Kansas Geological Survey, Current Research in Earth Sciences, Bulletin 253, part 2
Prev Page--Drainage System, Summary
Kansas Geological Survey, Current Research in Earth Sciences, Bulletin 253, part 2
Prev Page--Drainage System, Summary
In all, 192 7 1/2-min U.S. Geological Survey digital elevation models (DEMs) were used (fig. 3) and were grouped in turn into 16 "large" quadrangles labeled A to P (figs. 1 and 12). Each large quadrangle is composed of 12 7 1/2-min DEMs with pixels 30-m square and 1-m vertical resolution.
Several of the Level 1 DEMs exhibited pronounced banding patterns that required filtering to reduce the "artifacts." U.S. Geological Survey DEMs are classified as Level 1 or Level 2 depending on the presence of undesirable artifacts that arise in the digitizing process. Level 1 DEMs exhibit undesirable banding artifacts, whereas Level 2 lack them. The degree to which banding patterns are incorporated differs widely from quadrangle to quadrangle. Some are so poor that they are of limited usefulness.
Instead of using the seamless National Elevation Database (NED), we used individual DEMs because the NED was not available when we began our work. Although the NED has already been filtered to reduce banding, we suggest that our filtering procedures reduce artifacts more successfully than NED's procedures.
The artifacts include banding patterns, which were removed partially from Level 1 DEMs with a low-pass filter. The filter essentially smoothed the elevation surface by assigning new elevation values to each pixel, based on the average elevation of pixels located to the east and west of the pixel being processed. Because of the generally east-west extension of the banding patterns, the filter did not consider elevations of pixels located to the north and south of a given pixel. An ArcView© Avenue script was created for the filtering process, and the optimal parameters for the filtering were determined by trial and error. (Avenue is an object-based programming language built into ArcView© 3.x.).
Once the artifacts had been removed or subdued, the individual DEMs were combined into larger, composite DEMs that form each of the 16 large quadrangles. In turn, the large quadrangles were combined into two large composite DEMs, one representing the experimental area, and other the entire study area, as outlined in figs. 1 and 5. Hillshading for each of the composite DEMs was created using the Spatial Analyst extension of ArcGIS©. The Spatial Analyst extension is a software add-on that provides raster-processing and spatial analysis capabilities to ArcGIS©.
Actual chert-gravel deposits known from geologic mapping were represented digitally by drawing map polygons by hand on 7 1/2-min quadrangle topographic maps. The polygons were then digitized with ArcView© employing a large digitizing tablet and were stored as an ESRI Shapefile. The Shapefile is a vector-based spatial data format in which discrete points, lines, and polygons can be used to represent geographic features. Lineaments were manually marked with draftsman's tape on a large-format version of the shaded-relief map illustrated in fig. 5 and digitized in the same manner as the gravel deposits.
The two theoretical chert-gravel layers as illustrated in figs. 8, 9, and 11 are represented by Shapefiles, similar to the representation of the actual gravel deposits. Creation of the Shapefiles to represent the two theoretical gravel layers involved several processes, and three spatial data models, namely vector, triangulated irregular network (TIN), and raster. An Avenue script was created to automate these processes and the ArcView© user interface was customized to allow the input of the numerical parameters that define the geometrical properties of each of the two theoretical gravel layers, as follows:
This information was used to create points in three-dimensional space in vector format, four points being created for the lower gravel layer and four for the upper layer. Each set of four points was then used to create a TIN to define a plane in 3D space representing the bottom of the gravel layer. The two TINs then were converted into raster grids to facilitate the raster-based map algebra used in defining the location of the theoretical gravel layers. Two additional raster grids were created to represent the top of each gravel layer and were positioned in 3D space by adding the thickness assigned to each layer to the grids representing the underside of each gravel layer.
These operations defined the top and bottom grids for the upper gravel layer, as well as the top and bottom grids for the lower gravel layer (figs. 8 and 9). Map algebra was performed on these four raster grids, together with the composite DEM for the experimental area, to provide the location and extent of the two theoretical gravel layers.
The logic operation used to generate a binary yes/no grid for each of the two theoretical gravel layers is as follows:
binary grid = composite DEM >= bottom grid AND composite DEM <= top grid
This procedure resulted in grids for each theoretical layer that represented the grid cells of the composite DEM where the topographic elevation is equal to, or greater than, the bottom of the theoretical layer, as well as the cells where the topographic elevation is equal to, or less than, the top of the theoretical layer. Then, the grid cells that "contain" the theoretical gravel layer were labeled with ones, and the nongravel cells labeled with zeros.
The final step involved the conversion of the two binary grids into Shapefiles that look almost identical to the raster grids, in that they define locations where the theoretical gravel layers are "preserved." These Shapefiles were created because they require less storage space and processing power than the raster grids from which they were generated, and can be displayed and printed more readily.
Prev Page--Drainage System, Summary
Kansas Geological Survey
Web version Dec. 26, 2007
http://www.kgs.ku.edu/Current/2007/Harbaugh/06_app.html
email:webadmin@kgs.ku.edu