GLORI-B Processing
A New Processing Technique Developed to Remove Track Parallel
Artifacts in a GLORI-B Bathymetric Dataset
Preliminary Use of New GLORI-B & SeaBeam 2000 Bathymetry
Developed By: P. D. Johnson
& F. Martinez
Introduction
The GLORI-B system (Somers and Huggett, 1993), a modified configuration
of the original GLORIA system that is now capable of collecting swath bathymetry
as well as sidescan, was used to survey the region between the Easter and
Juan Fernandez microplates. However, this turned out to be a trial cruise
for the GLORI-B system and numerous problems existed in the swath bathymetry
collected.
Click on the image to see an expanded view.
Figure 1. Raw GLORI-B bathymetric data as received from
the Institute of Oceanographic Sciences in December, 1993. Notice the track
parallel artifacts (labeled in the lower left corner) and the discontinuous
nature of the swath, especially obvious in the upper left corner where
what should be a continuous high instead appears as a discontinuous ridge.
Also present at every 90 degree turn is an artificial "seamount"
and "crater".
The GLORI-B bathymetry contained very prominent track parallel artifacts
which were spaced across the entire swath (see Fig.1 & Fig. 2). The
artifacts which appear as 400 meter, or more, ridges running parallel to
shiptrack resisted initial attempts at removal by various forms of filtering
and gridding. Therefore we have developed an entirely new technique, done
entirely within the GMT system (Wessel and Smith, 1991), to remove any
trace of these ridges.
Figure 2. Cartoon illustrating the track parallel artifacts
present in the GLORI-B bathymetry. The arrow is the direction of travel
for the ship.
Processing
A key element in the success of processing the GLORI-B bathymetry is
that the SeaBeam 2000 system was operational the entire cruise . This meant
that we had an independently collected bathymetric dataset which could
be used to remove the GLORI-B artifacts. Unfortunately the SeaBeam 2000
swath (3.46 x water depth) is much narrower than the GLORI-B swath width
(~20 km) so the outer swath artifacts of the GLORI-B data could not be
removed this way.
Therefore, we developed a method which used the SeaBeam 2000 as a standard
on which we could check the GLORI-B bathymetry that was also capable of
not only checking the inner regions of the GLORI-B swath but the outer
reaches as well. The first step in this process involved creating a regional
geographically registered bathymetric grid from the SeaBeam 2000 data,
shiptrack bathymetry from other cruises, and digitized hand contoured SeaBeam
and SeaMARC II data. (NOTE: the next several plots have been rotated so
that the ship track azimuth of 70 degrees is now in an east-west orientation.)
The raw GLORI-B data in xyz triplets (as received from IOS) were median
filtered and had all values greater than 4500 meters and less than 100
meters depth clipped. This data was then transformed into a raw
GLORI-B GMT grid (also geographically coordinated) and was then subtracted
from the regional grid to create a difference grid which contained not
only the track parallel artifacts, but also some seafloor texture. To isolate
the long wavelength artifacts from the shorter wavelength texture we used
an asymmetric gaussian filter, that was 10.5 km long parallel to ship track
and 1.5 km long perpendicular to track. This filter was passed over the
difference grid, thereby isolating the artifacts (see Figure 3). The artifacts
were then subtracted from the original GLORI-B data producing a processed
grid .
Click on the image to see an expanded view.
Figure 3. Isolated track parallel artifacts from the
GLORI-B bathymetric data.
However, this processed grid still contained high amplitude noise concentrated
at the swath edges. To remove this noise a new
regional grid was created by combining the processed GLORI-B grid with
the SeaBeam 2000 and strongly filtering it with a very wide (90 km) gaussian
filter. A noise
grid was then generated by subtracting the regional grid from the processed
grid. The noise grid was then examined, both qualitatively and quantitatively,
and a series of clipping parameters were set. These parameters defined
any points which had values that deviated from an accepted medial range
as noise and then clipped
these points from the processed grid. The SeaBeam 2000 was than added
back on top of the clipped GLORI-B grid and a gaussian filter, 4.5 km wide,
was passed over it to produce the final
GLORI-B bathymetric grid (see Figure 4) which was then free of both
the track parallel artifacts and the high amplitude noise.
Click on the image to see an expanded view.
Figure 4. Color shaded relief map using the newly processed
GLORI-B bathymetry, SeaBeam 2000 data, hand contoured SeaMARC II data,
and shiptrack bathymetry. This image has the SeaBeam 2000 overlaid on to
the other data.
Processing Images (Described Above)
Papers (or Abstracts) on this Topic
[ Project
Summary | GLORI-B
Processing | Who's
Involved ]
[ Marine Geology &
Geophysics | Department
of Geology & Geophysics ]
