Jeana Bissonnette

                                                             Put Your Money Where The Change Is:

Updating The Nation's Wetland Maps Without Exceeding The Budget

Abstract

The Federal Government currently needs information on the country's wetlands, but the process of remapping all National Wetlands Inventory (NWI) Maps is costly and time consuming. This paper presents a nontraditional solution by integrating GIS/remote sensing tools to prioritize NWI map updates. Single date satellite imagery and digital vector data are statistically combined to locate and map areas highly likely to have changed. This approach ensures that mapping resources are used in an optimal way where they are most needed.

Paper

Earth Satellite Corporation (EarthSat) has developed a wetland change analysis procedure called Cross-Correlation Analysis (CCA).  CCA can be used to identify changes and losses in wetlands that were mapped by the U.S. Fish and Wildlife Service�s National Wetlands Inventory (NWI) program.  The procedure uses a recent Landsat Thematic Mapper (TM) imagery (or other digital imagery) and the digital wetland vector data created by NWI to determine wetland changes and losses which have occurred since the wetlands were delineated and mapped from visual interpretation of aerial photography. 

CCA consists of a two-step process.  In the first step, a spectral response is calculated for each of the wetland types that were delineated by NWI for the study area.  In the second step, pixels that have a spectral response significantly different from the expected spectral response are identified.  During the first step, the means and standard deviations for each wetland type delineated on the NWI wetlands maps covering the study area are calculated.  Either all bands of the TM image or selected bands can be processed.  For most study areas, over 200 different wetland types (where each wetland type is a unique combination of system, subsystem, class, subclass, and modifiers of the wetlands and deepwater habitats classification system (Cowardin et al. 1979)) were mapped by NWI.  A table of the means and standard deviations for each Landsat TM band (spectral response) and for every wetland type encountered is produced in the first pass through the Landsat TM image.  In the second pass a �Z-Statistic� is calculated for each Landsat TM pixel that corresponds to a wetland pixel from the NWI digital data set.  The Z-Statistic is computed by calculating the sum of the normalized differences between the means of the Landsat TM bands for the specific wetland type and the actual Landsat TM values for the pixel.  High Z-Statistics indicate dramatic change or losses in the wetlands.  Low Z-Statistics indicate no wetland change.  Z-Statistic thresholds are established to identify four classes of wetlands as shown below:

no significant change

low likelihood of change

moderate likelihood of change

high likelihood of change  

The Z-Statistic thresholds must be established for each study area and vary by study area, quality of TM data, and season during which the TM data were acquired.

The US Fish and Wildlife Service and the Environmental Protection Agency have conducted two pilot studies using CCA.  The first study was for a small area (study area included only five 7.5 minute topographic maps).  In the second study, CCA was calculated for three counties in the Coastal Plain of North Carolina.  In both studies, it was shown that CCA was able to effectively identify wetland changes and losses.  The CCA technique often cannot distinguish between wetland losses (i.e. conversion from wetlands to uplands) and wetland changes (i.e. conversion from forested wetlands to scrub/shrub wetlands that is caused by timber harvesting).  Despite this limitation wetland managers will find this tool extremely effective in appraising the current status of wetlands.

CCA products include maps and digital databases that depict the wetlands as mapped by NWI and provide the current status as the four classes of change as shown above.  Maps can be produced at 1:24000 scale, which will overlay the NWI maps, or at smaller scales that can be used to overlay images of Landsat TM data.  A full Landsat TM scene will provide data for over 200 NWI maps.  Cost of CCA is approximately $50 per map.  For most government organizations, this price includes the cost of the Landsat TM data.  For further information on CCA, please contact the following EarthSat staff by phone, fax or e-mail.  

            EarthSat Staff:

Gregory T. Koeln, Ph.D.         gkoeln@earthsat.com

Roger Mitchell                         rmitchel@earthsat.com

Jeana Bissonnette

Application Scientist

Earth Satellite Corporation

6011 Executive Blvd., Suite 400

Rockville, MD 20852

Phone:  301-231-0660 ext. 236

Fax:  301-231-5020

Email:  jbissonn@earthsat.com