Noise
Suppression Using Wave Field Decomposition Method
One of the
key tasks of office processing of GPR data using traditional methods, without
using BSEF analysis, is suppressing reflections from objects located on the
ground surface. Such air reflections have a high amplitude level. This leads to
masking of useful signals from subsurface structures.
Antenna
shielding of the GPR does not provide complete removal of air reflections.
These types of noise are most pronounced on radargrams acquired using dipole
low-frequency antennas whose design does not include shielding. Additional
noise sources that interfere with extracting useful deep reflections are
diffracted reflections from high-contrast local objects in the near-surface
layer.
To remove
noise, including air reflections, the wave field decomposition method is
implemented in the software. The idea of the method is to decompose the GPR
profile into independent components corresponding to different types of wave
objects. After excluding components that contain noise, the profile is
reconstructed from the remaining components. As a result, the excluded noise
waves are absent on the reconstructed profile.
An
example of removing wave components formed by air reflections is shown below.
On the left, the original profile recorded with a center frequency of 150 MHz
is shown. On the right, the result of applying the decomposition method is
shown.
To study local heterogeneities, components with
diffracted waves are preserved. Components of reflections from predominantly
horizontal medium boundaries are suppressed. Below, on the left, the original
data are shown. On the right, the result of extracting diffracted waves is
shown.
Noise
Suppression Using Spatial Filtering
Along with
the wave field decomposition method, GEORADAR-EXPERT implements noise removal
using spatial filtering. Filtering is performed by analyzing the
spatial-temporal parameters of diffracted reflections and extracting waves with
a certain apparent velocity and propagation direction. The signals of the
extracted waves are then subtracted from the signals of the GPR profile.
Spatial
filtering effectively removes noise in the form of air waves and intense
diffracted reflections from local objects. At the same time, reflections from
medium boundaries are preserved. The algorithm allows flexible adjustment of
filtering parameters according to the features of the wave pattern of each
specific profile.
An
example of using spatial filtering to remove numerous reflections from
submerged logs and boulders located on the riverbed surface and in silt
deposits is shown below. On the left, the original profile is shown. On the
right, the result of spatial filtering is shown.
Next, a GPR profile crossing tram tracks is considered. On the left, the original profile is shown, where intense reflections from the metal rails and tram track infrastructure objects are superimposed on relatively weak reflections from subsurface layer boundaries. On the right, the result of spatial filtering is shown, where noise is suppressed and does not mask reflections from the boundaries.
Each of the methods for improving the resolution of GPR data implemented in GEORADAR-EXPERT has its own advantages. Depending on the characteristics of the wave pattern of the GPR profile and the objectives of the survey, the user can choose which of these methods to use in each specific case.