Bigelow EM2040 Issues

FRV Henry Bigelow R225
EM2040 install issues.

John E. Hughes Clarke
Center for Coastal and Ocean Mapping
University of New Hampshire
Durham, NH

Harper Umfress LTJG/NOAA
Field Support Liaison - Pacific
Hydrographic Systems & Technologies Branch
7600 Sand Point Way NE, Seattle, WA 98115

This is based on data samples provided by Harper Umfress from the February 2021 assessment of the performance of the new 0.5x1.0 EM2040 recently installed on the Dyson Class FRV, Henry Bigelow (R225).

I've had to guess a bit on the dimensions of the Bigelow. From what I could glean from other FRV reports/Wikipedia, the Oscar Dyson Class are 64m long, with a beam of 15m. They have a draft of 5.9m, growing to 9.2m with the centreboard down.


from Cruise Report, Cutter and Demer, 2008, California Current Ecosystem Survey 2006 - NOAA-TM-NMFS-SWFSC-415	" The keel could be used in retracted, intermediate, or extended positions, placing the transducer faces at: 0m, 2m or 3.75m below the ship’s hull, respectively (5.4m, 7.4m or 9.15m below the sea surface). "

For the animations below, I've assumed the 2040 is 33m back from the bow mounted ~ 2.2m to starboard of the centreline., and the centerboard is on the centreline, is 1.0m wide and is fully (3.75m) extended.

Details of the Hull Blister

Timothy Lichter of KUTI kindly provided the following photos of the blister, acquiring during the install (before final painting). It appears to be installed flush with the hull, just offset to starboard and ahead of the centreboard.


view of downstream edge (see SV sensor abaft)	view of upstream edge

According to the parameter file in the .all files, the RP is truely at the waterline (?), and the Tx and Rx are 5.6m below and ~ 2.2m to starboard. Both are pitched up ~ 1.7 degrees.

WLZ=0.000 (deliberate?)
Transmitter:

S1X=0.042
S1Y=2.349
S1Z=5.621

Receiver:

S2X=0.045
S2Y=2.043
S2Z=5.605

The Dynamic Integration (Wobble) Issue

A conspicuous motion-correlated residual is present in the real-time data that needs to be better understood. Additionally the water column imagery indicated extensive bubble fields. Fortunately, however, masking or attenuation of the bottom tracking bathymetry or backscatter was not particularly significant.

Patch Test Sample Data

As part of the patch test on the 14th February 2021, a series of lines were collected in 21m of water. Apart from a conspicuous wreck for timing and pitch, the surrounding area was remarkable smooth and thus ideal for revealing dynamic integration issues (wobbles).

a pair of lines were run N-S,
a pair were run NW-SE
and several pairs were run E-W over the wreck.

These are presented below, with the associated water column imaging that reveals the natural bubble plumes already present in the ocean.

What is the across -track ribbing (wobble) in the Bigelow 2040 data?

In the real time data (see map above), there is a very apparent heave artefact running through all the data. What is causing that, and how big is it?.

On the 15th of February, conditions were :

4-5 ft chop,
5-7 ft swell @ 340
30 kt winds at @ 340.

In all, not a nice day, but the FRV are designed to collect ICES compliant acoustic data is such marginal seas. They of course have the luxury of the centerboard 3+m below the main hull, which the 2040 does not.

How was the vessel responding to these seas? As the artefacts run through nadir, they are believed to be some sort of heave residual due to either a false lever arm or imperfections in the real time heave. The second hypothesis proved to be the cause. To investigate that the difference between the real-time and delayed heave is presented.

The data is always initially viewed through SIS using the real-time data steam, From that, while the full integration is done, it has to use the real-time heave feed rather than the delayed heave (logged in the group 111 telegrams in the POS data stream ~ 200 seconds after real-time).

Comparing the two heave outputs (figures to the left) over the ~ 4 hour period of the patch test, one can see that there was typically ~ +/- 0.5m of heave. Analysis of the heave during the actual lines compared (see figures below) show that it started at +/-0.4m for the NS lines, increasing to +/-0.8m of heave for the E-W wreck lines.

While the two heave signals look similar, it is important to examine:

their instantaneous differences.
their long period drifting

1 - The instantaneous difference (center plot) shows that there usually was ~ +/- 0.1m of difference that was changing over the typical wave period. And it is that which is subtracted in the figures below.

2 - The long period effects can be seen from the 30 second low pass filtering of the two solutions (the overlain green line). From those one can see that both data sets suffer from long-period drifting (even though both are high pass filtered).
The delayed heave uses an acausal filter and thus this is minor, but will just reflect any slow speed changes. One can see near-static offsets of a few centimeters as the lines reversed.
The real-time heave is by necessity causal and thus one sees larger impulse responses during the turns (when the vessel speed changes and the squat adjusts).

The actual differences are a combination of the short period (within wave period) and long period differences. What shows up in the grids that the eye notices, are the short period differences. Exactly why they are there is not clear. Often a too short heave bandwidth setting can be a problem, but the motion here was only ~ 6 seconds.

The difference in heave values peaks at about 15cm but is usually within +- 10 cm, which may be within IHO specifications. But on these, otherwise featureless, Gulf of Mexico seafloors they show up clearly.
The noticeable difference between real and delayed heave coming out of the POS has long been known and reported as part of the NAVOCEANO open ocean TAGS testing (first noted in 2007, and still a problem in the 2020 trials). Their operational solution is, of course, to apply delayed heave. Interestingly in 2011 and 2012, NAVOCEANO tested a competing sensor that also provided both a real-time and delayed heave solution. Notably, that sensor did not exhibit nearly the same level of difference between real-time and delayed have (and all bandwidth settings were tested).

The figures below show three lines into the seas run at the three headings, alternating between the real-time and delayed heave solution. The improvement is very apparent and removes almost all the wobble. This will, however, add an extra level of necessary post-processing to FRV-acquired data which, while not an issue for standard OCS operations (which do full PPP and reprocessing through POS-Pac), is concerning for the reduced manning being envisaged for the FRVs.

The Bubble Wash down or Wave Plume Issue:

In addition to the real time wobble problem, it was apparent that the marginal sea states were impacting the quality of the data. To a first order these were suspected to be motion correlated, perhaps related to wave-driven bubble sweep.

To investigate this suspected bubble sweep, three pairs of reciprocal lines (N-S, E-W and NW-SE) are presented below to assess the performance of system going into or against the seas. In each case, a vertical section of the water column scattering under the vessel is presented, geographically registered to the corresponding seabed bathymetry and backscatter. And for each, the spectrum of the roll and pitch is shown together with a histogram of the roll and heave distribution.

To illustrate any correlation between vessel motion and bubble generation, for a short 6 second window (~ one wave period) 100 swath (50 dual ping cycles) are presented as an animation.

North South Lines:
There is a significantly higher bubble plume frequency and intensity going into rather than with the seas.
Nevertheless, at either azimuth, the data was not compromised.
heave is ~ +/-0.4 m.

line 04 -steaming North	line 05 -steaming South
shorter period pitch, less rolling, same heave	longer period pitch, more rolling, same heave
note undisturbed BS and bathy.
this subset was chosen to be where the water column scattering was the highest. Nevertheless, there is no apparent impact on the seabed bathymetry or backscatter.	this was chosen where interference from another sonar was suspected. But again no masking of the seabed backscatter intensities, nor modulation of the bathymetric tracking was apparent.

SE-NW Lines
This second pair of lines were run SE-NW. The pitch period wasn't significantly different on the reciprocal, suggesting the waves were on the beam?. Nevertheless, bubble masking was now seen, but only steaming to the NW.
Heave now increased to ~+/-0.55 m

line 02 -steaming NorthWest	line 03 -steaming SouthEast
1 deg roll bias to port longer period roll, more heave, more roll	1 deg roll bias to stbd shorter period roll - less heave
note bubble masking stripes in BS
complete masking due to intense cloud over to starboard (extending from hull to 4m below). It never appears to actually reach the transducer.	period of almost no motion - ye the deepest local cloud seen. But not intense enough to impact either the bathy or backscatter modulation.

E-W Wreck Lines
The third pair of lines, conducted slightly later, were all roughly E-W crossing the wreck site. Pitch period was shorter going east suggesting into the seas. Interestingly though, the bubble masking only occurred when going to the west.?
Heave now increased to a local maximum of ~+/-0.75 m

line 11 -steaming East	line 12 -steaming West
1 deg roll bias to stbd shorter period pitch, less roll - same heave	2 deg roll bias to port longer period pitch, more roll, same heave
	note bubble masking stripes in BS
much deeper bubble cloud, but no high intensity regions and thus absolutely no impact on bathy tracking or backscatter modulation. (nice view of wreck)	The most intense masking seen. Note that it is a result of a very intense cloud over to starboard that never actually reaches the transducer. But it is still 2-3 below the hull (7-8m below the surface).

What is very apparent is that the centerboard avoids almost all the bubble cloud issues. The masking is seen for the 2040 data, but only when the heave gets above about +/-0.6m. also the masking is never directly below, the transducer, only offset at lower grazing angles. And it also seems to be preferentially on the starboard side (recall the array is offset to starboard)..

While pulsed bubble clouds are very apparent, it seems that they are not related to the vessel motion. Rather they may be due to waves breaking well forward of the vessel. Fortunately, most of the time, while very clear in the water column scattering, the bubble clouds are not intense enough to have a detrimental effect on the 2040 seabed backscatter intensity or the bathymetric bottom tracking. The motion observed for this limited trial, however, is unusually short (5-7 second periods more typical of inner shelf), and thus probably wouldn't reflect a deeper outer shelf environment where the FRV's more commonly operate.

The few times you do get intense masking, it seem triggered by a "clump" of much denser bubble that are passing by. Notably, those clumps are not necessarily right up against underside of the hull, rather they are usually offset of the side and extend several meters below.

It would be fascinating to temporarily add a forward-looking multibeam to the front of the centreboard to see what is coming- and from where...

page created by JEHC, June 2021