A second source of error is the error associated with the GIS layers used in making the delineations. The main layers that were used and an estimate of the horizontal accuracy for each layer is listed.

Counties (+/- 50 meters); see note 1
DEM (30 meter resolution, +/- 15); see note 2
DEM (90 meter resolution, +/-100); see note 2
Geology (+/- 700); see note 3
Hydrography (+/- 50); see note 2
Hydrologic units (+/- 700); see note 4
Land Cover (+/- 250); derived from Thematic mapper dataset; see note 3
Thematic Mapper (+/- 200); original dataset was not terrain corrected
and was resampled up to 100m resoultion; see note 3

Note 2. Accuracy estimate from the "Geo-Positioning Selection Guide" developed by the U.S. Department of Interior, Bureau of Land Management (U.S. Government Printing Office: 1992-676-687).

Note 3. Accuracy estimate based on a comparison of the positions of large lakes and reservoirs and major rivers in the GIS layer with the positions of the corresponding features in the USGS 1:100,000 hydrography DLG coverage.

Note 4. Accuracy estimate based on a comparison of the line along the continental divide in the Wind River Range with the corresponding lines in the counties coverage, which forms the boundary between Sublette and Fremont counties.

Some of the arcs in the Landtype Association coverage were taken directly from the source layers, so the errors associated with these arcs carried over to the Landtype Association coverage. Other arcs were digitized on-screen with a variety of background displays. In this case, there was not only the error associated with the GIS layer used for the background display, but also a digitizing error. The most common background display was shaded relief derived from 30 m DEM data, with the display zoomed to a scale of about 1:24,000. When there were distinct topographic features to follow, the lines drawn were within 1 to 2 pixels of the desired location (30-60 m with 30 m DEM data). When the features were less distinct, the position of the line involved more judgement. Often, selected geology polygons were displayed on top of the shaded relief with hatched patterns to guide the digitizing. In this situation, the horizontal accuracy could vary significantly depending on whether the delineation closely followed the relief or the geology.

Since the accuracy associated with a given delineation depends on many factors and since these factors may vary for different parts of the delineation, it is difficult to specify a single number that reflects the true accuracy situation. Lines based on distinct features visible in the 30 m shaded relief displays are going to be more accurate than lines based on 90 m data or lines based on the geology coverage or the hydrologic units coverage. In an effort to document the source/criteria for each delineation and to provide some sense of accuracy, the arcs in the Landtype Association coverage were attributed with a 3-digit code in the SOURCE field. Codes ranging from 101 to 199 indicate lines obtained directly from other coverages, while codes from 201 to 299 indicate lines digitized on-screen with the specified background display. In addition to the SOURCE field, each arc has a NOTE field that contains a brief description of the criteria used in making the delineation.