Spiro is one of the most important Mississippian sites in North America. Located in eastern Oklahoma, the site is characterized three types of mounds; one burial mound, two temple mounds, and nine house mounds; ceremonial plazas and supporting city environs.

This study focuses on Brown Mound, a temple mound which has been both looted and excavated in the past. A goal of the project was to generate a higher resolution contour map than the existing 25 cm contour interval map shown below. Additional objectives of the project included generation of a five cm resolution digital elevation model (DEM) and creation of a 3D model of the mound.

Spiro Site contour map showing location of 1979, 1980, 1981 excavation units. Adapted from Rogers, 1982.

The equipment used for this study is the Optech Intelligent Laser Ranging Imaging System (ILRIS) 3D Scanner, a laptop with the appropriate software and a Genie TZ-50 boom unit. The scanner has an accuracy of 3 mm in the x, y, and z directions, a range of 3-350 m and a data sampling rate of 2000 points per second. The Genie boom unit has a maximum vertical range of 55 feet and a maximum horizontal range of 30 feet.

Brown Mound is approximately five meters in height and is covered by grasses and sapling trees (below top). In early April tress were cleared and the site was mowed, but due to wet weather conditions fieldwork could not resume until after substantial regrowth had occurred. The existence of ground cover such as tall grasses, or visual obstacles such as trees, is a major issue in the production of accurate surface models at sites such as Spiro.

Since the survey area lacks identifiable ground control features needed for the alignment process (which will be discussed below), scanner targets were created.

Brown Mound, March 2004. Note paved path and interpretive sign at right.

Brown Mound after clearing and mowing – with substantial grass regrowth.	Placing target at survey scene.

These targets were then dispersed across the site to ensure the necessary ground control in the overlapping scans, required to merge individual scenes in the PolyWorks software. The targets were 16 x 16 inch plywood boards that had a black background and white designs on them (shown above right). This scheme was chosen to produce highly contrasting reflectivity (target vs. grass/bare dirt) making the targets easily visible in the scan.

To attain complete scan coverage of the mound, five boom locations were chosen. The average scan was taken approximately 45 m above the mound, and the scan times ranged between 15-30 minutes. Nine scans at a resolution of 2-3 cm were required to achieve complete overlapping coverage of the survey scene. The scan file sizes ranged from 20 to 30 MB. Field notes where taken for the azimuth (direction) of the scan and the tilt of the scanner head.

Final boom lift placement.

The scan files where then parsed and cleaned using Optech’s software.

Cleaning process: This image was taken looking south. Note the exclusion of the land outside the road boundary.

Note that the scanner creates a shadow within the image (center of the right image above). This is due to the laser returns from the mound itself which ‘block’ the surface behind. This situation is rectified by obtaining the scans from multiple locations around the mound, which when merged fill in the no-data holes. After the scans were cleaned, they were aligned and merged using the PolyWorks software (shown below).

Alignment Process: Scans were aligned first and then merged. Note the disappearance of the shadows. This is due to the multiple scans taken around the mound.

After the alignment and merge processes were completed, the data were reduced from 14 million points to seven million. This was done to ensure that the data would be manageable in other software programs. Because the scans were taken from the boom unit, the axes were tilted. To correct this problem, the merged product was rotated along the x-axis using the tilt angle (relative to the horizontal plane) of the scanner head.

Rotation Process: Rotating the data by a measured angle ensures proper gridding in the Surfer environment.

After the rotation was completed, the data was exported in an ASCII point cloud file. This file was imported into Golden Software’s Surfer 8 program, where a grid was created. Triangulation with linear interpolation method was used for the gridding process. It was used because it was the simplest process and most efficient, and took the least amount of time to run. Because of the vegetation, the results of the scan were coarse at best. An attempt to filter out some of the vegetation was made using a low pass moving average filter (Surfer’s Lowpass 3 utilizing a 3 x 3 neighborhood), removing high frequency noise to produce a smoother grid. A 0.5 m contour interval was then computed in Surfer. It should be clearly understood that the elevations shown in the figure below are relative to the ground plane chosen in the previous rotation step. Any deviations of this ground plane from the plane normal to the local vertical will result in incorrect contours. A 0.5 m contour interval was then run in Surfer.

Contours: Note that the contour lines still look course. The vegetation at the site was anywhere between ankle and knee length.

The next step was to import the grid created in Surfer into the Idrisi Kilimanjaro. A number of Idrisi grid modules produce output that can be of help to archaeologists and others interested in these types of studies. Another contour map was created, this time with a 15 cm interval. Slope and aspect maps were also created within Idrisi.

Contour: Contours are at 15 cm intervals. There is still some noise visible due to vegetation.

Finally, a basic flyover of the data was created in the PolyWorks software. The flyover provides an aerial view of the entirety of Brown Mound and allows the user to visualize the major topographic changes within the site.

- Spiro Animation –

Overall, it took approximately 25 hours to complete the project from start to finish. This can be broken down into:
6-7 hours to set up and obtain the scans
4-5 hours to clean and align scans
4 hours to create and reduce polygonal mesh
2-4 hours to grid data in Surfer (depending on interpolation algorithm)
2-4 hours to create grid and other derived surfaces in IDRISI (slope etc…)

Conclusion

We believe that the applications of the scanner/boom combination can be of great help to people that are interested in a multitude of studies. These include, but are not limited to, archaeology (predictive modeling, historic preservation), geology (weathering studies), and architecture and engineering (identification of structural anomalies, change detection and monitoring). However, several questions remain and require further research. First, the periodic movements of the boom mounted scanner due to wind, for example, need to be somehow minimized. Possibilities include securing the platform with tension lines and monitoring movement during the scan so that it might be removed in postprocessing. Second, the effects of vegetation need to be investigated further with an eye toward minimizing its affect on the resulting surface. Third, a better method and procedure for “leveling” the scans is need to ensure that contours, i.e. lines of constant elevation, are relative to the plane normal to the local vertical. One promising method is to use geodetic grade GPS technology to observe the 3D coordinates of each scan center and rotate the aligned scans using the points. This would replace the rotation step explained above.