Author
|
Submitted to: Ecology and Management of Rangelands
Publication Type: Peer Reviewed Journal Publication Acceptance Date: 7/27/2005 Publication Date: 11/1/2005 Citation: Clark, P.E., Hardegree, S.P. 2005. Estimating vegetation change from repeat landscape photography using angular cover. Rangeland Ecology & Management 58:588-597. Interpretive Summary: Repeat photography has been a valuable technique to evaluate trends in vegetation cover over time. Unfortunately, analysis of repeat photography has almost always been qualitative in nature. The objectives of this study were to develop analysis techniques to quantify vegetation cover and changes in cover over time and to identify cover type conversions and assess their frequency based on point sampling of digital images derived from repeat photography. As a test case, digital images were created from 3 black and white, repeat photographs acquired in 1917, 1962, and 2000 of the Whiskey Mountain area in the Owyhee Mountains of southwestern Idaho. Point samples or pixels representing landscape features were classified into cover types by visual interpretation. Classification accuracy for the 2000 image was estimated to be 98.9% based on ground-truth data collected in the field at a random subset of the sample points. No field data were available to ground-truth the 1917 and 1962 images, consequently, classification accuracies of these images were assumed to be similar to that of the 2000 image. Once verified, pixel classification results were then used successfully to quantify vegetation cover and cover changes and to assess frequency of type conversions between image acquisition dates. Although analysis of cover using repeat photography is constrained by the limitations imposed by oblique view angles and variable image quality, it can be useful to quantify vegetation change and stability thus providing a historical context for the current status of our rangeland resources and establishing realistic targets critically needed by natural resource managers and policy makers concerned with future rangeland condition. Technical Abstract: Repeat photography has been a valuable technique to evaluate trends in vegetation cover over time. Unfortunately, analysis of repeat photography has almost always been qualitative in nature. The objectives of this study were to develop analysis techniques to quantify vegetation cover and changes in cover over time and to identify cover type conversions and assess their frequency based on point sampling of digital images derived from repeat photography. As a test case, digital images were created from 3 black and white, repeat photographs acquired in 1917, 1962, and 2000 of the Whiskey Mountain area in the Owyhee Mountains of southwestern Idaho. Point samples or pixels representing landscape features were classified into cover types by visual interpretation. Classification accuracy for the 2000 image was estimated to be 98.9% based on ground-truth data collected in the field at a random subset of the sample points. No field data were available to ground-truth the 1917 and 1962 images. Pixel classification results were then used successfully to quantify vegetation cover and cover changes and to assess frequency of type conversions between image acquisition dates. In this test case, mountain mahogany was the most dynamic vegetation cover type, increasing from 10.1 to 31.2% image cover during the 1917-2000 period. Because landscape photography typically provides oblique angles of view, reported image cover values did not directly reflect actual vegetation cover values. Image cover values were, however, a useful quantitative index of change and stability in vegetation cover. This quantitative information is critically needed to provide a historical context for the current status of our rangeland resources and to establish realistic targets for future rangeland conditions. |
