Advancement in the Estimation of Gas Seep Flux from Echosounder Measurements Elizabeth F. Weidner, Thomas C. Weber, Larry Mayer Center for Coastal and Ocean Mapping, University of New Hampshire Durham, New Hampshire, 03824 eweidner@ccom.unh.edu Abstract Gas seeps can be precisely located utilizing multibeam echosounders (MBES) and split-beam echosounders. Beyond establishing seep location, understanding bubble fate and flux of gas into the water column and atmosphere is an important aspect of climate studies, ocean chemistry, and determination of gas leakage from fossil fuel production sites. Can we advance our acoustics methodologies further, towards estimation of gas flux? A key to determining gas flux acoustically is estimation of seep bubble size distribution. Traditionally bubble size distribution is determined via direct measurements using remotely operated vehicles. We describe an experiment carried out in the Arctic Ocean, establishing a methodology for quantifying seep flux remotely from acoustic measurements. This experiment used a calibrated, broadband split-beam echosounder. The splitbeam echosounders's extended bandwidth (17-24 khz) provided excellent discrimination of individual targets, allowing for identification of single bubbles in many seeps. In these cases, bubble target strength (TS) values were obtained using phase-angle data to compensate for beampattern effects. Bubble size distributions were determined by modeling the relationship between bubble size and TS. Coupling the acoustically-derived bubble size distributions with rise velocity, seep flux can be constrained. As acoustic methodologies advance, we hope to generalize the splitbeam echosounder methodology to other calibrated echosounders, including MBES. Introduction The release of gas in the ocean, either from natural seeps or leaky infrastructure, have become a topic of intense scientific study in the past decade. The spatial distribution and density of gas seeps in marine environments is of interest for researchers from many disciplines, as seeps have been linked to climate forcing, oil and gas deposits, and marine geohazards. Acoustic systems, such as multibeam echosounders (MBES) and singlebeam echosounders (SBES), have been utilized to identify seep features in the water column and map their spatial distributions. The high impedance contrast between gas and water makes seeps excellent acoustic targets and echosounders synoptic view of the water column makes them ideal systems with which to study marine seeps. Recently gas seep research has increasingly turned from a focus on location and distribution to the quantitative estimation of gas flux using a variety of methodologies. From a climate science perspective, estimations of hydrocarbon flux can provide an important and yet unmeasured input into the atmospheric carbon budget; marine seep flux may have serious implications for climate forcing models and future carbon reduction policies as well as the overall alkalinity of the ocean The oil and gas industry also stands to greatly benefit from effective gas flux methodologies, as quantitative measures of gas flux may also assist the cost-benefit analysis of infrastructure leakage. Engineers may be able to utilize estimates of gas flux in assessing the value of hydrocarbon 1
reservoirs for prospective energy sources or determine the geohazard potential of a reservoir, informing policy makers. Current flux methodologies rely on direct measurements such as gas traps or optical systems, to determine the bubble size distribution (BSD) and bubble rise velocities in a seep. These pointsource measurements are often coupled with echosounders and can provide calibration for acoustic measurements. Unfortunately, utilizing point-source equipment is time consuming and expensive, as they take time to position and set up, require extensive training for operation, and require multiple deployments to fully capture the variability even in a small area. A methodology relying on acoustics systems alone to collect measurements of the BSD and rise velocities would be an efficient alternative. Acoustic systems are versatile; calibrated echosounders can synoptically map the water column, obtaining data at point sources or over large areas. Most research vessels come equipped with an acoustic system, reducing the cost of operation that comes with point-source equipment. This paper describes an innovative methodology for the acoustic derivation of gas flux with a calibrated broadband split-beam echosounder. Our methodology does not rely on any form of optical system or other point-source measurement to define the BSD or bubble rise velocities; instead, we calculate these flux parameters from acoustics data coupled with analytical models. Experimental Setting To verify the validity of our methodology an experiment was carried out in the East Siberian Arctic Ocean onboard the Swedish Icebreaker ODEN in August of 2014 as part of the Swedish-Russian- US Carbon-Cryosphere-Climate project (SWERUS C-3). The experimental data were collected in an area of the East Siberian Arctic Shelf called Herald Canyon. Herald Canyon is north of Wrangel Island, on the western edge of the East Siberian Sea in the Russian Exclusive Economic Zone. The main hydrocarbon reservoir in this area is inundated permafrost formed during the last glacial maximum (Semiletov, 1999). Methane gas is released in this area is a function of one of several potential processes: inundated permafrost degradation leading to methanogenesis in the surface sediment, shallow gas hydrate dissociation, or free gas escape through seafloor fissures (Judd, 2004). The mechanism of gas release is somewhat uncertain, however previous studies in the area suggest that gas ebullition is nearly 100% methane (Shakhova et al., 2010). Our methodology relies upon the use of the Kongsberg EK80 wideband transceiver (WBT). This system was connected to the Kongsberg ES18 split-beam echosounder onboard the ODEN, replacing the standard narrow-band EK60 transceiver. The ES18 is hull mounted behind an icewindow, aft of the ice knife on ODEN s bow and has a central frequency of 18 khz. Experimental work in an acoustic test tank have demonstrated that the EK80 WBT and ES18 splitbeam echosounder is capable of producing a broad bandwidth pulse ranging from approximately 15-30 khz. During this experiment the system was operated with a frequency modulated pulse ranging from 15-30 khz. The broad bandwidth provides excellent target resolution as a result of an improved range resolution. As shown in Figure 1, the scatterers in the water column image of the EK80 are resolvable to a much higher degree than those in the EK60. The EK60 system has an approximate bandwidth of 2 khz and calculated range resolution of 75 cm, while the EK80 s range resolution is approximately 10 cm. 2
Figure 1. Water column data collected with the ES18 transducer connected to both the EK80 (left) and EK60 (right) The crux of our methodology depends on the improved range resolution of the EK80 system, as with it we are able to identify individual bubbles in the EK80 echogram. Bubble Model By identifying individual bubbles in the EK80 echogram we can measure an individual bubble s acoustic response. This is central to our methodology, as there are analytical models that relate the acoustic response of a bubble to its physical properties, such as size. The acoustic response, or target strength (TS), of a bubble is a function of the frequency of ensonification, temperature, salinity, pressure, gas composition, and bubble size (Ainslie and Leighton, 2011; Clay and Medwin, 1977). Figure 2 shows the expected TS curves for methane bubbles at 50 meters depth between 1 and 5 mm, in typical Arctic Ocean environmental conditions. These TS curves were based on an analytical model derived in Ainslie and Leighton (2011). Figure 2. Frequency modulated target strength curves for methane bubbles ranging from 1 to 5 mm in an Arctic Ocean setting. 3
The TS curves all exhibit a rapid increase in TS to resonance peaks between 3 and 8 khz, and then a decrease to a fairly constant TS response at higher frequencies. Bubbles of different sizes should be distinguishable from each other based on either the location of the TS resonance peak or the magnitude of constant TS response at higher frequencies. For the EK80 system, producing a FM pulse from 15-30 khz, the expected TS curves should exhibit fairly constant TS values. We can model the expected TS of a given bubble by incorporating measurements taken during the SWERUS cruise and compare those modeled curves to data from individual bubbles we identify in the EK80 echograms. In other words, if we know something about ocean conditions and we can identify an individual bubble in the EK80 echogram we should be able compare the acoustic response to modeled curves, like those shown in Figure 2 and calculate bubble size. Seep Data The SWERUS EK80 dataset consists of approximately 70 hours of split-beam echosounder data collected between August 23 rd and 25 th, 2014. 86 seep features were identified in the data set (Figure 3), of the 86 seeps 68 have recognizable individual bubble scatterers. Figure 3. Overview of SWERUS-C3 survey location at Herald Canyon on the East Siberian Arctic Shelf. Locations of identified seep features in the experimental dataset (inset). The characteristics used to identify individual bubbles include: 1) An increase of at least 15 db in sound pressure from background noise levels, 2) A bubble record length of approximately 20 cm, 3) Relatively constant sound pressure across record length, 4) Distinct decrease in sound pressure to background noise levels, 5) Nearly linear increase in depth with increasing ping number. These characteristics distinguish individual bubbles from records with multiple scatterers in the beam or fish (Figure 4). 4
Figure 4. Seep data from August 25, 2014 (center), individual bubble scatterer (right), multi-bubble scatterer and fish (left). The majority of the seeps in the SWERUS data set contain multiple examples of individual scatters, however in every seep record there are some scatterers that do not fit the description of individual scatters. These are most likely the acoustic response of multiple bubbles within the beam of the EK80 or scattering of sound back from other mid-water targets, such as fish. Figure 5 shows examples of seeps, showing the variation in seep structure, depth, and acoustic response across the experimental data set. Figure 5. Examples of seeps from the SWERUS data set. 5
Although we can identify individual bubble in the majority of the seep features in the SWERUS dataset, we cannot measure bubble size from an individual bubble s acoustic response without first applying a calibration. Calibration Before we can measure bubble size we must correct our data for beam pattern effects and transducer sensitivity. The sound pressure level of an ensonified scatterer is a function of the location of that scatterer in the transducer beam. Imagine two identical scatters: one placed on the main response axis (MRA) or boresite of the transducer, and one placed closer to the edge of the beam. The recorded acoustic response of the scatterers will change depending on location within the beam; the scatterer at the MRA will have a higher sound pressure level than that of a scatterer located farther out in the beam. This is not a function of the physical properties of the scatterers (as they are identical), but a function of the beam pattern of the transducer; therefore, we must correct for the beam pattern effects in our data. Fortunately, calibrating a split-beam echosounder is a well-documented procedure. A target strength beam pattern calibration was performed onboard the ODEN on August 21, 2014 following a standard methodology described by Foote et al., 1981. A 64 mm copper sphere of known acoustic properties was suspended on a monofilament line and moved through the splitbeam echosounder field of view. The calibration data were collected in relatively calm seas and atmospheric conditions while the ODEN drifted in deep water to reduce bottom reverberation. All propulsion systems were secured during the calibration procedure in order to reduce noise in the water column. A CTD was collected immediately before calibration operations. Split-beam echosounders have the ability to locate a target within their acoustic beam through the use of split apertures. The EK80 echosounder is divided into four apertures, or quadrants, and can measure the phase shift of the return from the target. If the target is on the MRA of the echosounder, there will be no phase shift at the individual apertures; but if the target is off the boresite, the scattered wave will arrive to the apertures at different instances in time (Figure 6). This phase shift is exploited to determine the angle between the target and the echosounder. Figure 6. Overview of the split aperture correlation. 6
During our calibration procedure we utilizing the split aperture correlation to locate the calibration sphere in the beam of the EK80 to map out the beam pattern. The goal of a calibration is to collect sphere data throughout the beam of the echosounder to fully map the beam pattern (Figure 7, left). The sound pressure level (acoustic response) of the sphere is expected to decrease with increasing distance away from the MRA with radial symmetry. Unfortunately, due to environmental conditions and physical impediments associated with the size of the ODEN, the calibration file collected during the SWERUS cruise did not cover the full beam of the echosounder, nor was data collected at the MRA (Figure 7, right). Figure 7. Calibration data from a test tank (left) and from the SWERUS data, 4 ms pulse length (right). Utilizing a MATLAB software pack provided by Dezhang Chu (University of California, internal communication), frequency-dependent TS values of the calibration sphere were calculated (Figure 8). Oceanographic parameters, temperature of 5 C and salinity of 35 ppt at the sphere depth of 12 meters below the transducer face, were incorporated into the sphere TS calculation. Figure 8.. TS values for calibration sphere (inset) from MATLAB GUI. 7
The resulting TS curve defines the acoustic response for the calibration sphere, regardless of location in the EK80 beam or transceiver electronics. Angle dependent calibration offset values are calculated by subtracting measured sphere sound pressure values from the modeled TS curve. In order to create a full calibration offset pattern covering the transducer beam, the EK80 was assumed to have radial symmetry. This assumption has since been verified by tank experiments. The beam pattern at the MRA was extrapolated from data farther out in the beam. Again, these assumptions were verified by tank experiments. Two calibration data sets were produced for this experiment, one for data collected with 4 ms pulse length and the other for data collected with 8 ms. With the calibration offset data we can correct our measurements of the bubble acoustic response: correcting sound pressure values to target strength values, to compare to our bubble size models (Figure 9). Figure 9. Uncorrected data (left), location of bubble extracted from the split-aperture correlation (middle), and the resulting calibrated data (right). Measurements The estimation of seep flux requires the measurement of two parameters: bubble rise velocity and the BSD. From these measurements seep flux can be quantified over a bubble size range through the following integral: a max 4 Q gas (z) = 3 πa3 ρ(a)v(a)da a min Where a is bubble radius, ρ(a) is the probability of occurrence, and v(a) is bubble rise velocity. Individual bubbles are identified in the seep record by the characteristics described above. 8
Rise Velocity To calculate rise velocity, a single bubble is identified in the EK80 seep data in a range of pings. For each ping the time stamp and bubble depth is determined from the point of maximum response (Figure 10). The time stamp for each ping is embedded in the EK80 datagram as a computer time stamp. This corresponds to the clock on the EK80 computer, onboard the ODEN. We match this time stamp to an NMEA file, from which we can extract GPS time for a given ping. Bubble depth can be calculated by extracting the transducer range value associated with the ping datagram. Transducer range gives the distance from the transducer face to the center of the beam, R. From the split aperture correlation we calculate the bubble location in the beam off the MRA, ø. Before calculating bubble depth we also must take into account ship heave; which we can extract from the motion data from the multibeam EM122 data. Figure 10, shows bubble depth (D) is given as: D = Rcos( ) H Figure 10. Bubble range, given by location of the bubble and transducer range. Rise velocity is determined from linear regression of the bubble time and depth values through the ping record. Rise velocity is calculated for every bubble in a seep, plotted as a function of depth, and a line of best fit is determined for the flux calculation (Figure 14). Bubble Size With our calibrated data we can now calculate bubble size, by comparing the frequency modulated TS to modeled TS curves from the analytical models. For the transducer frequency range of 15-30 khz, we should be looking for a relatively constant TS value (Figure 2). Once an individual bubble has been identified we manually extract the full bubble record and take the MATLAB discrete Fourier transformation (Figure 11). This provides the frequency modulated sound pressure values for a given bubbles acoustic response. Based on the bubble location in the beam, chosen from the sample with the peak value sound pressure value, we apply the appropriate calibration offset to determine the TS curve. 9
Figure 11. Frequency modulated target strength of a single bubble record. This procedure is repeated for every bubble record between the identified initial and final pings. Any bubble with a location in the beam beyond the 3 db beamwidth, 5.5, is removed due to calibration offset restrictions. An average TS curve and bubble depth for a given bubble is calculated from all measurements for comparison to the model (Figure 12, top). Figure 12. All TS curves for bubble scatterer, with average TS curve (top). Average TS curve plotted against model TS curves for 1-5 mm (bottom). 10
The bubble size model is run with values of temperature and salinity from the most recent CTD cast as recorded in the seep datagram,. The bubble depth value is incorporated from the measurement processes. All bubbles are assumed to be 100% methane and bubbles sized between 1 and 5 mm by 0.1 mm increments are used to calculate TS curves. The average curve from the bubble record is then fit to the modeled curves using a Least Squares fit (Figure 12, bottom). Using this methodology bubble size is calculated for every bubble in the seep record (Figure 13). For a full seep or a given depth, the probability of occurrence of a given bubble size can be calculated by determining the number of bubbles of a given size divided by the total number of bubbles (Figure 14). For example, Figure 14, gives the probability of occurrence of bubbles at 40 ± 5 meters, in 0.5 mm bins. Most bubbles at this depth are small (<1.5 mm) and the curve reflects this with an increasingly decreasing likelihood of probability as bubble radius increases. Gas Flux Figure 13. Seep bubble size distribution as a function of depth. For a given seep record and target depth, volume flux can be calculated using the equation: a max 4 Q gas (z) = 3 πa3 ρ(a)v(a)da a min The bubble rise velocity as a function of size is represented as a best fit curve of all bubble rise velocity values (Figure 14, bottom). The probability of occurrence as a function of bubble size is calculated from the measured BSD (Figure 14, middle). And for the range of bubble sizes 11
(a min a max ), bubble volume is calculated (Figure 14, top). The three curves are multiplied together and the area under the resulting curve is calculated. This area represents the seep volume flux at the given depth in cubic centimeters per second. Figure 14. Measurement curves of bubble volume (top), probability of occurrence (middle), and rise velocity (bottom) for a seep record. Uncertainty As outlined above, for a given seep record with individual scatterers we can extract a value for gas flux from a given horizontal plane (depth location); however, the calculation of a volume flux value has little meaning for applications in climate modeling or industry use if we cannot provide some level of certainty for the measurement. Our current research efforts are centered on the 12
quantification of uncertainty values from each step of our methodology and the eventual propagation of these uncertainty into a final flux uncertainty. The uncertainties associated with the rise velocity measurements can be split into two values: variation in rise velocity values for a given bubble in the echogram and uncertainty in the bubble location as calculated from the split aperture correlation. The variation in bubble rise velocity both for a single bubble target through successive pings and for different bubble of the same calculated size, will introduce uncertainty into the final rise velocity curve. The uncertainty in the split aperture correlation is well documented as: σ angle = 2 SNR, Uncertainty in target angle (σ) is a function of the signal to noise ratio (Burdic, 1991). The calculated bubble size has uncertainty related to two sources: variation in the measurements of the frequency modulated target strength curves and uncertainty in the modeled curves due to sensitivity to temperature and salinity values. For a given bubble there is variation in the magnitude of the target strength curves (Figure 15). This introduces uncertainty into the calculated bubble size, when comparing our average TS curve to the modeled curves. The uncertainty introduced from TS variation is frequency dependent and must be accounted for in the final flux calculation. The modeled TS curves are calculated from analytical models that require both temperature and salinity values for the given bubble environment. These values are determined via CTD cast, however CTD casts are often neither spatially nor temporally close to the seep location. As such, we must investigate the sensitively of our analytical models to both temperature and pressure to determine the extent of uncertainty introduced into our bubble size calculations. Figure 15. All TS curves for a given bubble record (top) and the variation in TS magnitude in these records (bottom). 13
Acknowledgements The authors would like to acknowledge the Captain and Crew of the Icebreaker ODEN, as well as the science party of the SWERUS-C3 program for their effort in collecting the data for this research. Additionally, the authors would like to thank graduate students Alexandra Padilla and Scott Loranger for their input on the EK80 methodology and calibration procedure. References M. A. Ainslie and T. G. Leighton, Review of scattering and extinction cross-sections, damping factors, and resonance frequencies of a spherical gas bubble, J. Acoust. Soc. Am., vol. 130(5), pp. 3184-3208, Aug. 2011. C. S. Clay and H. Medwin, The Bubble, Acoustical Oceanography, 1 st ed. New York: Wiley, 1997, ch. 6, sec 6.3, pp. 194-203. D. A. Demer, L. Berger, M. Bernasconi, E. Bethke, K. Boswell, D. Chu, R. Domokos, et al., Calibration of acoustic instruments, International Council for the Exploration of the Sea, Copenhagen, Denmark, Tech. Rep. 326, 2015. D. Chu., private communication, 2014. A. G. Judd, Natural seabed gas seeps as sources of atmospheric methane, Env. Geol., vol 46, June 2004. N. I. Shakhova, I. Semiletov, I. Leifer, A. Salyuk, P. Rekant, and D. Kosmach, Geochemical and geophysical evidence of methane release over the East Siberian Arctic Shelf, J. Geophys. Res., vol. 115, Aug. 2010. W. S. Burdic, Statistical Basis for Performance Analysis, in Underwater Acoustic System Analysis, 2 nd ed. Englewood Cliffs, New Jersey: PTR Prentice Hall, 1991, ch. 13, sec. 13.3.3, pp. 384-388. I. P. Semiletov, Aquatic sources and sinks of CO2 and CH4 in the polar regions, J. Atmo. Sci., vol. 56(2), pp. 286-306, 1999. Lead Author Biography Elizabeth Weidner is a Master s student in Earth Sciences with the Center for Coastal and Ocean Mapping at the University of New Hampshire. She graduated with a BS in Oceanography from the University of Washington in 2012. After graduating she worked for C&C Technologies as a geophysicist for three years, running a variety of geophysical and conventional surveys, before starting graduate school. 14