ABSTRACT

Mapping based on LANDSAT Imagery was undertaken for three test sites along the Texas coast selected for contrasting vegetation, environmental geology, and active geologic processes. Standard LANDSAT products were utilized in a procedure dependent on optical enlargement of part of a band 7 1: 1,000,000 image to fit an existing l: 125.000 map base. Land-water and other distinctive boundaries were extracted from the enlargement to an overlay on which unit boundary data were interpreted from both other bands and the false-color composite of bands 4, 5, and 7.

Each resulting map unit was then classified according to a bi-level classification scheme modified from published systems to meet the priorities of this study. These are, in order of greatest emphasis, the inventory and monitoring of wetlands, other land use, beaches and dunes, and bay systems. Level I classes include urban areas, agricultural land, grassland, woodland, water, wetland, and barren land. Twenty-three Level II categories include five wetland classes: topographically low marshes, topographically high marshes, tidal flat, grass and algal flats, and vegetated spoil. Water was separated into four Level II classes on the basis of turbidity to emphasize flow pattern and suspended sediment distribution in preference to water-body type.

Three test sites have been mapped thus far: the Freeport-San Luis Pass area, the upper San Antonio Bay-Espiritu Santo Bay vicinity, and southern Padre Island and adjacent Laguna Madre. Comparison of maps derived from LANDSAT with color and color-infrared aerial photography, published maps, and field data has shown that the limits of Level II category delineation are primarily dependent on size of the map unit relative to LANDSAT's limit of resolution and on reflectivity contrast with surrounding units. Other factors influencing map accuracy include homogeneity of the map unit, spatial position with respect to surrounding units, season of the year in which the image was taken, and image quality.