A logit model for predicting wetland location using ASTER and GIS

  • Authors:

  • E. Pantaleoni;R. H. Wynne;J. M. Galbraith;J. B. Campbell

  • Affiliations:

  • Institute of Agricultural and Environmental Research, 3500 John A. Merritt Blvd. Tennessee State University, Nashville, TN 37209, USA;Department of Forestry, 319 Cheatham Hall (0324), Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA;Crop and Soil Environmental Sciences Department, 239 Smyth Hall, Polytechnic Institute and State University, Blacksburg, VA 24061, USA;Department of Geography, 109 Major Williams Hall (0115), Polytechnic Institute and State University, Blacksburg, VA 24061, USA

  • Venue:
  • International Journal of Remote Sensing
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) were used to develop a logistic regression model to predict the location of wetlands in the Coastal Plain of Virginia. We used the first five bands from two ASTER scenes (spanning 0.52-2.18 µm) covering the same area, acquired 6 March 2005 and 16 October 2005. March Band 3 contributed the most in discriminating wetlands over the other ASTER bands (marginal effect = 7.277), and it predicted the location of 60% of the total wetlands. We used a canonical discriminant analysis to test the significance of GIS variables in separating wetlands from uplands. Soil Survey Geographic Database soil data had the highest correlation with the first canonical component (0.876), followed by March Band 3 (-0.803), and the National Hydrography Dataset water (0.725). We included GIS data layers into the logit model. The resulting model predicted the location of over 78% of total wetlands, highlighting the potential of models incorporating ASTER data for speeding the wetland mapping process, lowering costs of map production, and improving wetland mapping accuracy.