|Home||Archived October 29, 2018||(i)|
This update of the Miami-Dade County soils map released in 2001 includes an attribute table for the soils polygons included in the spatial data layer.
The U.S. Geological Survey (USGS) Eastern Geographic Science Center (EGSC) developed a means for digitizing and attributing historic soil survey data for GIS analysis.
Techniques described in the process description should be viewed as suggestions only. Because source graphics vary in color and quality and because scanners vary between manufacturers, it is highly unlikely that the techniques described will achieve the same results in all circumstances. This project will however, describe several basic tools and techniques to use these tools that, with some trial and error and experimentation, will improve productivity.
This documentation describes methods used to collect linear map features. The same commercial products could be used to collect polygon features. However, some different tools and techniques would be used. Polygon delineation, though obviously important, will not be described here.
The methods used to complete this project were developed out of necessity. Time constraints imposed on the data collection process mandated that innovative solutions be developed in order to meet these time constraints.
The original maps were produced by the USDA Soil Conservation Service in cooperation with the Florida Agricultural Experiment Station.
Any use of trade, product, or firm names is for descriptive purposes only and does not constitute endorsement by the U.S. Government
Ballard, J. K., Evans, C. B., Galberry, H. S., Hinton, J. J., Powell, D. P., Truett, E., Watts, W. L., Willson, G. C., Leighty, R. G.
Scanning the Maps:
The scanner was an Ideal 48 inch pinch roller scanner. The Ideal scanner came with proprietary image processing software. However, the Ideal software was not used, because using the proprietary software would make the scanner unavailable to other users. Adobe PhotoShop was available on workstations and was a familiar tool to the data collection team. Consequently, PhotoShop became the selected image processing software.
Initially, the maps were scanned at 400 Pixels Per Inch (PPI). This preserved the greatest amount of graphic detail but resulted in files that were too large to be practical. For the larger maps stored with 24-bit image depth, file sizes were in the neighborhood of 500 megabytes. These 400 PPI images were saved for archival purposes only. The images used to extract the graphic features were resampled.
In Photoshop the 400 PPI images were resampled to 200 PPI. The maps could have been rescanned however, there was virtually no time savings to be gained by rescanning so the images were resampled to produce the 200 PPI images.
At this point in the processing it is necessary to view the color pallet and select an easily distinguishable and contrasting color to the majority of the colors in the scanned image. By using one of the "Paint Brush" tools or "Pens" it is necessary to place a moderately large swatch of this color in the margin of the image. An ideal location for this swatch is in the upper left corner of the image. This swatch will serve more than one purpose the second of which will be described in greater depth later in this document. Further processing can now be done to change the images "Mode" to 8-bit "Indexed Color". By converting the image from 24-bit to indexed color, the image file size was dramatically reduced.
The "Indexed Color" image should be viewed. The contrasting swatch should be visible and in an acceptable color. If the color of the swatch has changed and is no longer unique, the painted area should be made larger so that it is not changed during the indexing process. The number of unique colors in the image is now reduced from potentially millions to approximately 25.
The number 25 was selected as the number of unique colors to be retained in the output image. This number was somewhat arbitrarily selected, but represents roughly twice the number of unique colors used to print the original maps. The intent of indexing the colors is to reduce the colors in the image to a manageable level while not reducing them to the point where similar colors begin to coalesce.
For instance, the image will now have more than one color of blue, but it will not have many more than perhaps four or five. The same should be true of all of the other colors in the image, there may be more than one shade of black but the color black is stored separately from blue. Colors should be distinguishable. The contrasting color added to the margin should be present and unique. Once satisfied with the pixel size, the number of colors, and the presence of a unique user added color, feature processing can begin. Save this image with a new image name.
Set-up and Color Manipulations:
Before beginning the feature extraction process some Photoshop parameters need to be set. Click on the "Selection Tool", look for the "Magic Wand". In the "Magic Wand Options" window, set the tolerance to 1 and make sure the "Anti-Aliased" box is not checked.
For this project, virtually all of the polygons to be collected were bound by black outlines. Therefore, the color black is the color that needs to be distinguishable from all others.
If there appears to be a great number of pixels that are not black but should be black, they should be changed now. Most of the time all of the colors are well defined and distinguishable so that they can be processed without additional color manipulations.
Sometimes though, the color of pixels changes slightly where color changes on the map occur. For instance, where a black line borders a white area, some of the pixels may have been stored as gray. With the Magic Wand select a gray pixel that should have been black. Notice that all adjacent pixels of the same color are also highlighted.
From the "Select" pull down menu choose similar. All pixels of that color will be highlighted. With the "Hand Tool" scroll through the image to evaluate whether or not these pixels should be changed to black.
Assuming that some gray pixels should be converted to black, click on the "Eye Dropper" tool. With the Eye Dropper a new foreground color can be chosen. Select black as the new foreground color from either the color pallet window or from the image. To change the gray pixels to black, click on "Fill" from the "Edit" pull down window. Use the "Foreground Color" and select "OK". All of the gray pixels that should be black have been changed to black.
Assuming that either the colors were acceptable as scanned or that similar colors were converted, processing can continue.
Creating an Image Overlay:
With the "Magic Wand" select a black pixel. Notice that all adjacent black pixels will also be selected.
Choose "Similar" from the "Select" menu. Notice that all of the black pixels are selected.
Click on the "Select" pull down menu. Choose "Modify" then "Expand". In the dialog box enter "1" and click on "OK" to expand the selected set by one pixel. The selected set will now include all of the black pixels as well as any pixel that is within a radius of"1" from a black pixel.
Click on the "Eye Dropper" tool and select the bright, contrasting color that was added to the image before converting the image to "Indexed Color".
From the "Edit" pull down menu select "Fill" with foreground color and click on "OK". All of the selected pixels should have changed color.
Save this image to a new name. There should be three versions of the image saved at this time. Image 1 should be the 200 PPI image with 24-bit color. Image 2 should be the 200 PPI image with 8-bit indexed color. Image 3 should be 200 PPI 8-bit, indexed color with all of the black line work, and probably text too, in the contrasting bright color.
Overlaying the Images:
Open both Image 2, the indexed color image and Image 3, the image with the brightly colored delineations.
With the contrasting color image (Image 3) active, use the "Magic Wand" to select one of the brightly colored pixels. From the "Select" pull down menu, choose "Similar". All of the brightly colored pixels should be highlighted.
From the "Edit" pull down menu, choose "Copy" to put the selected set into a buffer.
Make Image 2 the indexed color image active.
From the "Edit" pull down menu, choose "Pastel" to place the information that was stored in the buffer on top of the indexed image.
Close Image 3 the contrasting color image.
Remember to save Image 2 the indexed color image that now has the brightly colored selection pasted on top of it to a new file name so that none of the previous images are overwritten.
Before beginning to edit the combined image be sure that there is nothing selected. Click on the "Select" pull down menu and choose "Deselect". Edits will only be possible within the selected set. If nothing is selected, the entire image is editable.
The image may appear somewhat messy. Text may be displayed in the bright contrasting color along with other features that would not be desirable in the final vectorized product.
This technique relies strongly on the concept that it is much easier to delete elements that are not wanted than it is to add elements manually.
With the "History Brush" tool selected choose a brush size that is comfortable from the "Brushes" pallet.
By erasing all of the elements that are not to be vectorized with the "History Brush", what remains will be the line work for the raster to vector conversion.
If line work is accidentally deleted, it can easily be recovered by going backward one step in the history pallet.
If additional line work needs to be captured, it can be added by tracing over the image with a "Paint Brush" or "Air Brush" as long as it is added using the selected bright color.
Line work should be inspected for connectivity. It may be necessary to connect broken line segments especially if dashed line symbology was used to produce the source graphic.
This method of capturing line work may seem time consuming and laborious, it is! However, through experimentation and practice considerable time-savings can be achieved.
Be sure to capture the neatline or corner tics. Without them geo-registration will not be possible.
Each survey contains descriptive text and tabular information about each soil type. To create the attribute information associated with each soil polygon, the tabular information was manually typed into an Excel file to take advantage of Excel’s forms completion capability. Soils suffixes were retained during the assignment of soil codes to polygons and the GIS was used to generate text files of all unique polygon identifiers or labels. These were used to hand edit the Excel files to make certain a record existed for every soil symbol/land cover class combination found in the digital soil map. Attribute information other than land cover class was copied from the table entry with the same soil symbol to each soil symbol/land cover combination with the same soil type.
The data in the Excel worksheets were exported to text files and edited to remove embedded Excel codes, insert delimiters, and place quotations around all character data. Then, within the INFO component of ARC/INFO, attribute file templates were created for each survey. Two templates were required because the types and formats of attribute information in each original table were different. The text files were then imported into INFO to create the attribute data files. For the final processing step, the soils attribute data file was joined to the soils boundary data file using the "soils attribute field" (i.e., soil symbol combinations for Miami-Dade cases).
The attribute data for the vector polygons were created by John W. Jones also of the USGS.
The attribute table is available separately at <https://sofia.usgs.gov/exchanget/jjones/Dade_soil_attributes.xls>.
|Home||Archived October 29, 2018|