Imagine a scenario in which street centerlines are stored in a GIS database with an accuracy of ±5 feet. In the same database, parcel data are stored with an accuracy of ±0.10 feet. And what if environmental data are stored at ±50 feet? Is that doable? As a matter of fact, presently this scenario is more the rule than the exception.
Imagine a scenario in which street centerlines are stored in a GIS database with an accuracy of ±5 feet. In the same database, parcel data are stored with an accuracy of ±0.10 feet. And what if environmental data are stored at ±50 feet? Is that doable? For example, data in Google Earth can have any range of accuracies.
One scenario that illustrates this would be the following: Imagine a GIS database where a city boundary is stored within 100 feet, the corresponding street centerlines within 10 feet, sidewalks within 2 feet, hydrants within 0.5 feet, and mechanical detail in the hydrants within 0.01 feet. Here, different layers would have differing accuracies.
Another scenario would be the recording of road centerlines outside the city within 20 feet, in the suburbs within 10 feet, in the city generally within 5 feet, and in the downtown core within 1 foot. Here, different accuracies would be found in different geographic locations.
As one can see, all these scenarios are possible and are current fact. Having an accuracy that relates only to a certain data layer, or to a certain geographic location, or a combination thereof, is eminently practical, since considerable amounts of money can be saved in adjusting accuracy to actual need.
Historically, maps had only one scale, and one related accuracy. That is the basis for the National Map Accuracy Standards. Still today, many people declare the accuracy of a database as being related to either as some nominal photo or map scale. In the digital realm, this doesn’t make sense anymore. Now, data are stored in real ground coordinates (scale 1:1) and displayed or worked on at any scale (using pan and zoom).
This leads us to the second topic on accuracy: That accuracy should be independent of any scales. By expressing an accuracy to be, say ±10 feet, one expresses an on-the-ground reality that most people understand intuitively. It then becomes reasonable to list the following for a database:
• City boundaries ±100’
• Street centerlines ±10’
• Sidewalks ±2’
• Hydrants ±0.01’
• Road centerlines (rural) ±20’
• Road centerlines (suburban) ±10’
• Road centerlines (city) ±5’
• Road centerlines (downtown core) ±1’
The principal impact of this approach is that the courts would stop seeing accuracy as related to map scale, which was a source of a lot of grief in the past, especially when “unscrupulous” map enlargements were used for certain projects. This would help protect the GIS practitioner who has no control at what scale a digital GIS product is going to be used.
This varying accuracy is made possible because most GIS products finally allow double precision data storage for coordinates.
Of course, this puts a greater burden on GIS database metadata. The above accuracy detail has to be inserted into database descriptions. Accuracy has to be covered for each layer/location for a complete description of a database’s “quality”. Each one of the above accuracies can be accompanied by additional information such as “as extracted photogrammetrically from 1:10,000 photography”, or “as digitized from ortho mosaic with a GSD of 1 foot and an accuracy of ±3 feet”. As a matter of fact, no metafile is to be deemed complete unless the major information sources are identified.
This method of handling of accuracies becomes more relevant at a time when GIS database are starting to hold more engineering and/or survey data. The original National Map Accuracy Standards were not designed to do that, and therefore some changes in this respect are warranted.