Geology 5660/6660 Applied Geophysics 28 Feb 2014 © A.R. Lowry 2014 Last Time: Ground Penetrating Radar (GPR) Radar = electromagnetic radiation (light) in the MHz (radio) frequency band  Governed by wave equation (  similar to seismic!)  Source & receiver are dipole antennae  Signal is a single pulse  Processing & display analogous to seismic section  High frequency  high resolution but also high attenuation  Images changes in electromagnetic impedance Z For Mon 3 Mar: Burger (§ )

Last Time: Ground Penetrating Radar (GPR) Velocity (usually) is not estimated; emphasis is mostly on the the imaging of structure rather than physical properties. Instead TWTT  depth is approximated from rough ~ V Radar reflections image variations in Dielectric constant  r ( = relative permittivity )  3-40 for most Earth materials;  higher when H 2 O &/or clay present Geology 5660/6660 Applied Geophysics 28 Feb 2014 © A.R. Lowry 2014 For Mon 3 Mar: Burger (§ )

Applied Geophysics “In the News”: Texas A&M researchers use GPR to image Civil War era fortress structure under Alcatraz… On the BBC.

For most applications (i.e., near-surface)  1 ≈  2 ≈ 1 ;  (10 -4 –10 -1 ) «  (10 6 –10 10 !), and hence (i.e., we are imaging velocity variations corresponding to changes in dielectric constant!) For the water table, R ~ 0.1 Recall seismic waves attenuate as where Q is quality factor; Radar waves attenuate similarly as ; where Attenuation is extremely high for shale, silt, clay, and briny water (which is why GPR rarely penetrates > 10 m!). 

Skin depth, or depth of penetration, is ~ 1/ . Hence main applications are in archaeology, environmental, engineering site investigation… Also used for cavity detection and other very near-surface applications GPR freqs

Frequency-dependence of the attenuation results in dispersion : High frequencies attenuate more rapidly; pulse appears to “broaden” and the phase is delayed: This has “appearance” of a lower velocity medium. GPR freqs

(From a very old cemetery in Alabama…) “Black-box” processing is simplistic so see some of the same features observed in low-level (brute stack) seismic processing:

Assuming a constant velocity can introduce a factor of 2 to 3 scale error in converting velocity to depth! ( But one could reduce velocity scaling error if  were calculated from, e.g., travel-time amplitude decay)…

V1V1 Alternatively can use moveout on Diffractions : h1h1 h2h2 x The equations are the same as they were for seismic, but since GPR is (nearly) always zero offset, x s = x g ! rsrs xgxg

Note some data processing steps are similar to seismic but lack some tools (such as refraction velocity analysis). Commonly do static corrections for elevation, filtering, automatic gain control; much less common to migrate.