Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: Analysis of Hyperspectral Imagery for Assessment of Wyoming Rangelands

Author
item Hunt, Earle

Submitted to: Intnl Conference On Geospatial Information In Agriculture And Forestry
Publication Type: Proceedings
Publication Acceptance Date: January 10, 2000
Publication Date: January 13, 2000

Interpretive Summary: Hyper-spectral sensors have a large number of bands, spaced close together in the visible (400 to 700 nm wavelength), near infrared (about 700 to 1300 nm wavelength), and the shortwave infrared (about 1300 to 2500 nm wavelength). Models of atmospheric properties were used to calculate reflectance spectra from the hyper-spectral data. One hyper-spectral sensor, the Airborne Visible Infrared Imaging Spectrometer (AVIRIS) from NASA's Jet Propulsion Laboratory, was flown over two sites during the summer of 1995 (a year with an abnormally high amount of precipitation) and the summer of 1998 (a year with a normal amount of precipitation). The first site was the Nature Conservancy's Red Canyon Ranch (RCR) near Lander, Wyoming, and the second was the ARS Central Plains Experimental Range (CPER) near Nunn, Colorado. At CPER, there was a long-term experiment manipulating grazing intensity: not grazed, lightly grazed, medium grazed, and heavily grazed. There were no significant differences in calculated reflectance among the grazing treatments in 1995; however, in 1998, overall reflectance was higher with increased grazing.

Technical Abstract: Airborne Visible Infrared Imaging Spectrometer (AVIRIS) data from NASA's Jet Propulsion Laboratory were acquired on July 25, 1995 and August 6, 1998, over two sites. The first site was the Nature Conservancy's Red Canyon Ranch (RCR) near Lander, Wyoming, and the second was the ARS Central Plains Experimental Range (CPER) near Nunn, Colorado. Normalized Difference Vegetation Indices (NDVI) was highly correlated to percent living (green) vegetation cover for both sites and both years. At CPER, there was a long-term experiment manipulating grazing intensity: not grazed, lightly grazed, medium grazed, and heavily grazed. There were no significant differences in ATREM-calculated reflectance or NDVI among the grazing treatments in 1995, which was a very wet year. However, in 1998, a year with a normal amount of precipitation, overall reflectance was higher with increased grazing, but NDVI's were equal among the treatments. Mapping vegetation communities with unsupervised and supervised techniques were difficult because of the effects of soil mineral spectral absorption features. Furthermore, mapping communities using the 1995 data were more difficult due to the strong influence of sun-surface-sensor geometry for different parts of the AVIRIS scenes (caused by the non-Lambertian bidirectional reflectance distribution function, BRDF, of the surface). Whereas the full potential of hyper-spectral imagery was not utilized, the results suggest that one-time-only hyper-spectral images would not help in assessing the status of rangelands resulting from grazing.

Last Modified: 10/1/2014
Footer Content Back to Top of Page