The U.S. Census Bureau is most well known for its decennial enumeration of citizens, but it’s also recognized within the geospatial industry as the developer/caretaker of nationwide GIS datasets. 

The U.S. Census Bureau is most well known for its decennial enumeration of citizens, but it’s also recognized within the geospatial industry as the developer/caretaker of nationwide GIS datasets. In the 1960s, the bureau developed a GIS called Dual Independent Map Encoding (DIME) to handle spatial data. This system was replaced in 1990 with the Topologically Integrated Geographic Encoding and Referencing (TIGER) database, which was linked to the Census Bureau’s Master Address File (MAF) and enabled geocoding of street addresses.

As both public and private industries have more frequently relied on the availability of such information and are continually enhancing the accuracy of their own data, the need to ensure the accuracy of the MAF/TIGER data has increased. In 2002, the Census Bureau embarked on an initiative to produce a highly accurate (7.6 meters or better) street centerline “digital map” (geographic data base) of the entire United States, Puerto Rico and the associated Island Areas. Called the MAF/TIGER Accuracy Improvement Project, or MTAIP, the project aimed to correctly locate every street and other map feature in the TIGER database as well as each MAF address and implement an effective automated feature change detection methodology. The Census Bureau’s Geography Division recognized that for the MTAIP initiative to achieve this goal, there was a crucial need for accurate geospatial control that could be used to perform qualitative assessments of current and future submitted datasets seeking inclusion into the bureau-maintained MAF/TIGER database.

To develop that control, the Census Bureau launched the Accurate Coordinate Datasets Collection (ACDC) project in 2005, led by Michael Baker Jr. Inc., a professional engineering and consulting services firm headquartered in Moon Township, Pa. The firm was chosen based on its proven capabilities, prior experience in related activities, its ability to handle shifting workloads, and its familiarity with the concepts, practices and procedures of collecting accurate street-intersection points for TIGER horizontal accuracy verification due to prior participation on the bureau’s MTAIP initiative. The objective was to provide spatially accurate ground control throughout the entire United States and U.S. territories for use in support of the Census Bureau’s MTAIP. However, the Census Bureau quickly came to rely on Baker for more than data. “Very soon after the beginning of the contract, we realized the added value in Baker’s selection as trusted advisors when predetermined procedures/requirements put forth by the Census Bureau presented certain issues in the field,” says J. David Bush, the bureau’s contracting officer’s technical representative (COTR). “In each case, Baker not only presented us with an issue, but they delivered alternative solutions from which to evaluate and consider.”

Mission planning, sound logistics and interdepartment/interagency coordination were integral facets of Baker’s approach. 

Serving as the program manager, Baker led a team of eight large, small and disadvantaged businesses that were assembled based on proven expertise, geographic location, and/or prior participation in ACDC-related activities. In this role, Baker coordinated the teams’ resource pool to function as a seamlessly integrated workforce for task order execution. To guarantee that only the highest-quality data was submitted to the bureau, Baker designated one team member to perform final quality assurance/quality control (QA/QC) assessments throughout the entire project. This approach ensured a continuous impartial and objective eye reviewing forthcoming submissions and further granted a level of familiarity with delivered datasets through standardization and consistency. All other team members were used to perform GPS field collections through the contiguous United States and Puerto Rico and were responsible for capturing 40 percent of the tasked collections.

To aid each collector in the determination of viable intersections and to promote continuity/consistency between the team members, Baker developed unique data capture methodologies, techniques and standards that were demonstrated first-hand to the COTR during a week-long trial. Baker also developed and implemented standardized training programs for field safety, GPS equipment testing (performed at geodetic bench marks or monuments) and data collection. Staff training consisted of both classroom and field coursework and also entailed biannual “ride-alongs” by GPS supervisors to verify that the required process outlined in the task orders were being executed as directed.

By understanding the vital need for a nationwide dataset of accurate control points, Baker and the Census Bureau formed a working partnership that was well-equipped and capable to perform the largest such geospatial undertaking of its kind. During the four and a half years of field collections, the Baker team was issued 22 separate tasking orders and performed coordinate collections within 1,018 counties/municipalities spanning 37 states and five U.S. territories. Coordinates and observed road name attributions were captured to a horizontal accuracy of +/-1 meter at the center of “well-defined” photo-identifiable street intersections. All coordinate collections performed by Baker’s team were derived from GPS equipment coupled to Baker’s own GPS/GIS software, but Baker was also tasked to investigate each assigned collection area for alternative sources from which coordinates could be derived. These included public land surveys, cadastral data, aerial photography and satellite imagery. To determine the viability for accurate coordinate extraction, each potential data source was assessed for spatial accuracy, age, resolution, cloud cover (if applicable), availability, price and subject area coverage.

The Baker team performed coordinate collections within 1,018 counties/municipalities spanning 37 states and five U.S. territories.

Mission planning, sound logistics and interdepartment/interagency coordination were integral facets of Baker’s approach. By implementing and, more importantly, following a quality project management plan (PMP) throughout the duration of the field operations, Baker efficiently managed the team’s resources and adjusted collection procedures to minimize the impacts of numerous challenges, which included volcanic eruptions, blizzards, hurricanes, restricted military areas, sacred Native American tribal land and tidal changes affecting ferry schedules.

To facilitate the point-collection process, Baker employed its wholly owned and internally developed GeoLink GPS/GIS Mapping System software, which enabled the direct import of GPS signals and displayed the “live” GPS position over GIS data (both vector and raster) on the computer mapping display. Using GeoLink, each collector was able to easily navigate from one location to the next yet remain flexible enough to make informed routing modifications based on real-world variables such as traffic congestion, safety issues and road closures.

Upon navigating to a “well-defined” intersection, a collector would position the GPS antenna over the center point of the converging street segments and begin the collection sequence. The robust capabilities of GeoLink were again used to average a minimum of 75 GPS positions, facilitate the association of terrestrially observed attribution to the newly georeferenced locations, and automatically filter errant GPS readings that could degrade the accuracy of the derived point. The averaging and filtering functions were configured to ensure the highest and most repeatable level of positional accuracy, which routinely resulted in spatial locations 30 percent more accurate than those derived by GPS alone. To further enhance the collectors’ effectiveness, both audible and visual alarms were enabled to alert when GPS quality factors (number of satellites, signal strength, positional dilution of precision, velocity, acceleration, and real-time differential correction) were outside of acceptable tolerances.

To ensure each collection area comprised an acceptable spatial dispersion of control points while also focusing on point densities, intersecting road angles and road alignment curvatures, automated routines were developed to create a valid sampling of candidate intersections throughout the subject area. Using standard GIS analysis tools and point generation routines, the process began with the automated identification of all intersections throughout the subject area. Each generated candidate quality intersection point (QI point) was assigned a unique identification number (APID), which would serve as the unambiguous link between the submitted dataset and the GPS-derived coordinates supplied by Baker.

Coordinates and observed road name attributions were captured to a horizontal accuracy of +/-1 meter at the center of “well-defined” photo-identifiable street intersections.

After creating the QI points, another automated routine was employed to generate a statistical random sampling of 110 points throughout the subject area. The custom routine to generate the sample selected intersection (SSI) points was designed to assess the viability of intersecting centerlines and automatically filter candidate intersections that comprised road curvatures on any intersecting segment within 100 feet of the intersection center, and/or where intersecting segment angles were (less than equal to) 16 degrees from a true 90-degree angle. This process further ensured that the resulting control coordinates not only met the preferred Quality-1 standard but also garnered the desired spatial dispersion of points throughout the subject area, yielding more-efficient field collections by enabling detailed mission planning and navigational route generation for travel between desired locations.

Baker relied on its team of GPS professionals and teaming members to make the final determination as to the quality of each intersections’ configuration based on real-world, field-observed information. To qualify as a viable collection, each intersection was assessed for the following:

•  Centerlines of roads forming the intersection should meet at right angles (90 degrees +/- 16 degrees).

• The intersection must comprise only three three-way or four four-way intersection road segments; two-way or ≥5 intersecting road segment intersections would not be collected.

• There must be a clear view to the sky with no overhanging trees or tall buildings that could impede the reception of GPS signals or degrade the quality thereof.

• Roads should be constructed of a hard surface, such as asphalt or concrete, whenever possible.

• Selected intersections should have road names posted at the intersection.

• Selected intersections must not be composed of driveways.

After all collections within the subject area were completed, the field-captured data were transmitted to the data processing center in Baker’s Jackson, Miss., facility through Baker’s secure enhanced File Transfer Protocol (eFTP) site, where it was retrieved and placed through an arduous QA/QC regimen. These processes entailed the development of several custom-automated routines to assess, validate, bundle and deliver the final products.

The primary QA/QC utility was used to check sequence numbers, validate the attributed APIDs, spatially assess the nearest QI points, flag potentially errant values, autogenerate linear features, calculate the distance between the two correlating points and generate listings of road naming discrepancies. Additional QA/QC routines were used to autogenerate standardized screen captures displaying the collected intersections overload on the most recent aerial/satellite photography and perform the final spatial analysis to assess the submitted datasets ability to conform to the Census Bureau’s 7.62 meter (25 foot), 95 percent circular error probability (CE-95) requirement.

Baker also developed a custom project management Web portal that facilitated secure and efficient data transfer and further enabled the Bureau’s COTR to access current production activities, documentation, deliverable postings and work package histories/reports. Operating within the boundaries of the work breakdown structure, the client was never more than a few keystrokes from directly assessing the status of each work package or the project as a whole.

As with most collection area taskings, the Census Bureau had already been presented a locally or commercially supplied dataset seeking inclusion into the MAF/TIGER database, for which quality control points were needed to assess the submission. As part of Baker’s cost-saving initiative, the firm was able to regularly group collection areas to minimize mobilization costs and further perform coordinate collections for areas in preparation for future submitted datasets, which would eliminate unnecessary downstream expenditures.

Within each collection area, Baker was tasked with collecting 110 quality control points at or near the identified SSI locations. To exceed these requirements and ensure the viability of the supplied data for years to come, Baker captured an additional 10 to 20 percent within each subject area. SSI point locations were always assessed for collection first; however, on average, 25 percent of captured intersections were selected from the list of candidate QI points due to local conditions that were not favorable for the accurate collection of the original SSI point.

Baker provided consistent and standardized deliverables throughout the life of the project that yielded a 99.6 percent first-time acceptance rate. The firm completed the entire program ahead of schedule and $2.4 million under the budgeted allocation. What’s more, due to the delivery of high-level positional accuracies on generated control points (+/-1 meter) as compared to the spatial accuracy requirement for acceptance into the MAF/TIGER database (≤7.62 meters), Baker has ensured that the supplied control coordinates will be valid for use by the Census Bureau for years to come, even as it continues to enhance the spatial accuracy of its nationwide datasets.

“Baker met every one of the Census Bureau’s technical expectations,” notes Bush. “The quality and timeliness of their work exceeded our expectations, and the value of their management on this project truly spoiled us. They set standards to which we still refer in appreciation.”

Baker was selected as the Grand Award winner in the 4th Annual MAPPS Geospatial Products and Services Excellence Awards for its work on the ACDC project. For more information about Baker, Additional information about MAPPS and the Geospatial Excellence Awards can be found at