Submitted to: Rangeland Ecology and Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/14/2006
Publication Date: 9/1/2006
Citation: Ustin-Stitt, S., Root, R., Brown, K., Hager, S., Mladinich, C., Anderson, G.L., Dudek, K., Ruiz-Bustos, M., Kokaly, R. 2006. Classification of leafy spurge with earth o;bserving-1 advanced land imager. Rangeland Ecology and Management. 59(5): 507-511.
Interpretive Summary: Assessing rangeland weed populations is difficult in many ways. One problem is being able to distinguish the weed from other vegetation and another is collecting data at a scale that can be used by front line managers or regional planners. This study focuses on the Earth Observing –1 (EO-1) Hyperion satellite. EO-1 is unique in that it was the first earth-orbiting hyperspectral imaging spectrometer, which simply means that this remote sensing tool collected reflectance from the earth’s surface at 220 points along the light spectrum. The actual size of each individual area imaged (pixel) was similar to Landsat (30m by 30m), however the added spectral bands 220 verses 7 substantially improved managers ability to map the exotic weed leafy spurge. Overall accuracies ranged from 63% to 78%, but this difference was the result of trying three different analytical techniques. Therefore, the best method of analysis resulted in an accuracy of 78%, which is very good for natural resource mapping. This analysis helps point the way for the development of similar systems, capable of meeting the needs of front-line managers and regional planners.
Technical Abstract: In a series of research studies sponsored by The Ecological Area-Wide Management of Leafy Spurge program, and as part of the 1998 DOI/NASA joint demonstration project, various remote sensing sensors were tested for their ability to map the invasive weed, leafy spurge (Euphorbia esula L). This study focuses on the Earth Observing –1 (EO-1) Hyperion satellite. EO-1 is unique in that it was the first earth-orbiting hyperspectral imaging spectrometer. A NASA-sponsored science validation team evaluated the hypothesis that improved plant mapping could be achieved using the 220 bands of the EO-1 sensor even though the 30 m resolution was considered quite coarse. Three different analytical approaches were applied to the image data collected on 6 July 2001 over the western portion of Theodore Roosevelt National Park, near Medora North Dakota. Overall classification accuracy ranged from 63% to 78%. In addition, leafy spurge stands mixed with as much as 65% of other grass and shrub species were detectable. Hyperion’s coarse spatial resolution (similar to LANDSAT) was offset by the added spectral resolution (198 usable bands) and the tendency for leafy spurge to grow in spatially extensive stands. The preliminary design of this first satellite imaging spectrometer is a big step toward collecting data over large areas; however, future sensor designs with improved signal-to-noise ratios and a wider swath width may eventually enable this type of sensor to address one of remote sensing’s most illusive tasks; quantification of weed/plant infestations at both the landscape and regional scale.