Detection of Temporal Changes in Vegetative Cover on South Padre Island, Texas Using Image Classifications Derived from Aerial Color-Infrared Photographs

Supervised image classifications developed from 23 x 23 cm aerial color-infrared aerial photographs (1:5,000 scale) were used to evaluate temporal changes in vegetative cover occurring within three 150 x 300-m research sites on South Padre Island, Texas. Use of high-resolution digitized imagery (ground pixel resolution of ca. 0.1 m) and survey-grade GPS for positional measurements of ground control points (20-25 1.0m2 targets within each research site) resulted in consistently high levels of geometric accuracy, with root mean square errors (RMSEs) ranging between 0.397 – 2.867. Similarly, use of relatively simple information categories (dry and wet sand, live and dead vegetative cover, and water) resulted in supervised image classifications with consistently high levels of overall thematic accuracy (90.0 – 98.0%). Temporal comparisons of image classifications using a cross-tabulation procedure indicated that changes in total vegetative cover had been minimal at most locations during the course the two-year study (2003-2005). However, a localized but significant disruption of native vegetation caused by the dumping of sand by highway maintenance crews was detected in one of the study sites (2004), and ground inspection revealed numerous other sand-dumping sites at various locations on the barrier island. In the first situation, comparison of CIR aerial photographs acquired before and after the sand-dumping incident provided a reliable means by which to evaluate the extent of damage caused by the dumped sand, and to monitor its eventual recovery as a result of recolonization by native plant species