Author
|
Submitted to: Society of Range Management
Publication Type: Abstract Only Publication Acceptance Date: 9/19/2007 Publication Date: N/A Citation: N/A Interpretive Summary: Very-large-scale-aerial (VLSA) imagery is useful for measuring rangeland cover attributes, but the location of the image is imprecisely known. Knowledge of the actual location on the ground is important for paired-method validations. A method has been developed at the U. S. Experiment Station for georeferencing VLSA images to within a few meters of the actual location. This is sufficient for locating, on the ground, features that were observed in the imagery. This level of georeferencing will allow the accuracy of imagery based measures of cover attributes to be directly compared on an image by image basis with ground based measurements. Until know these comparisons could not be made. Technical Abstract: VLSA imagery is captured with a digital camera, mounted on a light, piloted aircraft. VLSA images are high quality and have been used to measure cover of plant functional groups and some species, bare ground, litter, and rock, but the actual image location is known imprecisely (± 30 m). This imprecision is of little concern, if the images are used as population samples. However, if the image location must be accurately known, then georeferencing methods are needed. I refer to 3 locations: 1) the planned image center point location, chosen before flight; 2) the aircraft based location, measured from instruments in the aircraft when the camera was triggered; and 3) the actual location, where the center point was on the ground. This demonstration will show how VLSA imagery was georeferenced to within 2 m of the actual location using lower resolution, georeferenced imagery. Two VLSA images were simultaneously collected with approximately 6,000 and 100 m2 footprints. The area covered by the smaller footprint image was nested within the larger footprint image. The larger footprint image showed vegetation, disturbance, and geological patterns that were also perceived in the georeferenced imagery. The pattern in the large footprint VLSA image was matched to the pattern in the georeferenced imagery by translating, rotating, and scaling the VLSA image using GIS software. Similarly, the small footprint VLSA image showed patterns that could be perceived in the corresponding large footprint, and now georeferenced, image. The same method as above was used to georeference the small footprint image to the large footprint image. The small footprint georeferenced location was within 2 m of the actual GPS location measured in the field. This precision is sufficient for finding image features (e.g., a shrub or large rock) on the ground. |
