INFORMATION TECHNOLOGY/Automation protects quality of GIS data

Cities and counties across the country are using computerized maps to streamline tax parcel accounting, work order processing, construction permitting and emergency evacuation planning. Obviously, the data for these tasks needs to be accurate; map producers and users are counting on it.

But no matter how sophisticated or thorough the programs for converting geospatial data into digitized maps, numerous opportunities for errors typically exist within the production process. For this reason, automated quality assurance is stirring increasing interest.

In developing automated quality assurance software, the idea is to build a map production process that both checks for and traps errors. First, the production process is examined to identify steps where errors or omissions could occur.

If possible, work procedures are modified or added to eliminate the possibility of either. Checking software is then developed to screen data at remaining points of vulnerability. For example, the software might check that lines are touching each other, that each coded graphic item has a unique and matching database record attached and that there are no orphan database records in the file.

Manual checking is far less accurate than automated checking, especially when large volumes of information are involved. That is because time and manpower constraints generally allow only random checks, and errors are often not detected or corrected until they are reported by unhappy end users.

Moreover, it is the rare map conversion today that takes its data exclusively from existing hardcopy documents; in most cases, information must be pieced together from local maps, public works department files, private development records and USGS maps. The necessity of piecing together overlapping and conflicting data of varying quality only increases the opportunity for errors.

Subjecting geospatial data to an automated review process, however, is only one way to improve quality. Automating the way data quality is reported is another. Metadata, or data describing data, is useful in this regard.

For map producers, metadata provides valuable information about the quality of integrated data, offers a preview of a particular map and will tailor it to suit the needs of the user. It may tell, for example, when map data was captured, where it came from and what the scales of the source documents were. Prescreening with metadata files enables users to avoid acquiring and loading data that is not needed.

On smaller mapping projects, information about the quality of data components can be conveyed by embedding it in the files themselves. For instance, information on the location of fire hydrants and water lines along a city’s streets has been pulled together from different sources.

Map producers can then insert into the relational database descriptive fields that rank the accuracy of each graphic feature. Though the descriptive fields can be necessarily subjective, such rankings can clue users to possible trouble spots.

Information on when a piece of data was last updated can also be imbedded — a useful feature in datasets that tend to change incrementally over time. Thus, users can accept deliveries virtually without question, whereas, without automated checking, it would be necessary to review all the data — a task than can delay public release of the information.

Obviously, not every municipality has large mapping projects. But even with small projects, quality control issues can be critical. And, no matter what the size of the mapping effort, the relative cost savings that can be realized from error-free data are significant.

The authors are senior associate and associate, respectively, Dewberry & Davis, Fairfax, Va.