Knudsen 320 200 kHz keel-mounted sidescan trials
Results from 2000/2001/2002 field operations

John E. Hughes Clarke
© Ocean Mapping Group, Dept. Geodesy and Geomatics Engineering
University of New Brunswick, CANADA
506-453-4568 (ph.), 506-453-4943 (fax.), jhc@omg.unb.ca

Mike Crutchlow and Bruce MacGowan
CHS -Central and Arctic and Atlantic
 October 2002



CONTENTS
  Overview

Since 1996, the Canadian Hydrographic Service has been slowly replacing inshore single-beam hydrographic survey launches with launches equipped with multibeam echo sounders. At this time (November 2002) there are 5 Simrad EM3000S multibeam sonar deployed across the country (Atlantic 1, Quebec 1, Central and Arctic 2 and Pacific 1). These are deployed on a range of platforms. Originally all systems were installed on 31ft P-class vessels. At this time, only two remain in this configuration. The  other three have been moved to larger 40ft and smaller ~24 ft platforms. In all cases, these systems are primarily used for work in depths less than 50m.

Initially it was felt that these multibeam systems would completely replace single beam hydrography for most applications. It quickly became apparent however, that, whilst they could provide near 100% coverage for the first time, the cost effectiveness was limited whenever depths were much less than 15m.  For the systems in use by the CHS (Simrad EM3000S's), it is operationally impractical to acquire multibeam data with sectors greater than ~ +/- 60 degrees  for three reasons:

As a result, swath widths of ~ 3.5 x water depth are seen here as the absolute maximum realistically possible. This is exasperated by the CHS's conservative habit of routinely acquiring ~ 200% coverage (to guarantee sounding density) with a resulting line spacing of only 1.75x the sonar altitude (depth -draft). As a result in depths less than 20m, line spacing is always less than 35m (7m in 5m of water! (1m draft)). Older CHS standing orders allowed single beam line spacing to be controlled by chart scale with typically no tighter than ~ 50m spacing required for most coastal environments (without sidescan coverage as it is not used by CHS).  Furthermore, the required MB line spacing is  significantly smaller than the typical half swath width of a conventional sidescan sonar (most agencies that use sidescans operate at between 75 or 150m per side with  200% coverage).Thus under these shallow water conditions it is actually taking more line miles to achieve the coverage requirement. Arguably the multibeam can be acquired at higher speeds (although focussed sidescans are changing that).

Thus to date, extreme shallow water survey is taking longer using the requirement of 100% multibeam coverage over the old single beam model. Thus the CHS has been considering the use of sidescan instrumentation. But a number of concerns exist:

And, more generally, the CHS has always been skeptical about the use of sidescan as the result requires subjective interpretation.

To get away from the concerns listed above, if the sidescan can be rigidly mounted to the survey vessel, using in part existing hardware, then those concerns can be minimised. Keel-mounted sidescans are an ancient concept. Indeed the earliest sidescans systems developed in the 1960's originally were all rigidly mounted. Most notably, NOAA currently mounts a Klein 5500 sidescan on the base of one of their survey launches (although this obviously does not alleviate the cost concerns!).

Herein we examine the use of keel mounted sidescan systems derived from a Knudsen 320 200 kHz echosounder as a viable means of providing a proxy for the coverage requirement in depths less than ~ 20m.  The CHS have operated Knudsen digital echosounders in all regions for over a decade now. They thus already have the required topside hardware for digital acquisition of dual channel 200 kHz backscatter intensity information. In 1999  Knudsen Engineering Limited (KEL) first offered the use of an Airmar 200 kHz 0.5 x 50 degree sidescan stave as an option. This actually followed on from similar offerings by Odom and others in the same market (although at that time the other offerings were all analog logging only). 

A review of the developments in software at UNB and hardware at KEL over the past few years, as well as field deployment issues are herein described.



Y2000 operations

The Ocean Mapping Group (OMG) took delivery of one of the first 200 kHz sidescan staves offered by KEL in September 1999. At that time, with only a single stave and a single available channel (one of a 200/28kHz 320BP unit owned by the OMG), options were limited.

original Mary-O mount

The single sidescan stave was originally mounted on a swinging pole on the side of the vessel RV Mary-O (figure to left).  Using a switch box, the 200 kHz chanel could be transfered from the downward looking 200k Hz beam (6 degree) to the sideward looking 200kHz sidescan stave. The operator had to note when the switch happened as no digital record of which transducer was used, was retained.

Dynamic Range Considerations.

At this time, the 320 software was designed solely for conventional vertical incidence echo sounding and the binary trace data was designed purely for replay on a grey scale recorder. 1600 samples of linear received intensity were logged at only 8 bit resolution (even though the internal digitisers actually generate data at 14 bit dynamic range).  A 40 logR TVG was already available which was much more suited to incoherent rather than coherent reflections.

However, because of the strongly +vely biased distribution of scatterers (bright coherent reflections ("glints") being 20-30 dB above typical scattering levels), in order not to saturate the data in the mid range where the beam boresite touched down, the data in the far range quickly dropped down in to the lowest DN values (0,1,2 etc..). As a result, data beyond 50m slant range could not usefully be gained up in post-processing to retain any useful signature (unless the data in the mid-range saturated the digitisers). Thus the limited dynamic range effectively controlled the maximum usable swath width.

In part to alleviate the dynamic range and strong +ve bias of the intensity data, the 8bit data were logarithmically compressed to better illustrate the available data content.
 

Imaging Geometry Considerations

Based on the 50m maximum achievable swath, the usable line spacing was defined. The other critical factor was that with a single stave we could only image in one direction (starboard in this case). As a result, if a line spacing of 40m was used, alternate line directions meant that alternate survey corridors  were either 200% covered or not covered at all. In order to get a complete image, line spacing of as little at 20m was required.
Nevetheless, in waterdepths less than 10m this still provided a wider effective swath that the EM3000S as operated to CHS standards (200%coverage).

Hydrocamp 2000 results -the Kennebecasis Estuary

The first operational deployment of this system was part of the SE4083 undergraduate field camp in May 2000. As noted above, a 20m line spacing was used with alternate lines shot using the 200 kHz sidescan at 50m slant range. The in-between lines were run using the downward looking 200 kHz transducer to image the strong halocline present in the estuary. More details are available at http://www.omg.unb.ca/GGE/Kennebecasis_Project.html

Grand Bay Sill Survey - June 2000

In June of the same year a continuation of the Kennebecasis survey was extended to the west into Grand Bay using exactly the same imaging geometry. In this case the shallower sill (5-10m) was imaged over which it was too costly to survey with the EM3000S.. In these depths, the rate of survey was significantly higher than that achieved (and abandoned) when using a single headed EM3000. More details can be seen at : http://www.omg.unb.ca/GGE/Sill_Survey.html.
 



Y2001 operations

Dynamic  Range Considerations

For the 2001 season, whislt the data remained 8 bit, at the request of the OMG, the original 14 bit intensity values were square rooted before being reduced to 8bit to reduce the range of values and thus better preserve the data within the available dynamic range. This single change allowed us now to log data out to 100m slant range without saturating the data in the main beam pattern boresite (at 30 deg off horizontal) whilst still having sufficient dynamic range for the weaker signal in the far range.

Imaging Geometry Considerations

For 2001, whilst we still only had one 200 kHz channel available, we had purchased a second 200 kHz sidescan stave and a switch box.


As a result we were able to make more efficient use of shiptime by alternately transmitting to port and starboard so that neighbouring lines complimented each other, rather than overlapped. We had to pick a single illumination direction, but this stilll allowed us to effectively increase our rate of coverage to a total of 4 x the Y2000 operations (2x for slant range, 2x for switching sides).

Whislt 100m range was achievable, it was found that the extremely low grazing angle data was routinely compromised by imaging tidal fronts and other oceanographic phenomena (surface waves, thermoclines etc..) thus the use of 100m slant range was not ideal for reliable target detection.

Hydrocamp 2001 - The Musquash Estuary Survey

Our first operational deployment in 2001 was for the DFO sponsored survey of the Musquash Marine Protected Area.
    see : http://gge.unb.ca/Research/OceanGov/musquash/hydro/SS/knudsen.html for more details ....

July 2001 Limekin Bay Aquaculture site survey.

Later in 2001 we decided to try out the potential of these pole mounted sidescans for investigations of organic enrichment (primarily faecal material and excess food) under aquaculture site (salmon) cages. this was a follow on of the Letang Estuary EM3000 survey conducted in November 2000 by CSL Plover for Dave Wildish of SABS (which we processed).

Hughes Clarke, J.E., 2001, Remote Acoustic Characterisation of Mariculture Site Sediments: in Hargrave and Phillips Eds., Environmental Studies for Sustainable Aquaculture (ESSA), Canadian Technical Report of Fisheries and Aquatic Sciences No. 2352.

We were hopeful that the 200 kHz imagery would provide an adequate substitute for the  EM3000 data. The data were collected after the cage sites had been moved and they were only occupied by juvenile fish at the time of the survey. The inital results of the survey were extremely promising. Results were presented at the January 2002 ESSA meeting at BIO and at the CHC 2002 meeting in Toronto.

Hughes Clarke, J.E., Wildish, D. and Akagi, H., 2002, Monitoring near-field changes at mariculture sites: a comparison between multibeam and pole-mounted sidescan: in Hargrave, B.T (Editor). 2002. Environmental Studies for Sustainable Aquaculture (ESSA): 2002 Workshop Report. Can Tech. Rep. Fish. Aquat. Sci. 2411: v+ 112 pp.

Hughes Clarke, J.E., Wildish, D. and Duxfield, A., 2002, Acoustic Imaging of Salmonid Mariculture Sites: Canadian Hydrographic Conference Proceedings CDROM. 



Y2002 trials
 

Dynamic Range Considerations

For 2002 an additional upgrade was now available allowing us for the first time to log 16 bit data (actually only 14 bits is used, but still a huge dynamic range advantage over the prior 8 bit data). As first implemented in May, the 16 bit data was available only on linear intensity data (the square-root compression was not implemented again until September of the year). Despite this, with the increased dynamic range, both the high intensites in the main beam pattern and the much reduce data outboard  were all within the newly expanded dynamic range (and thus both could be compensated for using either empirical or predictive beam pattern models) and thus the detail retained is far improved.


log compression

Unfortunately, the real time displays showed only  linear intensity and so the operator could not usefully use the displays for real-time target detection. However, with logarithmic compression in post-processing the data was seen to have all the available signal:

In September of 2002, the square-root compression was again reintroduced allowing even greater effective dynamic range as well as improved real time imaging.

As will also be shown below , by the fall of 2002, swept frequency chirp pulses became available at 200 kHz allowing vastly improved signal to noise conditions.
          




heron mount

Imaging Geometry Considerations

In 2002 we were loaned an extra 320M first by Pacific Region CHS and then subsequently KEL. This allowed us now to have two dedicated 200 kHz channels (the second 320 now has one 28 kHz channel and one 3.5 kHz channel).  We could thus for the first time, image both sides simultaneously as would a regular sidescan.

Additionally in April 2002 we took delivery of CSL Heron, an ex. CHS H-Class hydrographic Survey lanuch. We have now mounted the two sidescan staves permanently in the concave section on either side of the keel (Fig. to left).




Hydrocamp 2002 - St. Andrews Harbour Survey

In May the first deployment of the Heron keel-mounted sidescan was undertaken. The floor of inner St. Andrews harbour was imaged using 40m line spacing and 50m slant range. The line spacing was chosen on the basis of the desired 3.5 kHz sub-bottom profiilng line spacing. As a result of the 200% overlap two images were available imaging both to the east and the west (see figure below).

St. Andrews by-the-sea, HydroCamp, 2002
Final ProductsThis image uses a 40m line spacing (dictated by the required 3.5 kHz subbottom profile line density), and thus has the luxury of using either east-looking or west-looking imagery only. The major problem visible in this image is the fact that there was significant electrical interference in the data (source not understood, not the 200 kHz interphase sounder though).  The second problem noted was that the bubble wake from the propellor on the previous survey line was routinely imaged by the side looking back over to the previous line (thus the image looking away from the last line always proved superior in quality).

   The preliminary Hydrocamp 2002 data is available at the  web site  at which you can compare this image with the EM3000 bathymetry and backscatter data from the same area (not collected simultaneously, rather collected on different tides and requiring much higher line density).

May 2002 - Limekiln Bay Aquaculture Site Survey.

In May of 2002, the KEL sidescans were again deployed in Limekiln Bay (this time simultaneously with EM3000 data on the Heron). The hope was to add an extra time sequence to the study of the site. Whilst the K320 200 kHz sidescan  data was very useful, at this time we noted two limiting factors.

  1. firstly, as with the earlier St. Andrews data, the presence of the propellor wake from the Heron was seen to be one of the strongest targets, so that data that imaged back over the previous track was often of little use.
  2. As the salmon in the pens were now much larger (larger swim bladders), the echo from the bladders was now actually stronger than that of the underlying organically enriched sediments....
For more details take a look at results of acquaculture imaging.

September 2002 - CHS -KEL sidescan project acceptance trials.  

In September of 2002, Bruce MacGowan and Mike Crutchlow of CHS came down to Saint John to formally test out the relative performance of the Knudsen 320 200 kHz sidescan imaging, with respect to the standard EM3000 multibeam operations (both of which area available from the Heron).

Standard Shallow Water Multibeam Operations

In this case we examine the results from using a 130 deg sector (4xWD) multibeam at 80m line spacing in depths ranging from 5 to 15m. As expected  the swath widths (16-55m, remember 1m draft!) do not provide full seafloor coverage. Standard CHS operations run 200% coverage and thus we are well undersampling the seafloor.


The resulting bathymetric surface, whilst incomplete, very precisely resolves small scale bathymetric features within the coverage. In these depths, features of as little as 1-2 decimetres high may be resolved as long as they are a metre across horizontally.
EM 3000 Bathymetry
<< depth encoded (5-15m)
sun-illuminated >>

 
EM3000 Backscatter
<< beam average
full trace >>



 At the same time, the keel mounted 200 kHz sidescans were operating on 100m slant range per side. At this range, 100% sidescan coverage could have been achieved using every second line:


just using lines into out of the page...

just using lines into the page...

K320SS 200 kHz sidescan
<< every second line, both sides
every other second line, both sides >>
both 160m line spacing
100m slant range per side.


With such a luxury of overlap, one can take the opportunity to optimise the visible sidescan image quality. Two major limitations exists with the prior pair of images:

With the overlap one can now generate two images that get around both these problems. For the first, one can replace the near nadir data with data from the far range of the overlapping image. For the second problem, with 200% coverage one can select to view data all on a common azimuth. Note that in both cases, for these approaches to work, one must be extremely confident of the image location. For a towfish deployed system this was rarely the case (how many operators routinely use USBL positioning for towbody location in shallow water, and of those how many have a high quality gyrocompass in the towfish?).
 


imaging to the west, far range overlapping nadir


imaging to the left, far range  overlapping nadir


K320SS 200kHz sidescan
<< just north looking
just south looking >>
both 80m line spacing,
100m slant range

Three limitations can be seen:

Although not seen in these images, on the day before when mirror calm conditions existed, classic Lloyds mirror artefacts showed up.

Chirped Pulse Trials

K320 200 kHz chirp example
As part of the CHS testing we wanted to see what benefits there were to chirping the 200 kHz pulse. The standard shortest pulse length available is 0.1ms (~10kHz bandwidth). We chose to try the longest available chirped pulse of 3.2ms with a nominal 8kHz sweep). If the matched filtering worked properly we should see about the same range resolution while having a 30 fold increase in signal to noise. I think the images to the left speak for themselves.
  • left 0.1ms CW pulse -great range resolution but poor S/N
  • centre 3.0ms CW pulse, way better S/N but lousy range resolution
  • right 3.2ms, 8Hz bandwidth Chirp pulse - great range res AND S/N

October 2002 - 3rd Limekiln Bay Aquaculture site surveys.

These data were collected on the 10th October to look for changes in the distribuiton of the seabed organic matter. No difference in the imaging geometry existed at this time. Results can be seen at: results of acquaculture imaging. And a paper on the quantification of the change is available as:

Wildish, D.J., 2004, Acoustic detection of organic enrichment in sediments at a salmon farm is confirmed by independent groundtruthing methods: Marine Ecology Progress Series, v.267, p.99-105

A more general discussion on the relative advantages and disadvantages of multibeam v. sidescan imaging of salmon acquaculture sites can be found at:

Hughes Clarke, J.E., Wildish, D. and Duxfield, A., 2002, Acoustic Imaging of Salmonid Mariculture Sites: Canadian Hydrographic Conference Proceedings CDROM.



 

Current status of KEL 200 kHz Sidescan System.

As of October 10th 2002.....

With improved real time displays (vertically scrolling, port and starboard reversed and with linear echo amplitude (sqrt(intensity)) the tools may now operationally  be used by the CHS. 4 pairs of sidescan staves have been purchased by the CHS for use in all regions. Significant software development  has been added to smoothly integrate with Hypack, (the CHS current line running and logging software). As the output data is now available in 16bit XTF format, any standard sidescan mosacing package could be used. As the CHS using HIPS/SIPS in all regions, SIPS is  the most likely software postprocessing tool to be used.

Remaining Limitations:

Heading integration
The 320 Echo-control software still does not directly log heading input into the XTF or KEB format files (remains true as of 08/2004!). At this time the heading is logged into the Hypack data stream which is time synchronised to the KEL clock. Thus the heading may be extracted from the Hypack files in post processing. This will meet the immediate needs of of the CHS but leaves other users  who choose not to use the Hypack integration on their own.

Time synchronisation
Another remaining limitation is the stability of the clocks inside the 320. They can be synchronised to the PC controller clocks (which in turn can be synchronised to GPS time). But from analysis of common fix times recorded by the 320 and directly from a GPS receiver, it is clear that the internal clocks drift several seconds in a day (and not just linearly in one direction). This has created problems for newer research ((2003-2004) where we have integrated roll pitch, heading and heave, logged asynchronously through a simultaneously operating EM3000 multibeam system. Synchronisation issues are described in:

Hughes Clarke, J.E., 2004, Seafloor Characterization Using Keel-Mounted Sidescan: Proper Compensation for Radiometric and Geometric Distortion proceedings of the Canadian Hydrographic Conference 2004, Ottawa, CDROM.


Real time bottom tracking and slant range correction

Yet to be implemented in KEL Echo-Control...

At this time, the OMG software does the bottom tracking in post-processing. This can be a problem in high water-column noise areas though. As we usually log EM3000 multibeam at the same time, we often substitute the minimum slant range achieved by that system instead. This obviously is not available to most users however. An independent (and preferably not interfering) single beam echo sounder is a possible alternative, although the same problems exist concerning time synchronisation.

Adaptive real-time TVG's..

Yet to be implemented in KEL Echo -Control...

Without a real time bottom track, no bottom-locked custom TVG's to account for beam pattern affects can be applied. Within the OMG software, once the bottom track is established, ensonified area , beam pattern and grazing angle dependence is quantitatively removed  (described in paper quoted above).  


Activity since 2002

2003 - fully operational for:

2004 - fully operational for:



Last major modification October 14th , 2002, by John E. Hughes Clarke (jhc@omg.unb.ca) - minor mods and updates - Fall 2004.