A DOQ TEST PROJECT: COLLECTING DATA TO IMPROVE TIGER

Introduction

DOQ Test Project

One problem encountered while creating new header files for Newberry County was finding the appropriate value for skipbytes. It had been assumed that the value was fixed and would remain the same for all DOQ images. This was incorrect and affected the alignment of the DOQ image to other DOQ images and to the TIGEr/Line file in the ArcView display. To initially resolve this problem, the correct information for the new ASCII header file was located manually; each item was determined separately. Once the problem was pinpointed, investigation uncovered an executable that was downloaded from the Internet and that read the original header file and organized each item according to its name.

1. APID - A sequential number given to the DOQ anchor point from the TIGEr/Line anchor point file.
2. Rating - A rating scheme was developed by the GRaSS to assign each DOQ anchor point a grade that represented the quality of its placement. The following table shows the five ratings and their meaning.

   Rating	Meaning
   3	Excellent Placement
   2	Fair
   1	Unsure
   0	Best Guess
   -	Cannot be used

   Figure 3. DOQ Anchor Point Rating.

   Each rating was somewhat of a "guesstimate" by the analyst. Because of this, ratings varied among analysts.

3. Reason - The GRaSS felt it was important to know the reason why analysts found some DOQ anchor points easier to place than others (and therefore assigned the DOQ anchor points different ratings). The GRaSS developed a classification scheme for identifying these reasons.

   Problem Associated with DOQ
   D DOQ	R Road		U Unidentifiable
   	H Hydrography		V V-type intersection
   	I Intersection		S Star intersection
   	L Railroad		P Pinpointing of intersection is difficult
   			T Vegetation obscuring feature
   			L Parallel features make issues
   			B Blurs with background/other features
   			H Shadow obscuring feature
   			O Other
   Problem Associated with TIGEr/Line
   T TIGEr/Line		R Road	A Additional AP not on DOQ
   		H Hydrography	M Missing AP caused by merging AP
   		I Intersection	L Lost road or railroad with random AP
   		L Railroad	C Confusable with roads on DOQ
   			T Topologically misplaced
   			S Shape same as DOQ but displaced (>50m)
   			N TIGER road not on nearby DOQ roads.
   			D Digitized poorly
   			R Scale is poor but shape is ok
   			V V-type intersection – TIGER is jagged

   Figure 4. DOQ Anchor Point Classification Scheme.

   For example, the classification DHH denoted that on a DOQ the intersection of a stream and local road was difficult to plot due to a shadow obscuring the location of the stream. (The first D indicated the problem being associated with the DOQ, the first H indicated the stream as a hydrographic feature, and the final H indicated the specific problem, in this a case a shadow obscuring the stream.)

4. Type - The TIGEr/Line features that were intersecting at the TIGER anchor point were identified and recorded.

   Intersecting Feature Identification
   H	Highway
   L	Local Road
   J	Jeep Road
   R	Railroad
   S	Stream

   Figure 5. Intersection Feature Identification.

   For example, if a local road and a stream create a four-way intersection, it is identified as L-L-S-S.

5. Interesting - DOQ anchor points that were particularly interesting or an exceptional case were noted with a "Y" for further investigation or to be used as examples. Otherwise, the field was left empty.

Analyst Thoughts

Determining whether an anchor point was easy or difficult to plot depended on the qualities of both the TIGEr/Line feature and the DOQ image. GRaSS analysts discovered that DOQ anchor points were most easily plotted when:

• An intersection on a DOQ had defined edges, and the shade of the road, stream or railroad was not confused with other items in the nearby area.
• Vegetation cover did not obstruct features.
• The TIGEr/Line road feature was in close proximity to the same feature on the DOQ.
• The source from which the TIGEr/Line feature has been digitized was accurate and could be associated with a feature on the DOQ.

Analysts found it most challenging to plot DOQ anchor points when roads and streams were not identifiable due to the following reasons:

• Vegetation created a canopy over the road or intersection.
• Ground cover with little vegetation often appeared to be the same shade as a road.
• The digitized road in TIGEr/Line did not appear to follow the shape of any of the roads within the nearby area on the DOQ.
• The middle of the star-type intersection was difficult to identify.
• Vegetation and hydrographic features were both displayed in a dark shade of gray and it was difficult to distinguish between the two.
• Often there were streams that appeared to be parallel to one another but there was only one stream that appeared in TIGEr/Line.
• The digitized stream in TIGEr/Line did not follow the shape of the stream in the DOQ.

From the above lists, it was apparent that most challenges and problems that occurred during data collection were associated with feature visibility and DOQ interpretation.

When project data collectors have little DOQ experience it becomes crucial to select a GIS that is easy to learn and use. More specifically, it is important to have image tools available in the software that cater to a minimum level of understanding.

To plot DOQ anchor points effectively and efficiently, features on a DOQ must be clear and distinguishable from other features. There are several functions in ArcView that facilitate DOQ anchor point placement on the image. Equally, there are several aspects lacking in ArcView that may have made DOQ anchor point placement easier or more accurate for a junior level geographer.

Successes Using ArcView

ArcView provides tools that allow the manipulation of a DOQ image. These tools lie primarily within the Image Legend Editor. The Image Legend Editor is prompted by activating the DOQ image theme and clicking on the Edit Legend button on the tool bar. The tools found most useful while analyzing DOQs are the image contrasting tools and zooming tools.

A gray scale image can be adjusted in the Linear Lookup option in the Image Legend Editor. The Linear Lookup provides a means to perform a contrast stretch, increase the contrast, soften the image, increase the brightness, and darken the image by modifying the graph that is provided. In some instances altering the image improved the visibility of particular features on the DOQ. However, there was no particular aspect of the Linear Lookup that proved to be more beneficial than the other.

When analyzing a DOQ image and placing a DOQ anchor point, it is necessary to view the image at several different scales. The appropriate scale for an image to be at when placing DOQ anchor points varies depending on the particular DOQ image, the individual viewing the image, and the feature that is being analyzed. ArcView has several zooming tools that are helpful when placing anchor points such as Zoom In, Zoom Out, Zoom to Previous, Zoom to Full Extent, and Zoom to Active Theme.

Aside from image tools and DOQ analysis, ArcView shapefiles are documented and can be manipulated. For example, the GRaSS chose Microsoft Visual BasicÒ programming language to manipulate shapefiles by adding data and extracting data. The GRaSS was then able to load the manipulated files back into ArcView for visual interpretation.

Features Lacking in ArcView

Determining what tools are lacking in ArcView is difficult when an analyst has limited knowledge of photographic interpretation and GIS capabilities. How does an analyst decide what would benefit future DOQ projects without sufficient experience? If DOQs were adopted as the method to update TIGER at a national level, junior grade staff would be employed as primary operators. It is essential that the GIS software used for the project aid the analyst who is not an expert, nor wishes to be.

In an ArcView View window, zooming in and out of an image several times for the purpose of placing one anchor point can be time consuming. Each DOQ image is approximately 50 megabytes in size and it can take several seconds to redraw a simple image. Although several seconds does not seem significant, there are thousands of TIGER anchor points per county and the image can typically be redrawn five or more times per TIGER anchor point. In order to perform tasks in a timely manner, only one to two DOQ images can be active while placing DOQ anchor points. It would be helpful if ArcView were equipped to load large images at a faster pace.

Oftentimes, the clarity of features can vary within a single DOQ. It is necessary to manipulate the image several different ways to accommodate each DOQ anchor point that is being placed. When altering an image in ArcView, the entire image (one DOQQ) is affected. ArcView does not have the capability to adjust only one section of interest on the DOQ.

Overall, the ability to customize an ArcView project that caters specifically to gray scale image enhancement would be ideal. The above evaluation was written with the knowledge that a new ArcView Image Analysis Extension exists and contains features that may be of use to the DOQ project. However, the extension was not available to the GRaSS during the first phase of the DOQ test project and it is unclear that the extension will solve all the problems mentioned.

The analysts for the Newberry County, SC test project took approximately 200 hours to place 3,723 DOQ anchor points at a rate of approximately 20 DOQ anchor points per hour. After completing the first phase of the test project, the GRaSS believes DOQs are an efficient medium for use in data collection. However, the use of DOQs on a national level cannot be decided until the second phase of the project is complete, the results are validated, and all problems that have been encountered are addressed. Although the DOQ test project has thus far been successful, there are several components that need to be explored further. These components include potential software developments, image enhancement tools, and improving analyst interpretation skills.

1. The use of brand names within this paper does not represent an endorsement of a company or its products.

2. The Federal Geographic Data Committee's endorsed 1998 National Standard for Spatial Data Accuracy (NSSDA) is intended to replace the 1947 National Map Accuracy Standard. The NSSDA was not available when the GRaSS was selecting the medium.

1997 TIGEr/LineÒ Files Technical Documentation/prepared by the Bureau of the Census-Washington, DC, 1997.

Environmental Systems Research Institute, Inc. ArcView GIS Version 3.1 Documentation. 1998.

Federal Geographic Data Committee. A Proposal for a National Spatial Data Infrastructure Standards Project. URL: http://www.fgdc.gov/standards/documents/proposals/progpas3.html. 1999.

Godwin, Leslie. (1998). Improving TIGER: A DOQ Test Project. In: Proceedings, GIS/LIS '98, Annual Conference & Exposition.

United States Geological Survey. Digital Orthophoto Program. URL: http://mapping.usgs.gov/www/ndop/. 1998.

Kristen O'Grady
U.S. Bureau of the Census
Geography Division, GRaSS
GEO 7400
4700 Silver Hill Road
Washington, DC 20233-7400
Telephone: (301) 457-1056
E-mail: kogrady@geo.census.gov