GSA Annual Meeting in Denver, Colorado, USA - 2016

Paper No. 101-1
Presentation Time: 8:00 AM

AN IMPROVED WATER INDEX FOR MAPPING SUBPIXEL WETLAND INUNDATION AND ITS UTILITY IN UNDERSTANDING THE GEOLOGY OF LAND AREA CHANGE


AMER, Reda1, KOLKER, Alexander2 and MUSCIETTA, Annelise1, (1)Earth and Environmental Sciences, Tulane University, 6823 St. Charles Ave., Blessey Hall, New Orleans, LA 70118, (2)Louisiana Universities Marine Consortium, 8124 Highway 56, Chauvin, LA 70344, ramer1@tulane.edu

Determining the rates and processes that drive land loss and land gain in the coastal zones, is a critical issue in this era of fast relative sea-level rise. This study presented a modified Normalized Difference Water Index (NDWI) for mapping the percentage of wetland loss and gain in the lower Mississippi river within Cubit’s Gap Subdelta. The published NDWI calculated from (green - SWIR)/(green + SWIR) is widely accepted for mapping land/water boundaries because it has the most stable threshold (Ji et al. 2009). Herein, we used satellite imagery and spectral data obtained from Landsat-7 Enhanced Thematic Mapper (ETM+) and Landsat-8 Operational Land Imager (OLI), to derive highly accurate land/water boundary maps at the 30m resolution.. Spectral analysis of ETM+ and OLI images indicated that water has the highest reflectance in blue-wavelength and the lowest reflectance is in the SWIR. We found that the NDWI calculated from (blue - SWIR)/(blue + SWIR), where SWIR is the shorter wavelength region was able to determine the water/land boundaries, and separate mixed water pixels better than the previously published NDWI. Google Earth images were used to derive high accuracy reference map to evaluate the results of NDWI calculated from the ETM+ and OLI images at 30m resolution. The overall accuracies of the ETM+ and OLI classified maps were above 90% confidence level, and Kappa statistics are well above 0.9. The classification accuracy reflects how well the water, non-water and mixed-water classes were identified from the newly developed NDWI. The GIS model show that, in a fifteen year (from 2000 to 2015) there are about 84-km land gain and 38.1-km land loss within in the Cubit’s Gap Subdelta. A related field study indicated that areas of land gain were associated with the import of sediments, either from Mississippi River crevasses or from dredging whereas land loss typically occurred in fringing marshes with greater wave exposure.