In search of a Historic Grave: GPR Investigation near the Yellowstone Lake Store: 7/15/2010 Steven Sheriff Professor of Geophysics Department of Geosciences University of Montana Missoula, Montana Introduction On July 15, 2010 we collected 61 ground penetrating radar (GPR) transects using a MALA Ramac GPR system mounted on a cart with a 250 MHz shield antenna. We chose this frequency because our previous investigations with a 500 MHz antenna beneath nearby roads contained significant noise from construction disruption. The target of this investigation is a historic grave known to be beneath the pavement. We acquired data for each transect in the southeast to northwest direction. We started each transect on a tape measure laid out along the road edge nearest to the lake shore using a transect separation of 0.5 meters (figure 1) and a trace interval of 0.05 meters. For each transect we placed the trailing edge of the antenna where the asphalt drops steeply towards the lake. We extended some transects well beyond the northwest edge of the road but ultimately clipped all to the road width for 3D analysis and presentation. Using the monitoring well as a datum (Figure 1), the origin of the data volume is 4.6 meters from the center of the well at an azimuth of 324 o from magnetic north. The northwest end of the first line is 10.6 meters from the well at an azimuth of 320 o from magnetic north. Figure 1. The left image shows the GPR cart, location tapes, and location of grid relative to the gas station. The right image shows the well used as a datum; white dots (pancake mix) mark the beginning and end of GPR transect #1. Figure 2 shows the extent of the GPR survey along the road. In subsequent figures, the origin of the grid is at the left dot in Figure 2 with X increasing across the road and Y increasing to the southwest along the road.
Figure 2. White dots mark the beginning and end of GPR line #1 as discussed in the text. The left hand tape (southeast) marks the placement of the trailing edge of the GPR antenna. We acquired data for each transect from left to right in this image with successive transects at 0.5 meter spacing to the southwest (away from the camera). In subsequent figures, the origin of the grid is at the left dot in this figure with X increasing across the road and Y increasing to the southwest along the road. GPR Time Slices I subjected the data to standard processing techniques to reduce electronic and acquisition noises as well as to enhance visualization. The processing steps en route to interpretation include: dewow (e.g. DC correction) to remove instrumental drift, gain adjustment proportional to inverse energy decay to account for amplitude decrease with increasing geometric spread of the downward-propagating radar signal, bandpass filtering to remove noise outside the dominant frequency of the antenna and recovered signal, smoothing the images using running averages over several traces and lines to further reduce visual noise assembly of the 61 individual GPR transects into a 3D data volume. Time slices (sometimes called amplitude maps) are horizontal sections through the data volume. Thus, they represent map views of the reflection amplitude recorded by the radar waves as they reflect off subsurface changes in material properties. When observing the data in this manner, one looks for patterns of reflection amplitude that are not characteristic of the fluvial and/or dune features of the near shore sediments in the area. This visualization technique often makes features visible that would otherwise be missed during the inspection of successive individual transects. In these data several interesting features show up in the time slices. Figure 3 presents a time slice at 11.44 nanoseconds (ns) or a depth of about 0.60 meters. The fabric highlighted by magenta dashed
lines in the lower image of Figure 3 is oblique to the current roadway s pavement and probably shows the path of its predecessor. The northwestern low amplitude feature shows significant disruption throughout the radar volume and probably marks an old wide trench. The southeastern magenta line marks a more distinct trench of some sort. These reflections propagate deep into the data volume (Figure 4) and help obscure other features. Figure 3. Time slice of the radar volume at a depth of about 0.60 meters; the lower edge of the images follows the southeastern edge of the pavement. The high reflection amplitudes between the linear lows (marked by magenta dashed lines in the lower image) probably mark an older path of the existing road. The northwestern low amplitude feature shows disruption of sedimentary features throughout the section; this seems to be an old trench. The southeastern magenta line marks a more distinct trench of some sort.
Figure 4. Fence diagram of five GPR transects and one time slice (horizontal) showing the disruption of reflectors which yields the interpreted trench in Figure 3. Deeper in the data volume an interesting feature begins to show up as a rectilinear outline of high amplitude reflections among the data at a depth of about one meter. The subsequent low amplitude part of the waveforms at about 1.25 meters deep shows the rectilinear feature well (figure 5). This is the only persistent rectilinear feature in the data set and is about the correct scale for a grave. It extends from about one meter in depth to nearly two meters. The corners, clockwise from the lower left, are at: X Y 1.5 8.0 3.5 8.0 3.5 7.0 1.5 7.0 Thus, the feature is nominally 1x2 meters in plan view. Complicating the issue is that it does not face west as one would expect from historic grave considerations and the northwest and southeast edges are on acquisition lines. The latter detail could simply be a result of the interpolation algorithm and the buried source could be displaced to either side of the lines by a bit.
0 2 4 6 11000 9000 7000 5000 3000 30 25 20 15 10 5 0 1000 11000 0 2 4 6 9000 7000 5000 3000 1000 30 25 20 15 10 5 0 Figure 5. Time slice from about 1.25 meters deep. Upper and lower images are equivalent with the lower showing a dashed white box around the rectilinear feature which is interpreted to be the most likely location of the known grave in these data. Summary The dashed white box in Figure 5 outlines the radar feature in this data volume which is the best bet for being the historic grave. The other features in the data volume appear to be from a previous road, construction (perhaps related to that or the current road), or sedimentary features. A good follow up to this will be a concentrated grid of 500 MHz data with 0.25 meter line spacing between the five and ten meter marks on the northeast-southwest axis. Of course, excavation would be a good test of the hypothesis.