Ground penetrating radar (GPR) is used to map hidden objects non-destructively. Recently, GPR is being used to look for uncounted bodies buried following  the Tulsa Race Massacre.

According to Wikipedia, GPR is kind of an old technology.  It was first demonstrated in the 1920s, including measurement of the thickness of a glacier.  It became a commercial product in the 1970s.

Typical frequencies are 10s of MHz to a few GHz. The "Wallabot" is a home use "GPR" (more like a "WPR" where W is for wall) that you can use to find studs, pipes and wires behind a wall so you don't drill into them. You can buy one for maybe \$60 and control it from your phone through a USB port.

Modern GPR equipment looks a little like a lawn mower with a display. As you roll along, data are displayed in an X-Y plot, where X is distance and Y is depth.  The natural units for "depth" are in time for the signal to return from a reflector (typically nanoseconds).  For a uniform material of dielctric constant DK, depth in feet can be computed as approximately

Depth=Nanoseconds/(2*SQRT(DK))

This follows the rules of thumb that distance in free space is nearly one foot per nanosecond, and the signal is delayed by twice the distance (out and back).  And of course, propagation of EM signals are reduced by 1/SQRT(DK).

One big unknown is the dielectric of the stuff you are penetrating. You can make an educated guess from tables (pick a number between 4 and 80...), or you can measure the return from an object of known depth, or you can send some sample material to a lab for analysis. In all cases this is not an exact science. The error in knowing the exact DK is reduced by square-root, so if you messed up the physical constant by a factor of two your measurement will be within 71% to 141% of the truth.

Reflectors can be metallic, or dielectric. In the case of dielectric reflections, what you see are interfaces between materials that have different dielectric constants.

GPR is similar to synthetic aperture radar, where airborne or spaceborne sensors map ground features using physical motion of the sensor platform in one axis.

There is an art to reading GPR plots, much like there is an art doctors use to read X-rays and sonograms.  You need to know what to look for: typically, objects will display as a hyperbola; look at where it is focused and ignore the tails.  Also, there is a ton of noise on the data.  Being an engineer, you know that reducing the noise can reduce the accuracy of the what you are looking for, so experts tend to look at the noisy data and only clean it up for the boss or for marketing (see smoothing is cheating). Below, Robert Freeland from the University of Tennessee gives us the basics of GPR and describes how to interpret the data on the resulting radargrams.

Robert Freeland from the University of Tennessee gives us the basics of GPR

Author : Unknown Editor