Verbyla, D. L. and C. Richardson. 1996.
Remote sensing clearcut areas within a forested watershed:
Comparing SPOT HRV Pan, SPOT HRV Multispectral, and
Landsat TM data.
Journal of Soil and Water Conservation. 51:423-427.
Abstract. We compared the accuracy of classifying
exposed soil/vegetation areas in 12 forested test areas using rectified
SPOT HRV panchromatic (PAN),SPOT HRV multispectral (XS), and Landsat Thematic
Mapper (TM) data. There was close agreement between the classified
satellite images and a reference grid of exposed soil/vegetation classes
that were derived from 1:15840 color aerial photographs. The mean
overall classification accuracy exceeded 85percent. Omission errors
of misclassifying reference exposed soil areas (derived from aerial photography)
were consistently less than 5 percent. Commission errors resulted
mainly from: 1)difference between the minimum mapping unit area used
in photo interpretation (0.1ha) and satellite image cell sizes (0.01, 0.04,
and 0.09 ha), 2) positional errors between the classified images
and the reference grid, and 3) some incorrect classification of spectrally
bright areas. Classifications with SPOT HRV PAN,SPOT HRV XS, or Landsat
Thematic Mapper can be useful for monitoring potential sediment sources
within large watersheds, especially if the exposed soils are large clearcut
areas and there is a high contrast between bright, exposed soils and surrounding
landcover types.
Email: D.Verbyla@uaf.edu
Last updated: August 2001