Planning GIS Database Accuracies

Agencies must evaluate their most important needs to select the right GIS accuracy

by / August 31, 1995
Sept 95 Level of Govt: State, local Function: GIS Problem/situation: Agencies have a wide range of database applications, each requiring different levels of accuracy

Solution: Agencies need to decide their most critical needs and design accuracy accordingly

Jurisdiction: Texas; San Diego, Calif

Vendors: none Contact: Roddy Seekins, Texas Department of Information Resources

512/463-6581 or Bill McGarigle Contributing Writer Several administrative, technical and political factors must be considered when deciding how much accuracy to have in a proposed GIS database. From an administrative point of view, the major considerations are the primary application and the price tag of the database - two closely related factors, given that cost rises almost exponentially as the accuracy of the database increases

Because electronic presentation accuracy definitions are different from paper products, it is important to clarify those distinctions. The accuracy of a hard copy map is defined by its scale. A Quarter Quad map on Mylar, for example, which is one-fourth the area of a standard USGS map (or quad, for quadrangle) has a 1:12,000 scale, which means one inch represents 12,000 inches, or 1,000 feet, on the ground. The electronic, or digital, version of that map retains its 1:12,000 scale, but since it is now a raster (video screen) product, its accuracy is defined in terms of resolution and horizontal distance (spatial data), rather than "scale." Even though the scale can be changed electronically with a flick of a mouse, the digital, ortho quarter-quad (DOQQ) has a resolution of one meter, with a spatial accuracy of plus or minus 10 meters - which means objects having a width of one meter or larger will be displayed with a positional accuracy of plus or minus 10 meters. Linear features smaller than one meter in width, but having one meter in length, may also be discernible. We will look at the importance of DOQQs for GIS database accuracy later in this piece

APPLICATIONS Agencies have a wide range of database applications, each requiring different levels of accuracy. Engineering needs the highest levels, perhaps quarter-meter resolution or even better. An environmental protection unit using GPS to locate rare and endangered species habitats require point accuracy to enable biologists to return to a specific breeding location. If regional planning groups are studying intelligent vehicle highway systems, they will need resolutions of three feet or better. Less accuracy may show the vehicle as being completely off the road. On the other hand, planning applications involving spatial relationships between features require much broader measurements. A comparison of different corridors for routing a new road or pipeline, for example, may call for a resolution of plus or minus 50 feet

Some applications require both fine and coarse resolution. As Roddy Seekins, statewide GIS Planning Analyst for the Texas Department of Information Resources, pointed out, "there is a trade-off for having too high a scale [resolution]. The larger your scale is, the more data you have to deal with, which is why in some cases an initial analysis is done using LANDSAT imagery with 30-meter resolution. Analysts will identify a problem area in vegetation, then step up to one-meter resolution." ACCURACY DICTATES BASEMAP The accuracy required for a particular GIS application usually dictates the original graphic, or basemap, from which the digitized version is made. A basemap can be simply a parcel outline or lot lines; it can be a Census TIGER [topologically-integrated, geographically-enhanced reference data] map, or a USGS 1:24,000 scale map. It can also be satellite imagery, black and white or false-color, infrared (CIR) aerial photography, from which digital, ortho quarter-quads (DOQQ) are made

A DOQQ is a highly-accurate digital version of the original imagery. It is produced from stereo pairs of aerial photos, either black-and-white or CIR

A high-resolution scanner translates the photographic image into electrical impulses, removes the distortions inherent in aerial photography, and produces a digital elevation model. The model is then translated into a language recognized by the GIS and loaded into the computer. A DOQQ enables users to measure distances and elevations - just as they would on a paper map - with accuracies of 10 meters or better

The aerial photography for DOQQs is provided by the National Aerial Photographic Program (NAPP), which is funded partially by a consortium of federal agencies and partially by the state to be photographed. NAPP's plan is to photograph each state once every five years. Without NAPP, the cost of acquiring precision databases would probably be out of the question for individual state or local agencies

COST "It's very expensive to create databases with a resolution of plus or minus one foot, as compared to plus or minus 50 feet," said Bob Parrott, research director for the San Diego Council of Governments, or SANDAG. "What I've seen happen is the engineering department wants plus or minus one foot, but planning doesn't need that type of accuracy - so the organization takes years to acquire a database because it is holding out for plus or minus one foot. [In the] meantime, there is plenty of need for the more general database." Parrott added that there are also political implications to consider: "You want to build into your database development actual applications and uses of that data. If it's going to take you five years to come up with the use, you may lose your following. So [the political factor] needs to be considered in terms of how you approach developing the data you need. You may not be able to develop the ultimate database right away." ALTERNATIVE FINANCING According to SANDAG's Parrott, many organizations are finding a way around the high cost of precision GIS databases through alternative financing methods. "Today we're seeing more data-development partnerships between local governments, local and state, local and federal - to share in the cost of developing data. Maybe the cost of data with plus or minus one-foot accuracy is out of reach for one agency," he said. "Through some sort of partnership, it could become a reality." MIXING ACCURACIES A strong argument for replacing older databases is the confusion resulting from loading differentially-corrected GPS data into it. "The problem," explained GIS lab manager for Texas State Parks and Wildlife, Dr. Kim Ludeke, "is that today's technology provides far greater accuracy than that of many older basemaps." For example, Ludeke has 1:24,000 USGS digital data in his computer showing the location of a highway. "Ninety percent of all the well-defined points on that map have an absolute accuracy of plus or minus 40 feet. If I use GPS with differential correction to record the location of an endangered cactus next to that highway, then put that data into the GIS, the cactus may appear on the wrong side of the road because I have more absolute accuracy on the location of the cactus than I do on the location of the highway

"If you display such a map in a presentation, or for publication, without displacing your GPS location in relation to the basemap, you're going to confuse people," he said

Ludeke added that Texas is trying to do a project in cooperation with the USGS to get DOQQs for the whole state. "The DOQQ will give us more credibility for our product. It is a more accurate basemap than either the USGS 1:24,000 scale digital line graph, or the census TIGER

"The idea would be to use the DOQQ as a photo map to update our data sources," Ludeke said. "It would enable us to realign roads, add roads, create a map of the state that has temporal as well as spatial accuracy

DOQQs are the best option we see right now."