Kansas Geological Survey, Open File Report 96-49
The data were downloaded from the DT6000A tape unit and converted from RADAN format into 4-byte SEGY format before importation into Seismic UNIX (SU) on a workstation. Once within SU, the data were time- and distance-scaled by a factor of 1x106 for viewing and processing purposes. The 60 ns scan data had 512 samples per trace, a sample interval of 0.117 ns, and a lateral distance (cdp trace spacing) of approximately 3 cm per trace.
The GPR data were treated as stacked seismic reflection data within SU, opening up the possibility of post-stack digital processing. A generalized data processing flow for the GPR data is shown in Figure 3. Front-end mutes removed high-amplitude first arrivals, allowing trace balancing to enhance low-amplitude reflection information recorded later in the data. Coherent noise filtering removed most of the lateral system noise recorded in the regions below 40 ns in the data. Bandpass and frequency-wavenumber (f-k) filters reduced the lower frequency information and enhanced high-frequency reflections. Trace balancing via automatic gain controls (AGC) allowed some of the low-amplitude events to become more visible. The data were then elevation corrected to an arbitrary datum at the highest elevation along the line to prevent the loss of data and remove the longer period elevation static shifts that inhibit correlation with the outcrop. Therefore, the top of the data on the GPR sections roughly corresponds to the surface. Due to muting, only 50 ns of data are displayed (Figs. 5 and 6).
Figure 3--Generalized GPR data processing flow chart.