Mapping Geographically Isolated Wetlands in the Dougherty Plain, Georgia, USA

Abstract

Increasing interest in isolated wetlands requires an understanding of the location, spatial extent, and configuration of the resource. The National Wetlands Inventory (NWI) is the most commonly used data source for this information; however, the accuracy is limited in some contexts, e.g., agricultural, seasonally-ponded, or forested wetlands. This study sought to improve isolated wetland mapping accuracy by integrating currently available NWI maps with other geospatial data sources (i.e., Soil Survey Geographic Database [SSURGO], Digital Raster Graphics [DRGs], and Digital Elevation Models [DEMs]). Relative to NWI, the optimal model (combining data derived from NWI, SSURGO, and DRGs) identified 1,874 additional isolated wetlands and 19,737-ha of additional isolated wetland area. A remote accuracy assessment indicated a high degree of accuracy for the optimal model (Kappa = 0.88, Sensitivity = 90%, and Specificity = 98%), as compared to NWI maps (Kappa = 0.70, Sensitivity = 68%, and Specificity = 99%). Small wetlands (< 4-ha) dominated the wetland population and reduced nearest neighbor distances between wetlands. Isolated wetlands were clustered and clustering was strongest among larger wetlands. This study identified a framework for improving the accuracy of wetland maps and serves as a foundation for ongoing wetland studies within the Dougherty Plain.