PARENT SESSION
Contributed Oral Session 64: GIS / Remote Sensing and Landscape Ecology
Tuesday, August 9, 1:30 PM - 5:00 PM, Meeting Room 519 B, Level 5, Palais des congrès de Montréal

Spectral separability of Ecological Systems in the East Gulf Coastal Plain.

Kleiner, Kevin^*,1, MacKenzie, Mark², ¹ Auburn University, Auburn² Auburn University, Auburn

ABSTRACT- The Alabama Gap Analysis Project and Southeast Gap Analysis Project are currently using satellite imagery to map vegetation in the Southeast using NatureServe's Terrestrial Ecological Systems as our target mapping units. The motivation for choosing this classification system is twofold: (1) it shows GAP's recognition of the need to move beyond individual species management and (2) it recognizes the need for a more realistic target mapping system given the difficulty in mapping alliances. This study focused on the East Gulf Coastal Plain (EGCP), and identifying the extent to which these Systems are spectrally separable using Landsat ETM+ imagery. NatureServe currently recognizes 51 Ecological Systems in the EGCP. These Systems, however, exist at a range of spatial scales across the landscape. Of these 51 Systems, 36 were determined to exist in sufficiently large patches to be recognized in the imagery. Point data were compiled from multiple sources, including state heritage records, aerial photography, and field work. Three seasonal mosaics (spring, fall, winter) were normalized and DN values were extracted from the mosaics for each image band at each point (21 bands per point). Hyperelipsoids (21 dimensions) were constructed for each System to assess class separability. Results indicate that some Systems are spectrally separable but many Systems overlap and are spectrally confused with one another. Spectral confusion is particularly high for Systems with a high degree of heterogeneity, due either to multiple life forms, continuous transition within the class (succession), or mixed pixel response (scale). Future work will focus on incorporating additional data such as land form and soils to further enhance class separability.

Key words: remote sensing, land cover classification, gap, terrestrial ecological systems