Visualization and Analysis of NW Rota-1 Eruptive Plumes Utilizing QPS-

Fledermaus Software Packages

Susan G. Merle Geo 510 internship

ROV frame grab showing CO2 bubbles rising from the eruptive vent during explosive bursts in

2006, from Chadwick et al. (2008).

Background

The NOAA Vents program and associates have participated in 5+ expeditions to NW Rota

submarine volcano in the Mariana arc. (Figs. 1 and 2) The first was in 2003 when the seamounts

of the Mariana arc were mapped using the R/V Thompson EM300 multibeam system. Figure 3 is

a 3d image of the entire NW Rota volcanic edifice from the summit (517m) to the base

(~3000m). Intense particle plumes were indicated by CTD tow-yos over the volcano (Fig 4).

The group returned to the area and conducted ROV dives in 2004, and that is when the explosive

nature of the eruptive vent, Brimstone, was first observed. Eruptive activity was again observed

during ROV dives made by members of the NOAA Vents program and associates in 2006 and

2009. When the group returned to NW Rota in 2010 for another series of ROV dives it was

apparent that something had changed. Surface differencing of bathymetric grids, comparing the

2010 pass over the summit to the data collected in 2009, revealed a large landslide event had

occurred after the 2009 cruise. An in-situ hydrophone deployed by Bob Dziak (OSU / NOAA

Vents) confirmed that an eruptive burst had triggered a major landslide in August 2009 [Dziak et

al., 2009; Dziak et al., 2011]. Chadwick states that the nature of the earthquakes in August, and

previous pre-cursor quakes, plus the fact that the earthquake swarm did not follow a mainshock-

aftershock pattern, indicates that the cause was magmatic intrusion, not tectonic [Chadwick et

al., 2012]. ROV dives in 2010 revealed a very different eruptive venting scenario than in

previous years, when Brimstone was the only eruptive vent near the summit. In 2010 Brimstone

was less active, but 4 other eruptive vents had popped up in a NW/SE line below the summit

extending ~100m from the westernmost to easternmost vents (Fig. 5).

Mid-water data

During the 2010 expedition we had the opportunity to use the new mapping system on the R/V

Kilo Moana. The EM122 (12 kHz) system collects seafloor bathymetry and backscatter data, as

well as data in the water column. The simultaneous collection of mid-water data is a new

technology only made available to the research community within the last couple years.

For the Geo 510 internship my goal has been to image and analyze the results of this new

magmatic plumbing system by observing changes in the eruptive vents near the summit over

time. Throughout the expedition we made numerous passes over the volcano summit and

observed the bubble plumes that rose off the eruptive vents. The bubbles consist mainly of CO2,

and have been commonly observed rising from the seafloor during ROV dives. A vast mid-

water data set was collected during the 2010 expedition over NW Rota. Numerous passes over

the summit to image the bubble plumes were designed to observe the variability of the plume

over time. The mid-water data set is big, 120 gigabytes, totaling >95 hours of observations over

a 12-day period (Fig. 6). This sonar data was mainly collected between daily 12-hour ROV

dives. Generally, we had the ship drive repeatedly over the eruptive vents at a range of ship

speeds (0.5-4 knots) and headings. In addition, we collected some sonar data during the last

three ROV dives when the ship was stationary over the eruptive vents to look for changes in the

plume at a fixed location over time. In total, we have ~120 passes over the eruptive vents and

~44 hours of data collected while stationary over the vents [from Chadwick NSF proposal].

Data processing steps

Software called FMMidwater has been designed to deal with this new mid-water technology.

Maurice Doucet (QPS Fledermaus) gave me a tutorial of the beta version of the software in

2010. He has shared valuable insights regarding midwater data visualization over the past 2

years. The first data processing step is to convert the raw multibeam files to a generic water

column (gwc) format. The gwc files can be analyzed with the FMMidwater tool in a number of

ways. (Fig.7). 3d volume objects can be created if the eruptive plumes contain enough gas to

isolate them from the surrounding water column amplitude (often referred to as intensity) values.

a) Stack view The first step is to view the gwc data perpendicular to the ship trackline in a stacked view. The stacked view is a valuable way to see the entire file quickly, start to finish, from left to right.

Plumes with high amplitude values are easily discernable in the stack view. The FMMidwater

manual states that the stack view takes all of the beams in the swath, collapses them down

together in an overlapped manner and displays the maximum signal level for every discrete range

increment in the display. Threshold filtering is used to isolate the plumes from the surrounding

water column. The stack view is not geometrically corrected (Fig 8).

I wanted to compare the 3d objects derived from the threshold filtering with the images I saw in

the stack view. I was aware that the entire plume could not be presented as a volume object,

only those voxels (3d volume cubes) that had high enough amplitude values to survive threshold

filtering comprised the volume object. I inserted the 2d vertical plane stack image into a

Fledermaus scene by determining the start and end coordinates of the line and using those

positions to define the bounds for the vertical plane (Fig. 9). Since the ship does not travel a

perfectly straight line from start to end, the placement of the stack in the 3d view is approximate.

One will notice a dark red trace on the stack that follows the seafloor. That is the seafloor, not to

be mistaken for high-intensity values in the water column. The difference between the high

amplitude seafloor values and high amplitude values in the water column is easily discerned by

the trained eye.

b) Beam fan A beam fan object can be exported from FMMidwater (Fig. 10 bottom). It is a geometrically

correct beam fan from the ship transducer to the seafloor, ping by ping, travelling through time

from start to end of the file. The beam fan object can be brought into Fledermaus and moved

with the time slider, making this truly a 4d time-aware object. Noting high amplitude values

near the seafloor that rise as one moves the time slider can assist in determining where the

plumes originate.

Other export options from FMMidwater include creating beam curtains, and point cloud objects.

3d objects created with FMMidwater can then be brought into Fledermaus where they are

geographically referenced in 3+ dimensions.

Data analyzed to date

Lines 209 258 were analyzed for this internship project, which is only 1/3 of the 150 total data files collected. This data set was collected at a slow speed (0.5 - 2 knots) therefore the data

density is good, with more pings to define the plumes. Also, for all but 5 of the lines, the same

direction was followed (NW/SE) along the summit in the same orientation as the 5 eruptive vents. That survey pattern led to better data density over the vents (lots of pings), which in turn

made it easier to differentiate plumes rising from individual vents on the seafloor. Lines 209 258 consisted of 28 passes over the summit vents in 18.25 hours. The interval between passes

over the vents varied from 20 to 50 minutes, with the exception of 1 two-hour interval between

lines 230 and 233, when the ship re-alignined to pass over the vents in a N/S pattern. For each

line a group of figures was created including a 2d stack view with annotation, a 3d volume object

above the summit, and a 2d stack view inserted into the 3d scene as a vertical plane. Figures 8,

9, and 10 are the group of figures created for Line 254.

Remaining data analysis

The next series of water column data to process is lines 102 122 (3/21/10 19:18 3/22/10

01:47). It is a series of 6.5 hours of continuous data collected moving 4-5 knots.

We also collected water column data while stationary over the vents with the ship parked, only

moving when ROV navigators needed to move. Those data include lines during dives J2-493,

J2-494, J2-495, as well as during a plankton haul. From the data Ive looked at during ROV

dives it is difficult to discern any kind of plume data in the stack view. Because the ship is more

or less parked in the same spot during a dive, perhaps not directly over the vents, eruptive

activity on the seafloor may not be seen in the midwater data. The best way to proceed with the

data collected during ROV dives is to determine when we saw bubbles on the seafloor and then

go to that part of the water column file to try to see that venting in 1 or 2 pings of the data. I

dont believe I will be able to create 3d volume objects from the stationary data, but I havent

looked at enough of it to determine that absolutely.

Statistics

An excel spreadsheet has been created to accompany the 3D visualization techniques. Also refer

to the poster that accompanies this document for images and information about each summit

plume.

Major attributes noted:

Line #, time, speed, direction, survey comments at sea, water column data comments, plumes

(yes or no), number of vents involved in plume creation, height of plume tops, height of plume

wisps, maximum amplitude (intensity) in the stack view, min/max height of 3d volume object,

min/max amplitudes in 3d volume object, whether the vents are deflecting and in what direction,

deflection from vertical and how far off the seafloor does that start, 3d reach of the volume

object above the seafloor.

Major attribute values:

Largest number of plumes seen on stack view: 4 plumes on line 230. All 5 eruptive

vents contributing.

One pass (line 256) only 1 plume was visible.

Every pass over the summit there were plumes visible in the water column data.

Highest amplitude value: A=26 Styx (line 250)

Lowest (high) amplitude value: A=-15 (Line 235) could not create 3d volume object

Highest plumes: (line 230) Z=175m, 415m rise from Charon; (line 233) Z=215m, 345m

rise from Phantom

Highest plume wisp: (line 233) Z=90m, 470m rise from Phantom

Only 1 line when the plumes were not deflecting (line 238)

Main plume deflection direction: WSW

Plume deflection started at the seafloor in > 1/3 of the passes. Deflection height above

the seafloor varied from 0 (at seafloor) to 200m above the seafloor.

Deflection of the plume from vertical varied widely.

Farthest plume reach above the seafloor, determined from 3d objects, was 200m from

Styx/Charon (line 242). That plume deflected from vertical 45 100m above the seafloor.

(Aside: 3d volume objects were used to determine deflection values although the actual

plumes do extend out farther and higher than the volume object does so this is a

minimum measurement.)

Significant findings from the water column data

I believe that the water column data can be used as a proxy to determine the level of eruptive

activity above submarine volcanoes that have robust CO2 bubble plumes. It may not be totally

quantifiable, but the water column data can give us some qualitative information about the vents,

like the major attribute data listed above.

Styx was the most active of the vents and often had the highest amplitude values. Sulfur was

next in the mix, often active and displaying relatively high amplitude values. Plumes rise from

the individual vents and often combine above the seafloor. The highest plumes were a

combination of Styx and Charon, the easternmost summit vents. Charon is the deepest of the

eruptive vents near the summit, 30m deeper than Styx and Sulfur. It was difficult to create 3d

volume objects that start at the base of Charon, but the beam fan object helped hone in on that

vents activity. The data indicates there may be more bubble plume venting to the east of Charon, but if so it is low amplitude, appearing like a haze on the eastern side of the stack and

beamfan objects. ROV dives confirm that there are vents to the east of Charon, but not the

eruptive type observed at the summit. Its possible that we have not discovered all the eruptive vents formed after the landslide. Given that, it is also important to remember that the vents are

ephemeral, and that the plumbing system could re-arrange itself again and we could find a whole new eruptive scenario in the future.

Most plume deflection was to the WSW and SSW, but the plumes also deflected to the W,

NNW, NNE and SSE. Plumes deflect at different heights above the seafloor and even change

deflection angle from vertical as the plume rises. Deflection from vertical is likely caused by

currents near the seafloor. On one line (238) there was little to no deflection of the main plume.

There was one instance, line 244, were the plumes split and deflect in different directions.

During lines 209 226 (6.5 hours) venting was fairly subdued, usually 2 vents erupting at the same time, sometimes 3. Plumes rise 150 to 300 meters above the vents. Then on line 228 more

vents are active and the plumes are rising higher in the water column. On the next pass (line

230) all 5 summit vents are active. This is the only instance in this series where all 5 vents were

erupting at the same time. This pass over the summit also saw the highest main plume rising to

z=175, a 415m rise from Charon. Plume wisps rose to within 115m of the sea surface. After that

there is a 2 hour gap in the data series as the ship took a different course (N/S). That next pass

over the summit (line 233) the highest plume wisps were observed 90m from the sea surface, a

470m rise from Phantom, but the amplitude values had fallen sharply (from 7 to -7). On the next

pass (line 235) the amplitude values were so low that a 3d object could not be made. After that

the venting gradually increases again until on line 250 the highest amplitude values of this data

series were recorded (A=26 from Styx). Venting after that is steady with 2 plumes visible. The

exception is line 256, when only 1 plume is visible above the summit. (See the poster that

accompanies this document).

Future plans

Bill Chadwick submitted an NSF proposal entitled Collaborative Research: Dynamics of eruptive plumes above a submarine arc volcano. Future plans include analyzing the rest of the water column data. All those data will be integrated with other temporal geophysical data sets

including:

Acoustic Doppler Current Profiler (ADCP) data will help unravel the current regime in

the area on NW Rota during the 2010 expedition

In-situ hydrophone data from a moored hydrophone near the summit and a small portable

hydrophone deployed 50m from the eruptive vent Brimstone. The hydrophone data will

be compared to variations seen in the water column data.

ROV observations on the seafloor will indicate when vents were active. Those

observations will be compared to the hydrophone data and the water column data.

Tidal data during the expedition in 2010 should also be analyzed. Tidal data may

influence plume heights, etc.

A 3d plume model will also be created for NW Rota tracking temperature, salinity and bubble

concentration during the bubble plume surveys. The results of this study will hopefully produce

peer-reviewed manuscripts fostering better understanding of eruptive processes at submarine

volcanoes.

References

Chadwick, W. W., Jr., R. P. Dziak, J. H. Haxel, R. W. Embley, and H. Matsumoto (2012), Submarine

landslide triggered by volcanic eruption recorded by in-situ hydrophone, Geology, 40(1), 51-54,

doi:10.1130/G32495.1.

Dziak, R. P., R. W. Embley, E. T. Baker, W. W. Chadwick, Jr., J. A. Resing, H. Matsumoto, J. H. Haxel,

S. L. Walker, and D. R. Bohnenstiehl (2009), Long-term explosion records from two erupting

submarine volcanoes in the Mariana and Tonga island-arcs, Eos Trans. AGU, 90(52, Fall Meet.

Suppl.), Abstract V44B-02

Dziak, R. P., E. T. Baker, A. M. Shaw, D. R. Bohnenstiehl, W. W. Chadwick, Jr., J. H. Haxel, H.

Matsumoto, and S. L. Walker (2011), Gas flux measurements from a year-long hydroacoustic

record at an erupting submarine volcano. Abstract V52C-07, presented at 2011 Fall Meeting,

AGU, San Francisco, Calif., 5-9 Dec.

Smith, W. H. F., and D. T. Sandwell, Global seafloor topography from satellite altimetry and ship depth

soundings, Science, v. 277, p. 1957-1962, 26 Sept., 1997.

Walker, S. L., E. T. Baker, J. A. Resing, W. W. Chadwick, Jr., G. T. Lebon, J. E. Lupton, and S. G. Merle

(2008), Eruption-fed particle plumes and volcaniclastic deposits at a submarine volcano: NW-

Rota-1, Mariana Arc, J. Geophys. Res., 113, B08S11, doi:10.1029/2007JB005441

Figure 1. Location map of the western Pacific highlighting NW Rota volcano. Mercator projection.

Base layer created from satellite altimetry data at 1800 meter resolution [Smith and Sandwell, 1997].

Figure 2. NW Rota volcano. EM300 bathymetry data. 25 meter resolution. 50 meter contours. Mercator

projection.

Figure 3. NW Rota volcano. EM122 bathymetry data. 25 meter resolution. Depths range from 517m at

the summit to ~3000m at the base.

Figure 4. Two-dimensional profiles of particle plumes at NW Rota-1 in (a) 2003 and (b) 2004, from

Walker et al. (2008), showing well-defined plumes above the summit, rising about 100 m above the

eruptive vent. Saw-toothed tow path of the CTDO during tow-yos is shown (light gray lines). Deeper

plumes in 2004 are from small landslide events. NTU are nephelometric turbidity units.

Figure 5. NW Rota volcano summit eruptive vents. SM2000 bathymetry data at 2 meter resolution

overlaid on EM122 multibeam data at 25m resolution. Distance between westernmost (PH) and

easternmost (CH) vents is ~100m. Vent abbreviations: PH=Phantom, SU=Sulfur, BR=Brimstone,

ST=Styx, CH=Charon.

Figure 6. Timeline in late March 2010 showing times of ROV dives (blue), when mid-water

multibeam data were collected (red), and when mid-water multibeam data were collect during

ROV dives (purple).

Figure 7. Screen shots of the FMMidwater software program, showing processing windows

displaying mid-water multibeam data from the 2010 expedition to NW Rota-1 volcano, in (a)

beam fan mode (displaying a single ping), and (b) data in stacked mode (displaying multiple

pings along a trackline).

Figure 8. 2d stack view showing 2 lines concatenated together. Vent positions and times are

noted as well as amplitude values and plume heights.

Figure 9. Top) Just the stack view inserted as a vertical plane into the 3d scene. Bottom) Stack

view inserted as a vertical plane and 3d volume object. This view allows one to see how much of

the total plume is preserved in the 3d object.

Figure 10. Top) 3d view showing volume objects representing the eruptive plumes above the

summit vents. Amplitude values for the 3d object are on the right. Those values are lower than

in the stack view. Bottom) 3d side view of volume object with the beam fan object in the

background. The plume wisp can be easily detected on the beamfan, but usually cannot be

preserved in the 3d volume object because of lower amplitude levels as it ascends.