TerraSearch Geophysical, in partnership with the New Storrs Cemetery, in Mansfield, Connecticut, has been conducting a systematic test of different ground penetrating radar (GPR) survey methods of cemeteries. The New Storrs Cemetery is a historical burying ground, established in 1864 by a grant from Charles Storrs (a co-founder of the University of Connecticut with his brother Augustus) to the Mansfield Burying Association. The cemetery has been in continual use since it's inception, and documents burials from the mid 19th century to the present. A selection of the headstones within this section of the cemetery are shown below in a three dimensional model we created during the survey.
For our systematic survey, we choose an area with 22 marked headstones as well as six footstones. The grid survey area measured 15 by 20 meters, and was collected using several different methods. The first method was uni-directional collection, at 25 cm (9.8 in) intervals, with reverse transects collected whenever an obstruction was encountered. The results of this survey method are shown below at approximately 1 meter in depth, with highly reflective areas indicating likely burial features (grave shafts and coffin reflections). Other interment reflections were also clearly visible, and rectangular in plan view, at shallower and deeper depths. In the lower left corner of the map, we've included the individual transect map of this grid, the only data gaps in this method of collection are from obstacles, such as headstones.
The second survey method included data collection at 25 cm intervals, but collected from both baselines, in a bi-directional manner. No reverse or continuation transects are included in this grid. The results of this method are shown below at approximately 1 meter in depth, and while highly reflective features indicating burial features are again visible, they are much less clear. There is also clearly a discontinuity in the location of these features when shown in plan view, based on the disconnect between starting the survey on opposite baselines. While this data collection method does not impact the two dimensional profiles of interments, it has a profound effect on the quality of the plan view maps. Again, we have also included the transect map in the lower left, and while the coverage using 25 cm bi-directional is still most of the grid, there are larger data gaps. These gaps are not what is driving the disconnect in the highly reflective features shown in the plan view map. This disconnect is driven by three main factors: 1.) The variations in the topography of the grid (which was overall very flat and only slightly sloping); 2.) The highly accurate encoder wheel on the GPR machine, which accurately tracks the movement of the GPR; and 3.) the method of interpolation, which is reading the position of buried features (based on return time in nanoseconds from the GPR antenna). By introducing a second baseline to collect GPR data, we have introduced an unnecessary error for the processing software in estimating the antenna's position as it encounters buried features. These differences in interpolation can of course be smoothed out using a more advanced method of interpolation, such as Krigging, but ultimately the data will not be as accurate and reliable as the uni-directional method of collection.
We plan to explore many other data collection methods with this grid, including 25 cm interval uni-directional survey with no reverses, 25 cm interval bi-directional survey with reverse and continuation transects, 50 cm (19.7 in) interval uni- bi-directional survey with and without reverse transects, and 1 m (3.28 ft) interval uni- and bi-directional survey with and without reverse transects.
The preliminary results of the survey indicate that 25 cm uni-directional GPR survey with reverse transects is a far superior method of GPR data collection; this is always our preferred method of data collection in cemeteries, particularly historical burying grounds where grave shaft and coffin reflection features may appear weaker than recent interments. Cole Peterson (pictured above) will be presenting the results of this survey and research at the upcoming Easter States Archaeology Federation meeting in Ocean City, Maryland this October.