The paper below describes a site selection analysis performed by Edgetech America, Inc. and Dewberry & Davis. It was presented at the 1998 Virginia GIS Conference.
The United States Immigration and Naturalization Service (INS) is maintaining active paper records on over 26 million individuals. As part of its modernization strategy, the INS evaluated centralizing all these records, which are currently kept in 80 locations across the country into a single facility. Dewberry & Davis and Edgetech America, Inc. developed a short list, for INS senior management and eventually the U.S. Congress, of the best and safest facility locations.
This paper describes the site selection process from establishing the selection criteria to searching trusted data sets and performing the analysis itself. It also evaluates the ability of ArcView to perform the analysis and offers suggestions for improving the software.
In the Spring of 1997, Denis Roose, President of Edgetech America, was asked to support Dewberry & Davis with its INS Centralized Records Management Facility siting analysis. Dewberry & Davis is the largest civil engineering company based in Virginia, USA. Edgetech America, Inc. is a small Esri Business Partner with extensive experience in GIS, economic development and site selection.
The Start-up Situation
The U.S. Immigration and Naturalization Service (INS) was investigating the feasibility of centralizing its 80 records management facilities in one location. These facilities manage the paper records of over 26 million people with U.S. residency status, immigrating to, or being naturalized in, the U.S.
These records are considered active in that they should all be deliverable to a court or other institution overnight. It is the INS's responsibility to manage them until they become inactive and are sent to the National Archive Centers for storage. In the Spring of 1997, the INS was in the process of modernizing its operations to improve the safeguard, delivery and overall management of these records. It had identified at least two ways to achieve this: by digitizing the records and by centralizing its 80 regional record management centers.
The proposed central office would take over all the current record management activities and digitize the existing records within three years of operation.
Due to the short time allocated to this study and its political sensitivity, it was decided that the analysis would be county-based and would use federal government data to the extent possible. Over fifty data layers (criteria) were investigated. An on-site study would take place after this analysis to verify findings and refine the analysis to the sub-county level.
The methodology would also be kept simple and would contain two major steps:
Tier 1 criteria involved:
Tier 2 criteria involved:
Each criteria was assessed utilizing specially measurable indicators. Data were collected and analyzed for each indicator.
Tier 1 Indicators & Measures
Tier 2 Indicators, Measures & Weights
Most of the top 12 counties are in the Mid West.
How Did ArcView GIS 3.0a Perform?
Overall, ArcView GIS performed very well. Its Graphical User Interface (GUI) made it possible to perform this extensive analysis, including researching the data, in just a month. Most data sets were delivered to us as county-based tables. It was easy to bring this information into ArcView using its "Join" button and append the data to the county shapefile delivered with ArcView.
ArcView could become even more functional if some of its Avenue functionality is exposed to the end user. For example, ArcView supports:
- Buffering, or
But in ArcView 3.0a, this functionality is only available to programmers using Avenue. When a project has to be delivered quickly, this becomes an important obstacle. We would also like to see a TIGER conversion extension included with the software. We used the sample Projector extension often to project data sources in a common geographic reference system.
The ArcView project file (.apr) created for this project became rather large (25 Megabytes) taking several minutes to load. Loading time can be reduced by limiting the use of definition and selection sets as well as saving layouts in separate object databases or in separate .apr files. Snapshot layouts, those that are not hot-linked to views, greatly increase the size of the project file.
At the inception of this project, we considered using the ArcView Spatial Analyst extension. Raster-based GIS are very good at performing weighted analyses. While Spatial Analyst would have allowed a sub-county analysis to be performed to some extent. We decided not to use it. Even though Spatial Analyst may have broken down extremely large counties in the West, it would not have made a difference in the final result. Furthermore, INS Administrators and the U.S. Congress, our clients, are more familiar with political boundaries that they are with a matrix of raster cells. And as most of the data analyzed had been collected on a county basis, it is also more accurately communicated on a county basis.
Due to its built-in GUI, ArcView GIS was the appropriate Esri platform to perform this analysis. We were very pleased with its overall performance and identified a few areas where it could be improved.
The study was very well received by the INS. Although initially requesting a handful of copies of the report, the INS ultimately requested over 50 copies!
The author would like to thank Dewberry & Davis for giving him the opportunity to join its team and perform this very interesting siting study. The D&D team included Rob Dollison, Xiaoyi Zhang, Sue Hoegberg and Helen Malo. All made important contributions to the project.
© 1998 Edgetech America, Inc.
Last updated: April 15, 2015