Spatio-spectral heterogeneity analysis using EO-1 Hyperion imagery

How does observed spatial structure change as a function of spectral wavelength and resolution? The fine spectral resolution data of a Hyperion image acquired over agricultural research fields in eastern Nebraska, USA, were analyzed using variography to characterize spatial patterns within each wavelength-specific image. The geospatial analysis was repeated simulating Landsat TM/ETM+ bands from the Hyperion data. Significant changes in spatio-spectral heterogeneity were observed across the red edge. The total amount of spatial variability is closely correlated to the contrast in brightness of objects within the scene. The spatial dependence in an image is also influenced by differential brightness, but to a lesser extent. Increasing bandwidth may blur spatial structure that occurs at finer spectral resolution, resulting in the loss of spatial structure that may relate to significant features within the scene.