TIPS: Enhancing GPR Targets by Filtering
close X
Nav Menu
 

TIPS: Enhancing GPR Targets by Filtering

TIPS: Enhancing GPR Targets by Filtering

–>

T

he menu item “Filter” is short for “Background Subtraction Filter”. It is applied to remove flat-lying signals (signals that arrive at the same travel time or depth) that are present in ALL traces in the GPR line; typically, the direct air and ground arrivals at the top of the section (Figure 1) or system noise bands. These undesired signals can interfere with desired signals such as hyperbolas from subsurface objects.

The most common background subtraction filter is to subtract the average of all the traces. While effective, some flat-lying features do not extend over the whole survey line and therefore are not removed. Reducing the width of the filter from an average of all traces to a shorter length of traces is very effective for removing more localized flat reflectors. These are typically real reflections from flat-lying objects and boundaries in the subsurface which can mask the response of more localized targets.

gpr background noise reduction filter
Figure 1
All CONQUEST®100, NOGGIN®, pulseEKKO®, LMX100™,
LMX200™ and FINDAR® GPR systems have the Filter button available on the screen during data collection and review in the field.
gpr background noise frequency
Figure 2
Level 1 of the Filter tends to remove flat-lying signals present in ALL GPR traces; such as the direct arrivals from the GPR transmitter to the GPR receiver visible at the top of all GPR lines (circle above). Compare Figure 1 (no filter) and Figure 2 (filter set to Level 1)

 

Why would I want the filter to do this? Most GPR data targets that people are interested in are point targets that create hyperbolic responses in the GPR data, either true point objects that archaeologists or forensic investigators are interested in, or the hyperbolas caused by crossing perpendicular to the alignment of linear objects, such as buried pipes and cables.

pulseekko gpr daragram 2
Figure 3
Level 3 of the Filter removes moderately long flat-lying signals in the data. In this example, notice how the 2-meter- long flat reflector, from position 3 to 5 meters in Figure 2, is now removed by the shorter filter width.
pulseekko gpr hyper stacking
Figure 4
With the Filter set to Level 5, all flat reflectors have been removed from the GPR line, revealing very weak hyperbolas that may be targets of interest.

 

Changing the width of the background subtraction filter removes flat reflectors of shorter width so that hyperbolas from targets are enhanced and more visible in the data.

You can quickly toggle through filter levels 1 to 5 to find the best filter setting for your data. The lower the number, the longer the filter width, and the more “relaxed” the filter; only longer flat-lying features get removed (Figure 2). The higher the filter number, the shorter the filter width, and the more “aggressive” the filter. This results in the removal of long and short flat-lying features. Figures 2, 3 and 4 show a gradual change of the filter width and the effect this has on data. Pressing OFF turns the filter off completely (Figure 1).

Figure 5
The Background Subtraction Filter button is accessible from the toolbar of the LineView software module.

 
The Filter button is present on Sensors & Software GPR systems (Figure 1) as well as the LineView module of the EKKO_ Project™ GPR Analysis PC software (Figure 5).

In the LineView module of the EKKO_ProjectTM software, the filter width is changed in the View > Settings > Gain/Filter > Filter Width menu option. In EKKO_Project™, the filter allows you to set the physical length of the background subtraction filter for finer tuning during post-processing.

Remember that applying the Filter to your data does not affect the saved raw data; the filter only modifies the displayed image, so there is no harm in trying it to see if it helps your data interpretation.

With GPR lines that have many reflections from boundaries masking hyperbolas, shortening the background subtraction filter width may be the critical step in identifying the target.

LinkedInFacebookTwitterEmail