Agreement between measurements of shrub cover using ground-based methods and Very Large Scale Aerial (VLSA)imagery-measured shrub cover.

Authors: Moffet, Corey

Submitted to: Rangeland Ecology and Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/26/2009
Publication Date: 5/1/2009
Citation: Moffet, C.A. 2009. Agreement between measurements of shrub cover using ground-based methods and Very Large Scale Aerial (VLSA)imagery-measured shrub cover. Rangeland Ecology and Management. 62:268-277.

Interpretive Summary: Measurement of rangeland cover using conventional methods is time consuming and appropriate numbers of data are lacking for land management agencies to use for guiding the management. Very large scale aerial (VLSA) imagery can be collected and analyzed quickly, but, due to imprecision in knowing the exact location of the imagery on the ground, has not been validated for shrub cover in the sagebrush steppe. A method of georeferencing VLSA imagery has been developed. Using these georeferenced images, ground cover could be measured in the same area, and estimates of shrub cover from the VLSA imagery has been validated for a mountain big sagebrush community. This method of georeferencing will be useful for other validation studies and will allow users of VLSA imagery to locate precisely the ground where the image was taken for applications when this precision is warranted.

Technical Abstract: New sampling methods are needed for measuring rangeland cover that are more efficient than conventional methods. Very large scale aerial (VLSA) imagery has been suggested as a tool for improving cover sampling efficiency. Because of aircraft pitch and roll, camera misalignment, and errors in the navigation system, the on-ground location where a VLSA image was taken is known with low precision. Error in locations from aircraft data can be as much as 41 m. This has made validation of this type of imagery in heterogeneous landscapes somewhat difficult. By georeferencing the VLSA imagery, location error can be reduced to less than 5.1 m, making it possible to find true on-ground locations and conduct paired measurement comparisons at the plot level. Shrub canopy cover measured from VLSA imagery had ½ limit of agreement widths that were slightly larger (<3.3%) than the repeatability coefficient of the conventional method. Pairing methods at the plot level, rather than the within management unit level, improved validation efficiency. The data from this study were used to characterize location errors, suggest a method for georeferencing VLSA imagery, validate VLSA imagery for estimating shrub cover, and show the benefits of conducting validation experiments with plot level, rather than management unit level, measurements.