Development of Techniques for Quantification of Cattail
Canopy Coverage of Wetland Pools at Cheyenne Bottoms Using Kite Aerial
Managers of the Cheyenne Bottoms wetlands are concerned that the narrow-leaved cattail has been proliferating at an unhealthy pace. Excessive cattail growth decreases botanical diversity and exclude access to mud flats where shorebirds feed.
These wetlands serve as one of the most important stopping places for migratory birds in the Western Hemisphere. The Bottoms, as they are known by locals, have been recognized as a Ramsar wetland of international importance.
Two administrative entities are responsible for the management of the 16,600 hectares that comprise the Cheyenne Bottoms wildlife refuge. The Kansas Department of Wildlife and Parks manages the largest pool to the east and the Nature Conservancy manages 2,915 hectares in the western portion of the refuge. This work is being conducted as part of Emporia State University's NASA EPSCoR project in cooperation with the Nature Conservancy. It is our goal to develop remote sensing techniques that will aid the management of these wetlands.
The available Landsat imagery may not provide the ground resolution (30 m) needed to elucidate the patterns of open water and cattail canopy. I have reviewed digital orthophoto quarter-quadrangles (DOQQ) with 1 meter ground resolution and digital color photographs from kite aerial photography (KAP) with sub-meter resolution. The objective of this work was to develop techniques to convert panchromatic or color photographs into a format that could be analyzed with the Idrisi32 program.
I wanted to quantitatively determine the ratio of cattail canopy to open water. The Idrisi program provides many powerful techniques and algorithms for image processing and GIS analysis, but it seems they are primarily oriented toward the interpretation of satellite imagery. I tried to use pattern recognition techniques associated with Idrisi32 but found that they did not separate the various flora and open water pools as I had hoped.
The first KAP photos that I reviewed were taken in mid-March 2002 while all vegetation was yet dormant. There was a visible difference in color between the cattails and the grasslands but the primary visual difference was the textural pattern. These patterns are created when the cattails lodge or fall down, presumably due to the effect of a stem boring insect.
The difference between the upland grasslands, cattail and open pools are apparent to the human eye which frequently integrates lines, textures and colors to identify an image. These differences are not so easily distinguished with the available computer software. John Jensen quotes Hardin and Thompson, 1992, in his Introductory Digital Image Processing text in saying that "most of the computer-assisted image processing to date has involved the use of only a few of the basic elements of image interpretation. In fact, the overwhelming majority of all digital image analysis appears to be dependent primarily upon just the tone and color of individual pixels in the scene using fundamental statistical pattern recognition techniques."
After trying several edge and pattern recognition techniques I decided upon a hybrid approach in which I used a Paint Shop Pro (PSP) to draw in the boundaries that my eye perceived. Then I experimented with methods to convert three-tiered RGB values into single values that would differentiate the cattails from open water. The area outside the pools or grasslands were excluded in a similar manner and given a different set of values through a masking technique. Then the images were color separated with PSP and the red layer was imported to Idrisi32 for GIS analysis.
The imported images were reclassified and given a new color code using the symbol workshop. Ground resolution in meters per pixel was estimated and added to the metadata. Then I performed a database query for AREA to estimate the number of hectares covered by cattails, open
water and the uplands outside the wetland pool.