The characterization of land surface conditions such as senescent plant materials, soils, green vegetation, and water bodies is essential information for understanding global change. The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) provides more detailed information about materials having specific narrow absorption features and possibilities for detecting very subtle variations than can be obtained from the broad bands of current sensors systems. This research was designed to assess the potential use of hyperspectral remote sensing data for characterizing soil surface conditions, particularly for crop residues and soils. Analytical tools used included the linear mixture model, block kriging, fuzzy c-means clustering, fractal analysis, and the adjustment for atmospheric effects with the ATmosphere REMoval (ATREM) program. To achieve this primary objective, several models and techniques were used for analyzing hyperspectral data to improve the capability to detect and classify crop residues and quantity and quality of variability in soil patterns. Several simulation studies were also implemented to test the procedures used for this study. The results from this study illustrate the usefulness of hyperspectral data for the classification of crop residue using the near infrared bands. Results also show the need for improvement of radiometric resolution of the system. The linear mixture model can be used for reducing the dimensionality and noise effects for the analysis of hyperspectral data. Detailed soil patterns derived from the hyperspectral imagery were well represented using the block kriging interpolation and fuzzy c-means clustering analyses. Fractal analysis indicates that the spatial resolution of AVIRIS is not sufficient for representing the surface roughness of crop residue. Consequently, the future of hyperspectral remote sensing may be very promising for deriving important information about land surface conditions with improved radiometric and spatial resolution.
展开▼
