LuAnn Wandsnider and Christopher Dore
In a recent SAA Bulletin article [12(5):13,22], Jim Ebert outlined issues regarding the conversion of statewide archaeological site databases, maintained by state historic preservation officers (SHPOs), to data layers to be included in a Geographic Information System (GIS). He identified factors that affect this conversion process, including the need for compatible databases between states, the specialized needs of individual SHPOs, and the relative low cost and wide availability of generalized GIS technology.
At the University of Nebraska, with the help of the Nebraska State Historical Society (NSHS), we have recently completed the initial conversion of the Nebraska site database to a GIS data layer. This conversion has taken place without engaging the critical issues identified by Ebert. Our strategy has been simply to bring the site and survey databases into a low-cost, easy-to-use GIS. We and those at NSHS responsible for responding to questions about the database can now experiment with the kinds of management and research questions that were previously difficult or impossible to ask but that are easily posed and answered with GIS technology. For Nebraska, the issue of compatibility with other archaeological databases has yet to be broached.
Nebraska has little federal land and no state laws governing the inventory and protection of cultural resources. For this reason, as of November 1994, information for only 5,665 archaeological sites recorded as far back as 1930 had been reported to the NSHS, which serves as the clearinghouse for information on state cultural resources. Many of the recently located sites have been reported by archaeological surveys associated with federally funded ground disturbance, especially highway improvements. The GIS conversion was supported by a grant from the Nebraska Department of Roads as part of the IS-TEA program, which recognized the need for an archaeological data layer to enable short- and long-term planning for highway construction and maintenance.
Here we report some of our conversion experiences, hoping to assist others contemplating or engaged in a similar process. Our report has two parts. First, we deal with the practical aspects of the conversion process and the lessons we learned. Second, with a cultural resource data layer in hand, the critical question becomes, "what is its accuracy?" GIS- and CAD-assisted road design is a binary operation: either the site is or is not within the planned right-of-way. It is imperative, therefore, to assess the quality of the locational information contained in the archaeological database.
As part of the Nebraska Cultural Resources GIS Project, we created two data layers, one for archaeological sites and the other for survey areas. For both tasks, we selected TNTmips (developed by MicroImages of Lincoln, Neb.), a menu-driven, PC-based product that was easy to learn, flexible enough to meet our needs (vector and raster databases can both be accommodated), and of low cost ($5,000). It also has the ability to export the files in Arc/Info (a GIS standard) and other formats, so by building the GIS database with this software, we were not committed to the product should our needs expand beyond it.
Archaeological Site Layer
Creating the site GIS layer involved converting the extant state cultural
resource database to GIS, which was relatively painless. The dBASE III+ file,
containing records on 5,665 sites and maintained by the NSHS, was easily
imported. For all sites, a legal description was coded, but for this locational
information to be useful in the GIS, it had to be converted to the
latitude/longitude graticule using a United States Geological Survey (USGS)
program developed for Nebraska. For 1,371 sites, UTM zone and coordinates were
available, and from these, three different layers containing sites in the three
UTM zones that span Nebraska were created. From these UTM layers and the
latitude/longitude layer, we developed a single site layer in the State Plane
coordinate system tied to the North American 1983 Datum and employing the GRS80
As we undertook the conversion process, we became reacquainted with several critical geographic concepts: geoid, ellipsoid, datum, coordinate system, and projection. The geoid is the actual earth's surface measured at mean sea level; the ellipsoid is an elliptical approximation of the geoid that takes into account the flattening of the earth at the poles; a datum is the arbitrary origin of the graticule or grid coordinate system; a coordinate system is a numeric system for describing locations in space; and a projection is a particular way of viewing a coordinate system.
Why are these important? When we describe the locations of sites, we use either latitude and longitude or coordinates in a particular grid system, e.g., UTM and State Plane. Each descriptor has its advantages and disadvantages. On one hand, archaeologists are comfortable with the UTM coordinate system, and coordinates can be assigned with some accuracy at field locations using USGS maps. On the other hand, statewide archaeological databases span several UTM zones, creating data management problems. In addition, many other GIS layers available from other agencies are based in the latitude/longitude graticule. Interpolating longitude and latitude from a USGS map is difficult and error prone.
A wide variety of ellipsoids are in use. These, however, tend to be consistent for any given area of the globe, since an ellipsoid is chosen to provide a "best fit" for particular areas of the earth's surface (geoid).
Each coordinate system by zone has at least one datum and relies on a particular ellipsoid. Herein lies a problem for archaeologists: when we record the coordinates of a site from a paper map, we rarely record the datum or note the ellipsoid. Importantly, USGS has recently accepted a new model of the earth with projection parameters that use satellite and terrestrial data and that are therefore consistent over wider areas. Maps older than 1984 use the Clarke 1866 ellipsoid and the North American 1927 Datum; more recent maps use the GRS80 ellipsoid and North American 1983 Datum. The spatial difference for identical coordinates determined with different ellipsoid and datum parameters can be as much as 300 meters! In addition to recording UTM zone and coordinates, it is important to record the reference datum.
However, with GIS and Global Position System (GPS) receivers, coordinates can be converted to virtually any of the ellipsoids, coordinate systems, and datums. Importantly, all three of these must be specified to fix an accurate location. If a field worker is using a GPS receiver to relocate a previously recorded site, the worker must specify the appropriate parameters on the receiver. Recording the datum, ellipsoid, and coordinate system along with the site coordinates is essential both for projects using GPS equipment as well as for those working from paper maps, where this information is usually available on the map itself.
Survey Block Layer
A layer containing boundaries of 258 blocks surveyed as part of 87 projects was
created through digitization. The NSHS maintains a library of survey reports in
which the locations and boundaries of survey areas were reported for about half
of the survey projects done in Nebraska (mostly after 1990). While an
electronic report database is maintained by the SHPO, it contains no
information on survey unit boundaries. Rather, this information is usually
transferred to the appropriate county map at a scale of 1:126,720, which we
found too coarsely resolved for our purposes.
When creating the survey area boundary layer, we relied on the more primary source, which was usually the original report. We found that photocopies of USGS or other maps included distortions of up to 100 m. Therefore, we usually transferred the reported survey polygon to a USGS quad sheet and digitized it using the UTM ticks to georeference the digitization. If USGS maps with the survey areas were available, we digitized from these. On rare occasions, high-quality photocopies were included in reports, so we digitized from these, using prominent features common to the report map and a reference quad sheet to georeference the survey polygon.
In addition to the survey of extensive blocks, numerous surveys have been conducted along road rights-of-way. A letter report with a photocopy of the county map is on file at NSHS. We are presently experimenting with ways to incorporate this information into the survey block layer other than through digitization. Under consideration is the batch processing of road segments delimited by mile-markers to which a linear ROW buffers could then be applied. Alternatively, tracing appropriate road segments as they appear on the screen is also being considered.
The accuracy of this survey boundary layer is variable, reflecting the quality of the digitizing source. Furthermore, for early work associated with the River Basin Surveys, survey boundaries and parameters are often known only through oral tradition. In the GIS data table, we include a variable that codes the quality of the digitized maps.
After we created the cultural resource layers, the next major problem to address was the accuracy of the coordinate data. The errors introduced in recording site data, the scale of the original legal descriptions, and errors inherent in UTM coordinates can affect the accuracy of data entered into the GIS layer. To address this problem, we needed to evaluate the accuracy of locational information.
Part II: Site Locational Accuracy, will be presented in the November/December issue of the SAA Bulletin.
LuAnn Wandsnider is assistant professor of anthropology at the University of Nebraska-Lincoln. Christopher D. Dore is a principal in the firm Archaeological Mapping Specialists and is an adjunct assistant professor of anthropology at the University of Nebraska-Lincoln.