GISdevelopment.net ---> GITA 1997 ---> Fundamental & Economic Issues of AM/FM/GIS

Lower Costs and Higher Quality Data Through Maintenance

Robert Kelley
Serion International, 1965 N. 57’1’Court, Suite 202
Boulder, CO 80301


Abstract
The importance of a coherent maintenance process is frequently underestimated. These procedures must form part of the overall GIS design and be put into practice during the implementation phase--before start-up. Yet maintenance all too often remains a second priority even after GIS implementation as new users and applications proliferate and resources stretch to the breaking point. If new data is not consistently integrated into the system, the enormous investment in a high quality GIS can quickly be undermined or lost. Hiring restrictions, insufficient trained staff, lack of adequate equipment and many other factors can prevent the establishment of a consistent maintenance program. However, the maintenance process actually represents an excellent opportunity to improve the quality of the information generated by a GIS, Using the same techniques employed during the conversion process to design a maintenance process helps guarantee that the same specifications and the same level of quality are maintained throughout the life of the GIS. The use of Total Quality Management (TQM) keeps costs down by focusing on the development of processes which prevent errors from occurring rather than relying on end-stage quality control steps to find and correct them. The results are lower costs, better data and shorter maintenance cycles.

Lower costs and higher quality data through maintenance
Data maintenance is not new. Spatial data had to be maintained when it was stored on linen, paper and mylar in flat files. Data management and products have changed dramatically, however, as manual spatial data has evolved into digital data and been used as the building blocks of geographic information systems. So, too, must maintenance processes.

It is much faster and easier to update a screen image than it is to redraw an entire paper map. Ironically, this very fact may be part of the problem. It is so easy that it is tempting to assign maintenance to a technician with little or no training and to apply few guidelines or standards. In fact, GIS maintenance seems so easy that it is often overlooked during the design of a GIS and, more importantly, during the allocation of resources, as well.

There are three excellent reasons to pay very close attention to spatial database maintenance:
  • As one of the greatest expenses over the life of a GIS and one which requires continuous funding, maintenance represents one of the most logical areas in which to cut costs, especially because there are ways this can be done without sacrificing quality.
  • The maintenance process offers excellent opportunities to improve the accuracy and quality of the data in an incremental and cost-effective manner.
  • It is possible to lose the entire investment in data if maintenance is neglected.
A GIS and the many services it provides simply accelerates and intensifies the need to maintain the database in a systematic and efficient manner. Merely mimicking manual data maintenance methods is inefficient and expensive. A good analogy for this common practice is the initial use of computers to do the same tasks in the same ways as adding machines and typewriters had always done them. It soon became apparent to some Bill Gates-types that the tool at hand was far more powerfil than the old manual devices once used to perform these fi.mctions. New processes -- even new paradigms -- were in order.

A paradigm shift in the design of the maintenance processes used to keep spatial databases current is also needed. Maintenance should not merely keep data current, but also provide an opportunity to continuously improve the quality and the accuracy of the database. And it should do so at a cost lower than the GIS is currently experiencing.

The Conversion--Maintenance Connection
The GIS processes which most closely parallel the maintenance process -- and it is a process, not just an onerous task -- are conversion and integration. Most data does not initially exist in electronic form; it starts out as a work order, a survey report, a diagram, a legal document--a piece of paper. If the data is not digital, it must be put into electronic form. Digital data then must be integrated into the GIS. This is essentially conversion and integration on a small scale, happening in an incremental manner throughout the life of the GIS.

When a GIS manager contracts for conversion or integration work, specifications are written, QA steps are put into place, accuracy standards are established, and so on. This occurs because it is recognized that the quality and the accuracy of the data are two of the most important factors which determine the overall quality of a GIS. Conversion and integration are thus seen as critical steps in the GIS creation process. Moreover, the creation of a high quality spatial database can constitute the single most expensive aspect of the GIS, often costing more than hardware and software combined.

This is equally true of the maintenance process. Streamlining this process can save many, many dollars, and if done properly, can simultaneously lead to quality that improves, rather than deteriorates. Process engineering techniques provide the keys to achieving both cost savings and quality improvements.

Attaining Zero Defect
Total Quality Management is an innovative approach to ensuring quality. It was developed during the 1950s by W. Edwards Deming, who is widely considered to be the “father of QC.” 324?At the heart of this paradigm shift in quality control is fixing the process when an error occurs, rather than merely correcting the error. It is, after all, cheaper to do something once than it is to do it twice.

Processes designed so that that errors cannot occur are known as Zero Defect Processes. Their use allows Statistical Process Control to replace the expensive and time-consuming 100 percent QC step that has traditionally formed the last lapin data processing. This approach can ensure that the quality and accuracy standards established for the conversion process are adhered to during the maintenance process, as well, at a cost significantly lower than traditional methods.

Making Maintenance Wrok
There is an entire palette of ways to improve the maintenance process once the underlying structure has been addressed by applying TQM principles and techniques. Three of the most cost-effective of these concepts involve establishing a complete set of clearly spelled-out specifications for all data in the system, the use of process engineering to redesign the entire data entry process, and the use of QA Metadata files. Each of these is briefly discussed below.
  • Specifications
    Specifications always start with the end user, who may be external or internal to the organization. Because quality and accuracy are expensive, it is important to establish accuracy specifications that meet the needs of the most demanding user--not more, not less. Conversion specifications are a good place to start when developing maintenance specifications. TQM and Zero Defect processes will ensure that these specifications are met in a consistent, timely and cost-effective manner. These clearly defined specifications also allow automated checking for data integrity--an important part of statistical process control.

  • The Maintenance Plan

  • Draw a flow chart that maps the route currently followed by each type of data, from inception to GIS. Then think about three things:

    1) At what points can data be lost or altered?
    Design processes that foresee all the ways in which this can happen and then embed preventive measures into the process itself. For instance, rather than have an engineering technician hand a GIS technician a piece of paper with data that needs to be entered, could the engineering tech submit the data in such a way that it does not have to be re-entered? At this stage, data is like food; the less it is processed, the better.

    2] How can new (or old) technologies streamline this process?
    Consider the Internet, handheld computer units -- even Post-it Notes, if these would improve the process and avoid opportunities for error. Also search carefully for ways in which technology may be producing, rather than reducing, error.

    3)What level ofservice is needed?
    This is the basis for the hlaintenance Plan. It is a written document that addresses each type of data in the system in terms of

    • how frequently must the data be updated?
    • how current must the data be?
    • how easy must access be?
    • how accurate must the data be?
    • what risks are run if the data is inaccurate or out of date?
    • how quickly must users have access to each type of data?

    Once these questions have been answered for every type of user and every kind of data, a cost analysis can be performed to establish the least expensive way to meet these parameters. In-house, partial out-sourcing and full out-sourcing are all options.

  • QA Metadata Files
    QA Metadata files area method of identifying the spatial accuracy of each segment or node, as well as its source. Having a QA Metadata file allows a GIS manager to quickly and easily prioritize quality improvement efforts, starting with the least accurate engineering drawings, for instance. The development of these standards and the identification of sources are jobs for the engineering department, but think carefully about the most efficient way to attach this information to the corresponding spatial data.

    QA Metadata files can be constructed wholesale during the conversion process; they can then serve as a guide for integrating data accuracy improvement work into the maintenance process. But the construction of the metadata file itself can also be part of the maintenance process, done on a sector-by-sector, incremental basis.
Well defined and consistently defined metadata standards also permit automated verification of data quality--feedback, if you will. If the metadata parameters are exceeded, there is a quality problem.

QA Metadata Example

Pipe-lD Diam Location Material Source Code ‘%0 of stretch Rotate
1234 .5 ft. 27.25ft. Pvc 124 -1,3% 3.6

The source code identifies the type of document the feature was captured from, as well as the conversion method used. The ‘Aofstretch represents the difference between the attribute and the GIS measured feature, indicating how much a node or line changed in size when it was entered into the GIS. Rotate indicates the angle of rotation that took place between the source document, transformed to the State Plane Grid, for instance.

Geographic information systems, being infrastructure tools used to offer many different types of services, often bridge numerous departments or sections of an organization. For instance, a water utility may superimpose its data over the city’s planimetric base map. Of course, this means that when the assessor’s office performs its routine maintenance of the base map, water main information will probably need to be updated, too. This type of interdependency requires the development of excellent data processes and well-established communication channels, Some organizations outsource all data maintenance for several layers from all user departments to a third party. This is being done more and more, particularly to ensure data integrity and to guarantee that all layers are adjusted to the same standards.

Maintenance is not a chore that can be handled ad hoc by low level technicians. Over a period of months or years, this approach to maintenance can easily pollute the data quality that was so carefidly established during start-up. This is particularly true in organizations plagued by understaffing, high turnover and too few training dollars. It can also come to pass simply by virtue of the success of a GIS. Unanticipated demand by unforeseen users can hamper the ability of a GIS staff to merely provide basic services; maintenance is put off until someone has time to get to it.

Maintenance may seem like a crisis sometimes. Lack of it can certainly cause crises. But, in Chinese, “crisis” also means “opportunity.” Innovative maintenance represents an opportunity to simultaneously save money and improve GIS data quality.

© GISdevelopment.net. All rights reserved.