Scale change evaluations for arid land image interpretation: A case study at Camp Williams, Utah

Different scales of aerial and satellite imagery were examined for mapping vegetation species/community and monitoring land degradation at three test sites of Camp Williams, Utah. Color-space conversions and filtering in frequency space were employed as pre-processing techniques. Large-scale Kodak and small-scale IKONOS, IRS and Landsat TM images were classified using an unsupervised classification algorithm. Kodak 4mpp and 5mpp merged classifications were compared with IKONOS and IRS cookie-cut image classifications. Resampling methods used during merging were tested against each other. A method to scale up vegetation cover estimates to larger geographic extents, known as 'Upscaling', was developed involving the transfer of information from high-resolution to low-resolution data. Kodak classifications were degraded and used as training data for the supervised classifications of the small-scale images and tested for accuracy using error matrices. Comparisons of classification accuracies indicate an increase in the overall accuracy and kappa statistic with merging and upscaling of images.