(Data Collection and Processing)
Data Collection Methods
To collect GPR reflections, paired antennas, located within a "box," are moved along the ground in transects (see figure below). One antenna generates the propagating radar waves and a second antenna records the reflection traces generated from below. When many hundreds or even thousands of reflection traces are stacked together, as they are collected along an antenna transect, a reflection profile is produced.
Dr. Conyers collecting GPR data at Petra, Jordan
Most systems can also be programmed to collect data with a survey wheel, or some similar device that can measure where the antennas are in distance along each transect, which can expedite data processing as all recorded reflection traces can be assigned a specific location within a grid:
Michael Grealy collecting GPR data at the CATS test site. Antennas can be attached to survey wheels, which can be programmed to collect a given number of reflection traces every programmed distance along a transect.
Usually antennas are placed directly on the ground surface or close to the ground within a fiberglass sled of some sort. If antennas are located too far above the ground, energy will not as effectively penetrate the ground as most will be reflected back to the receiving antenna from the ground surface.
GPR surveys are most often done by establishing a grid over the desired area. Rectangular or rectilinear grids are preferable to other grid designs for a number of important reasons. Digital reflection data from a rectangular grid can easily be exported to computer display and imaging processing programs that are pre-set for this gridding method. In this way the data can be quickly processed and interpreted without time consuming transect surveying and drafting. In addition, with a rectangular grid, important reflections in each profile can be immediately correlated to others and reflections can be "tied" to parallel or perpendicular transects throughout the grid. In all cases a sketch of the grid, with notes on the transect length, orientation and beginning and end locations should be noted. The antenna is pulled along transects within the grid. Transects are typically spaced about 25 to 100 cm apart, depending largely on the antenna frequency being used and the amount of coverage desired. Time and financial restraints are also common factors affecting collection prodecures, since smaller transect spacing will require more surveying time.
The most tedious, but also important, part of a survey is performed by the person pulling the antennas. This job is the most difficult during continuous data acquisition because the person pulling the sled must not only walk backward but must also make sure that the antennas are moving parallel to the designated transect line. Some people use a cart or other devices to move equipment across the ground.
If data are being acquired in continuous acquisition mode, where radar pulses are being generated at a programmed number per second, the antenna puller must also pay attention to when the antennas move past designated surface markers. At each pre-surveyed location a marker button must be pushed to place marks in the reflection records. When a survey wheel is used, or antennas are moved in steps, manual marks of this sort are not necessary and antenna pulling is an easier task. Another important aspect of moving the antennas along the ground is making sure that the antennas are in the same orientation and the same distance above the ground, or directly touching it at all times. Changes in antenna orientation with respect to the ground will potentially cause variations in the recorded reflections that can be confused with “real” changes in the ground. This phenomena is called antenna coupling loss.
Point-source Reflections and Hyperbolas
There can also be point-source reflections that are generated from one distinct point feature in the subsurface. The buried materials that generate these types of point-source reflections could be individual rocks, metal objects, pipes that are crossed at right angles, and a great variety of other smaller objects of this sort. They are visible in two-dimensional profiles as reflection hyperbolas.
Point source reflection hyperbolas, also termed diffractions, are generated because most GPR antennas produce a transmitted radar beam that propagates downward from the surface in a conical pattern, radiating outward as energy travels to depth. The pattern of energy dispersal will therefore spread out and be reflected from buried features that are often not located directly below the transmitting antenna.
The ability to resolve buried features is mostly a function of the wavelength of energy reaching them at the depth they are buried. A “rule of thumb” is that the minimum object size that can be resolved is about 75% of the downloaded wavelength reaching them. Downloading of radar energy always occurs as energy passes in the ground and decreases in frequency. For instance, a 400 MHz center frequency antenna will generate downloaded energy of about 300 MHz in the ground.
Features capable of being resolved include both:
1)Planar surfaces, which can be stratigraphic and soil horizons or large
flat archaeological features such as house floors. Elongated buried features
of this sort would usually have to be oriented perpendicular to direction
of antenna travel in order to be visible on GPR profiles, and would be
visible as distinct “point sources” with noticeable reflection
hyperbolas. 2)Point targets are features such as tunnels, voids, artifact
caches or any other non-planar object.
A common complication that affects resolution of reflections in the ground is background noise, which is almost always recorded during GPR surveys. Ground-penetrating radar antennas employs electromagnetic energy of frequencies that are similar to those used in television, FM radio and other radio communication bands, so there is almost always nearby noise generators of some kind.
summary the following steps must be considered prior to selecting an antenna
that will allow for the best subsurface resolution at any study site:
2. Define the depth of the prospective target features and their approximate dimensions and composition. Using estimates of RDP, the cone of transmission can be predicted and potential resolution of features of interest can be estimated from the footprint size using different frequency antennas. From this calculate whether energy can be transmitted to the depth necessary to resolve the features of interest with the antennas available.
3. Decide whether or not it is physically possible to use the selected antenna frequency at the site to be surveyed. Transportability to and from the site and deployment over and around obstacles and obstructions once surveying is begun must be accounted for.
4. If it is known that there is a substantial amount of radio interference present at a site, and if the source can be identified, then it may be appropriate to choose an alternate antenna frequency so as to minimize that influence. In general this is not a simple task because it is difficult to identify sources and the risk of compromising survey objectives exists if the wrong antenna is chosen for only this reason.
Unfortunately, it is often not known in advance what the target depth of archaeological features of interest is, their dimensions, or often the ground conditions. Most importantly, the ability to transmit radar energy to the depth necessary is often not known until one actually collects some reflection profiles. Often the best one can do prior to going to the field is to make some rough calculations from the best knowledge available, and take the antennas that will probably be necessary for the task. Antenna choice can therefore be a difficult decision. As a general rule, if the target features are within about one meter of the ground surface, antennas between 400 and 900 MHz will be adequate to transmit energy to that depth and resolve most features, and associated stratigraphy.