Laserfiche WebLink
<br />0)2393 <br /> <br />photography was acquired in July 1976, and provided a more up-to-date refer- <br />ence for evaluation of classification results than the 1973 black-and-white <br />photographs. <br /> <br />Visual comparison of the Landsat classification with the color-infrared <br />aerial photography indicated the following results. Level I land-use deline- <br />ation generally appeared to be accompl ished with very little misclassifica- <br />tion, with the following exceptions. One involved urban areas (p. 22). Oth- <br />er possible areas of misclassification occur at boundaries between two or <br />more classes; also, in certain areas in the basin, there appeared to be lim- <br />ited misclassification between agriculture and rangeland classes. This most <br />likely includes either fallow or stubble fields and adjacent areas of sparse- <br />ly vegetated rangeland. It also was observed that recently reclaimed sur- <br />face-mine areas (surface-mine areas were not selected as a class because o{ <br />their small size) were classified as dryland agriculture. These areas have <br />been smoothed over and seeded in grasses, thus appearing similar to agricul- <br />tural fields on the images. High-wall areas of active surface-mine opera- <br />tions were classified as cloud shadows (black) because of the shadowed areas <br />present. Older unreclaimed mine-spoil areas, were classified as rangeland or <br />sparsely vegetated rangeland. <br /> <br />Results also indicate that in many areas a reasonable level II <br />classification can be obtained for agricultural-land, rangeland, forest-land, <br />water, and wetland classes (table 2). Exceptions, in addition to those <br />discussed for level I classification, include some mixed categories and Some <br />classes .lith small areal extent. Because of lack of detailed ground <br />information, it was not possible to identify specific crop types for the <br />three dryland-agriculture classes (table 4). Areas classified as dryland <br />agriculture and confirmed using the July 1976 aerial photography also <br />indicated a considerable increase in areal extent of dryland-agriculture <br />activity along the Yampa River valley during the last several years, <br />especially between Craig and Milner, Colo. (fig. 1). This was determined by <br />comparison of dryland-agriculture fields appearing on the 1973 black-and- <br />white aerial photographs with those appearing on the 1976 color-infrared <br />images. The greater part of the increased area for dryland agriculture <br />resulted from conversion of rangeland to cultivated fields. Results in <br />level II classification of forest areas appeared to be quite accurate with <br />the exception of zones where oak brush and aspen trees tend to intermix, <br />causing an overlap of rangeland and forest-land classes. Also, irrigated hay <br />meadows, which would technically be included in an agricultural class, <br />virtually were spectrally the same as natural-grass areas along stream <br />drainages, which are in the class of wetlands. Water bodies greater than 5 <br />acreS (2 ha) in size were classified accurately in almost all areas, with the <br />exception of some lakes in steep-sided mountain valleys which were obscured <br />by shadows. <br /> <br />In summary, it appears that the Landsat system is a potentially valuable <br />tool for regional land-use classification and update in the Yampa River <br />basin, subject to the exceptions that have been noted previously. Land-use <br />maps for much of the Yampa River basin are already in existence, but these <br />maps already are outdated because of the rapid changes in land use that are <br /> <br />27 <br />