An expert system decision tree classification model for the location of blue oak in the interior Coast Ranges of San Benito County, California

California blue oak ecosystems are at risk. As one of the state's landmark endemic trees it appears common and abundant. Human encroachment, low regeneration rates, and strictly voluntary policies for protection, however, may be setting the stage for the loss of the lovely stands of blue oak woodlands that ring the Central Valley and grace the Coast Ranges. Recent work by the University of Arkansas Tree-Ring Laboratory has brought to light that (1) blue oak is an excellent precipitation proxy for paleoclimatological studies, (2) the existing population is of surprisingly old age, and (3) the scale of existing mapping is inadequate to define the size and distribution of remnant stands for scientific analysis and conservation management.; This project developed an expert system decision tree classification model on Landsat ETM+ imagery for blue oak ecosystems in San Benito County. San Benito County is one of the prime regions where large scale blue oak mapping is needed because of its extensive cover by blue oak habitats. The county's remoteness and large areas of undeveloped land also make it an ideal region in which to find relict stands of old growth blue oak.; The model mapped more than 66,000 hectares of blue oak woodland types with accuracies of greater than 85% on two independent assessments. There was reasonable agreement between the model and a Pinnacles National Monument GIS vegetation model even though the two had significant differences in classification organizations and vintages of source data and were different data types. When correlated with soils data the model found many more soil types that support blue oak ecosystems than the literature suggested.; The results indicate that an expert system developed from field and laboratory experience can be used to inform a very accurate decision tree classification on medium resolution remote sensing imagery to produce regional scale high resolution mapping of California blue oak. The results could be improved with additional field sampling and verification. A probabilistic approach would be less labor-intensive than the expert system employed. This mapping should prove useful for land management and for future vegetation and paleoclimatological studies.