methods from remote sensing and Artificial Intelligence (AI). It integrates input spectral measurements with diverse knowledge bases available from the literature, data sets of directional reflectance measurements, and form experts, into an intelligent and efficient system for making vegetation inferences. VEG accepts spectral data of an unknown target as input, determines the best strategy or strategies for inferring hemispherical reflectance, applies the strategy or strategies to the target data, and provides a rigorous estimate of the accuracy of the inference. VEG is also intended to become a valuable research tool with provisions for testing and developing new extraction techniques on an internal spectral data base, for browsing and analysis (multiple plotting schemes) of the data in the system's spectral data base, and for the development of discrimination learning algorithms that discriminate the spectral and directional reflectance relationships between user-defined vegetation classes.