Assessing vegetation types of Big Bend National Park, Texas for image-based mapping

Plumb, Gregory

Aerial photo interpretation and satellite image processing are techniques pervasive in contemporary vegetation mapping. Usually the vegetation categories selected for mapping are either previously defined from an existing classification or are formulated in a descriptive, or qualitative, manner. Quantitative analysis tends to be concentrated upon identifying the spectral characteristics of the categories from the remotely sensed data. In this work, a quantitative comparison is made among twenty prospective categories for mapping vegetation cover within Big Bend National Park, Texas. Vegetation types were subjectively identified at 131 sites within the Park. Each site was also examined numerically by measuring the extent of the floristic and lifeform constituents sampled within the site. A cluster analysis was performed using measures of floristic similarity and lifeform similarity among the sites. A dendrogram was constructed as part of the cluster analyses. Most of the category names corresponded to groupings interpreted on the dendrogram. Sites within these groupings typify the character of the vegetation cover, so their locations can also be used to derive spectral statistics for the purpose of mapping. While correspondences also suggest that vegetation cover may be reliably identified qualitatively in the field, several sites were either outliers of categories or were anomalous to the classification. Some vegetation types were inseparable on the dendrogram or were closely linked. Such categories may be difficult to spectrally distinguish on remotely sensed scenes.

Assessing vegetation types of Big Bend National Park, Texas for image-based mapping

Series