Paper No: 363

Title: GIS for Historic Analysis

Author: Barbara Hough

The U.S. Air Force Academy has experienced a great deal of development within surrounding communities. As Colorado Springs continues to grow, commercial and residential developers will continue to acquire desirable land adjacent to the Academy. It is important to understand the historical record of this urbanization in order to project trends and understand potential encroachment concerns as well as impacts to the natural resources of the Academy that will result from this development. This paper will address how GIS will be used to create, visualize, and analyze the data.

In 1954, Congress authorized creation of the Air Force Academy. The ten secretary of the Air Force, Harold E. Talbott, appointed a commission to assist in choosing the permanent site. After traveling 21,000 miles and considering 580 proposed sites in 45 states, the commission recommended three locations. From those, Secretary Talbott selected the site near Colorado Springs. The state of Colorado contributed $1 million toward the purchase of the property.

The staff and faculty of the U.S. Air Force Academy, in the interest of our future national security, molds outstanding young men and women into Air Force officers with knowledge, character, and discipline; motivated to lead the worlds greatest aerospace force in service to the nation.  Before its graduates enter various flying and support specialties, the Academy trains them to be, first and foremost, Air Force officers.  Of the 34,065 cadets who have graduated in 43 classes, more than 51.2 percent are still on active duty. In addition to the mission to develop outstanding young men and women, the Academy is also an active installation. The 10th Air Base Wing was re-activated activated on Nov. 1, 1994 as the support wing for the U.S. Air Force Academy, Colorado.

The Air Force Academy is located at the base of the Rampart Range on 18,500 acres. The elevation is 7,163 feet above sea level at the Terrazzo level of the installation.

This project involved the use of Geographic Information Systems (GIS) to develop an historic land use database. GIS has the ability to create, store, edit, visualize, analyze and present data needed to study the historical urban growth of the area. Understanding past land use trends and their environmental impact was crucial to improving current environmental management practices. Historic spatial and tabular data were developed to quantify the changes in area for all water bodies, watercourses, roads (dirt and paved), vegetation and major structures/built up areas (including paved areas), for each selected year.

An Area of Interest (AOI) was defined that included the Academy boundary and several miles in each direction. This ensured that development trends were clearly visible.

The creation of historic spatial data began with a large effort to research all existing resources. The investigation for historic raster or vector data began with federal, state and local agencies. These organizations include USDA, NIMA, BLM, El Paso County, the city of Colorado Springs, the Pikes Peak Council of Governments, USGS, the Army Corps of Engineers and several commercial vendors (including Pixxures, American Reprographics, Intrasearch and Airphoto USA). It was determined that there was no existing historic spatial data that would be of value. Historic raster data was available from a variety of sources.

It is imperative to clearly communicate to the USDA office in Salt Lake City the imagery required. This can be a confusing and time consuming process. It is recommended that anyone needing historical imagery should clearly identify their AOI and communicate that to the USDA office through maps and a concise list of the Sections, Townships and Ranges involved. USDA will return a quote for the imagery that includes a series of photo names. These names are quite ambiguous but can be verified through a map index that the USDA office can mail to you. The local USDA office may have some photo indices on hand for verification. This process can be tedious, but it is critical to quality control the order thoroughly. The time and expense of miscommunication can make a large impact on a project quite quickly.

Decisions regarding the quality of the imagery will also need to be made. The geometric resolution of the scanned imagery varies. This resolution is defined in terms of microns. To put it very simply, the smaller the number of microns the more Dots Per Inch (DPI) and better resolution. The price of the imagery is not affected by the scanning resolution. The cost does increase when considering the CD’s and shipping costs as the file sizes grow. For this project we decided on 17.5 microns. Each image is approximately 373 megabytes. A total of 235 CD’s were shipped at a cost of $10.00 per CD.

As the imagery began to arrive incrementally, we began to compress using Lizardtech software. The process to move the SID files to the accurate location was very simple. We added the USGS 1:24,000 Digital Raster Graphic (DRG) dataset quadrangles to the view with a historic image. Then we made the image the active theme for the georeferencing toolbar and added control points to move the image relatively close to the area of interest.

There are two very important steps to this process. It is essential to remember to set the appropriate image as active in the georeferencing toolbar. Without attention to this detail the GIS Technician will find that images are unintentionally warped beyond repair. It is also very important to remove each control point added while moving the image near the appropriate location.

If the control points are allowed to remain the image will warp considerably. If an error is made in either of these areas, the GIS technician should remove the image from the project and add it again.

Once the image was moved to the area of interest the map index was used to identify the location of the photo for accurate placement. It is very helpful to make the image transparent so that the USGS quads are visible underneath. The magnification window is also extremely helpful in placing imagery.

Once the image is in the appropriate location, control points are added to secure accurate placement.

The technician should get at least four control points evenly spread across the image to ensure that the pixels are stretched proportionately across the image.

At this point the image is ready to update. The Update Georeferencing option will create a world file for the image. A world file is an ASCII file consisting of 6 lines. The Eastings and Northings of upper left corner of the image are noted as two separate lines in this file. The pixel size is noted as an X and Y length for two additional lines. The final two lines indicate the second and third parameters for X and Y rotation. A world file for a tiff image will have the same name as the tiff, however the extension is tfw. For a sid file the extension is sdw. It is important to keep the world file in the same directory with the same name as the image. We discovered that the Update Georeferencing option will not work with control points placed with second or third order polynomial affine. We found it necessary to get this level of accuracy when working with images on a steep slope.

These images were then tested in a variety of ESRI software packages and operating systems to ensure that the SID files would appear accurately registered regardless of ESRI software version and operating system at the Academy. The images appeared shifted when viewed in ArcView 3.2 and ArcMap 8.2. Technical support for ESRI was contacted to identify the problem. The georeferencing tool creates an .aux file to carry the rotation of the warped pixels. ArcView 3.2 cannot read an .aux file.

ESRI technicians informed us that this is a common problem. They recommended that we use the "rectify" tool after creating the SDW file. This process required additional virtual memory and RAM to be added to the GIS computers. The rectify process was time consuming and only worked occasionally. Another call to technical support brought us to the conclusion that this process works best with images in the native format processed locally on a computer with a great deal of RAM and virtual memory. We were able to take the sdw files and change the extension to tfw. After copying the tiffs back into the directory with the tfw file the rectify process went smoothly. Files without an sdw file were converted to a tif format in ArcCatalog and rectified. Further testing confirmed that the rectification allowed the images to be viewed accurately by any ESRI software product.

We were able to adjust the symbology of the raster under the properties dialog window.

This code was used with the macro tools provided with ArcMap

Defining the spatial reference is the most difficult part of creating the SDSFIE geodatabase. These values should be set with consideration of all existing and potential data, as they cannot be changed in ArcGIS once defined. The spatial reference involves the coordinate system, precision and the extents of the geodatabase.

Precision defines the number of storage units in one map unit. The dataset precision is determined by dividing map units by storage units. If the data will be integrated with existing datasets, it is important to know the current precision so that the new data will not modify the accuracy of the existing data. Double precision in an ArcInfo coverage provides sub millimeter accuracy.

This process is well documented in several white papers available from the SDSFIE website.

With the geodatabases built, we were ready to begin digitizing our data. This process involved adding the raster catalog, the USGS DRG quads and the geodatabase. We worked at a scale of 1:10,000 to be sure that we were consistent in our level of detail when digitizing. We soon discovered that while the raster catalog was a convenient way of delivering the imagery, it was more beneficial to us to work with individual images because we could change the order of the images at will. When there is overlap of imagery, one photo might show the detail more clearly. It has been helpful to be able to restack the layers to find the clearest picture.

This project has been extremely interesting and challenging. The Air Force Academy will be able to use this data to visualize the development of Colorado Springs over the last 50 years. From decade to decade we can see the roads and developed areas increase while the vegetation decreases. It has been interesting to see the change in water bodies and watercourses relating to the additional roads and development. We have been able to quantify the changes to our community in terms of area and use this data to study trends in encroachment and natural resources.

1. Research historical data sources thoroughly to get the best vector and raster data to use for a foundation. The USDA is an excellent source for historic imagery.
2. Identify the Section, Township and Range of your area of interest to communicate the area for required imagery.
3. Ask the USDA to inform you of what imagery is available for your area and year of interest. Don’t forget that they handle aerial photography taken for the Forest Service as well.
4. When ordering historic imagery be sure to quality control the requisition so that you get the photos you expect.
5. Remember to order photo indices along with the aerial photography to translate cryptic file names into actual locations.
6. Determine the resolution needed before ordering imagery. Understand that the file size of the tiffs will require more CDs and will therefore increase the cost of the order. If possible, ask for samples of the same image scanned at different micron levels to compare.
7. Work with tif files if at all possible. If size is a concern, compress to sid as a final step before creating the raster catalog.
8. Rectify imagery to write the rotation of the pixels into the header file of the image. End users of the imagery that are still working in ArcView 3.X and this software cannot read the .aux file that holds the rotation of the pixel.
9. When georeferencing be sure to activate the image to be moved and remove all control points until the image is as accurately placed as possible.
10. Use at least 4 control points placed in corners of the imagery for best results.
11. When rectifying images, work on a hard drive and not across a network. RAM and virtual memory are more important than processing speed in terms of system requirements.
12. Adjust the symbology of the rectified raster to display background at a value of 255 to eliminate the overlap of the white boxes.
13. Don’t be afraid to call ESRI technical support and use code samples to work with raster files.
14. The CADD/GIS Technology Center has excellent tools and white papers available from their website for the development of SDSFIE compliant geodatabases.

Barbara Hough

barbara_hough@rexroadapg.com