100607 Using Land-Use Change, Soil Characteristics, and a Semi-Automated On-Line GIS Database to Inventory Carolina Bays.

Poster Number 474-134

See more from this Division: SSSA Division: Wetland Soils
See more from this Session: General Wetland Soils Poster I

Wednesday, November 9, 2016
Phoenix Convention Center North, Exhibit Hall CDE

Jeffrey G. White, Box 7620, North Carolina State University, Raleigh, NC, Dana Sullivan, TurfScout, LLC., Greensboro, NC and Michael J. Vepraskas, Department of Crop and Soil Sciences, North Carolina State University, Raleigh, NC
Poster Presentation
  • WhiteJG NCSU Carolina Bays 2016.pdf (2.8 MB)
  • Abstract:
    Carolina Bay wetlands are common in the southeastern US Coastal Plain and important to water quality, carbon sequestration, and habitat. Only South Carolina and Georgia have statewide inventories. We developed and evaluated a novel way to identify and delineate Bays using Bay-dense Bladen County, NC as a testbed. We posited that Bay land use had changed in the past 40 years. We classed decadal Landsat images as forest, agriculture, urban, and water. We used 812 previously delineated Bays to identify common Bay soils. From areas with both common Bay soils and land-use change, we delineated 548 new Bays using a semi-automated on-line digitization tool. We saved new Bays to a Google Fusion Table for download and integration within a geographic information system. To gauge accuracy, Bays were scored on soils, land-use change, wetland delineation, and landscape position. We assessed errors of omission and commission and estimated an overall accuracy of 74 to 86%.  Errors included omission of some small Bays (3.3-14.6% of total Bay area digitized) and commission of misclassified Bays (11% of 1,360 Bays).  With 1,360 delineated Bladen-County Bays, we estimated that as many as 478 have not yet been identified. Unclassified Bays are most likely small, <5.76 ha.

    See more from this Division: SSSA Division: Wetland Soils
    See more from this Session: General Wetland Soils Poster I