Using Gliders to Study a Phytoplankton Bloom in the Ross Sea, Antarctica

Vernon Asper Craig Lee, Jason Gobat Department of Marine Science Applied Physics Laboratory University of Southern Mississippi University of Washington Stennis Space Center, MS, USA Seattle, WA, USA

Walker Smith Karen Heywood, Bastien Queste Virginia Institute of Marine Science School of Environmental Sciences College of William and Mary University of East Anglia Gloucester Point, VA, USA Norwich, United Kingdom

Michael Dinniman Center for Coastal Physical Oceanography Old Dominion University Norfolk, VA, USA

Abstract-Over the last several decades, numerous approaches magnitude over three weeks or less [Smith et al., 2000, 2006; have been used to observe the rapid development of the annual Smith and Asper, 2001). phytoplankton bloom in the Ross Sea, including ship-based sampling, moored instrumentation, satellite images, and In contrast to P. antarctica, diatoms accumulate in computer modeling efforts. In the Austral Spring of 2010, our summer [e.g., Smith and Nelson, 1985; DiTullio and Smith, group deployed a pair of iRobot Seagliders equipped with 1996], and also are more likely to be observed near ice edges fluorometers, oxygen sensors and CTDs in order to obtain data [Arrigo et al., 1999; Garrison et al., 2003]. The spatial and on this phenomenon over the entire duration of the bloom. temporal partitionings of these groups may be related to Data from these deployments will be used, along with samples differential responses to available irradiance (which in turn is from the recovery cruise and satellite data, to model and better a function of ice cover and vertical mixing depth) and/or understand the dynamics of this phytoplankton bloom. micronutrient concentrations. It has also been speculated that increased stratification induced by anthropogenic Keywords-component; gliders; Antarctica; phytoplankton atmospheric forcing will alter the phytoplankton assemblage biomass to one that can optimize growth in high light environments [i.e., diatoms; Arrigo et al., 1999]. A tremendous amount of I. INTRODUCTION interannual variability in the biomass and distribution of The Ross Sea continental shelf is an extremely phytoplankton biomass and net community production has productive region in which much of the photosynthesis been observed [Arrigo and van Dijken, 2003, Smith et al., occurs within a seasonal polynya. This polynya generally 2006; Peloquin and Smith, 2007, Smith and Comiso 2008]. becomes ice-free during mid summer [Arrigo and van In order to understand the distributions of these Dijken, 2003; Barber and Massom, 2007], allowing organisms at any particular period and location, it is increased light into the surface waters. The melting ice also necessary to be able to access this remote region and to make decreases the surface salinity, increasing the density detailed measurements of appropriate temporal and spatial stratification and further enhancing phytoplankton growth. scales. Before the arrival of scientific icebreakers, this The blooms occurring in austral spring (November– access was sporadic at best but even with these vessels, it is December) are often dominated by the colonial haptophyte impossible to study the early stages of the blooms because of Phaeocystis antarctica [El-Sayed et al., 1983; Smith and the ice conditions. In order to overcome this limitation, we Gordon, 1997; Arrigo et al., 1999]. Its growth is initiated in elected to employ autonomous gliders to make these early November, and its biomass reaches its maximum in measurements as the polynya was forming and the blooms mid- to late December [Smith and Gordon, 1997; Smith et were in their early stages. Described below are the details of al., 2000; Tremblay and Smith, 2007]. These blooms this experiment, its results, and some suggestions for future disappear even more rapidly than they are generated, with projects. chlorophyll concentrations often decreasing by an order of

National Science Foundation ANT 0838980 II. METHODS

A. The gliders Although there are several excellent autonomous gliders on the market, we chose the iRobot Seaglider because it was the only one to offer the endurance we needed for this work. With a standard lithium battery, the Seaglider is capable of missions lasting as long as 9 months and covering several thousand kilometers. While the cold temperatures and shallow depths were expected to reduce this capacity in our application, battery power was not expected to be an issue for the 10 week planned duration. Like all autonomous gliders, the iRobot Seaglider functions by varying its buoyancy and shifting an internal weight (battery pack). When heavier than water and with the weight forward, the vehicle glides forward and down when positively buoyant and with the weight back, the glider flies up. The operator has the option of controlling the “thrust” of the glider by manipulating the magnitude of the buoyancy changes and the position of the internal weight, allowing more aggressive maneuvering when currents are present. Unlike some gliders, the Seaglider employs no external moving parts and, instead, uses the position of the battery, which is heaviest on its lower surface, to roll the vehicle. For example, to roll left during a climb, the battery is rotated to the left, causing the glider to roll in that direction and creating more turning force Figure 1. The glider was transported to a remote location on through lift in the wings. This mechanism, along with the the sea ice for compass calibration. optimized, streamlined shape, make the Seaglider one of the and use the third as a spare. The NSF contractor (Raytheon most efficient designs available. Polar Services) handled all of the shipment logistics from The standard Seaglider sensors were used for this project, Port Hueneme to McMurdo, making this process relatively including the on board Seabird CTD, Aanderaa 4330 simple for the science team. Concerns about safety of the dissolved oxygen sensor, and a Wetlabs BBFL2-VMT lithium battery packs required special consideration both in optical sensor which measures optical backscatter at 650nm the transit to the ice and for helicopter transfer to the launch as well as both chlorophyll and CDOM fluorescence. These locations. sensors are installed so that they produce minimal drag and yet are exposed to the undisturbed oncoming flow of water. B. Vehicle preparations Data are logged at different intervals depending on the depth Before launching the gliders into this formidable with higher rates in the upper water column. In the upper environment, we wanted to do whatever was possible to 250m, for example, all sensors were logged every 5 seconds increase the likelihood of success. In addition to the normal while between 250 and 500m, the CT sensor was sampled factors that can potentially result in a failed glider mission, every 10 seconds, the oxygen sensor was sampled every 20 we addressed three additional concerns that are specific to seconds and the optical sensors were not sampled at all. This the polar environment: scheme was employed to provide dense data sampling in the upper, euphotic zone and reduced sampling at depths where 1) Temperature biological effects were greatly reduced. Water temperatures in the Ross Sea can occasionally reach 0°C but are mostly around -1.8°C with air temperatures All data were transmitted via Iridium satellite to a server far below that. This results in several potential problems at the UW APL (University of Washington Applied Physics including the viscosity of the oil in the buoyancy engine, Laboratory) where they were posted to a web site that the battery capacity, current draw of the pitch and roll actuator team in the field could access. The glider was also piloted motors, and the overall function of the electronics. Each of from the APL, including transmission of revised waypoint these were addressed by placing the gliders in a cold room (- files, fine tuning the attitude and buoyancy settings, changing 9.48°C) during which all systems were exercised and proven thrust in response to observed trajectory offsets, to be functional. modifications to the diver terminal depth settings, and numerous other, controllable parameters. 2) Navigation and the magnetic compass Although three gliders were prepared and transported to McMurdo Station in Antarctica, the plan was to deploy two Figure 2. Location map showing the launch locations and trajectories of both gliders.

While underwater, the glider navigates by dead reckoning with GPS fixes obtained at every surfacing. This dead reckoning utilizes a Sparton SP3003D magnetic compass module for directional information and speed through the water is calculated using ahydrodynamic model that is based on the pitch, density differential of the vehicle, and observed vertical speed. Given this dependence on the magnetic compass, the steep inclination (nearly vertical) and declination (~140°) of the magnetic field, we took steps to ensure that the vehicle was able to function properly under these conditions. To calibrate the compass, we rigidly attached a ComNav Vector G2 satellite compass to the vehicle in place of its vertical fin (Fig. 1). This sensor determines its true heading using GPS satellites so it is immune to the vagaries of the magnetic field. With this compass installed, we took the gliders out onto the sea ice, Figure 3. Glider 502 was launched using a VideoRay ROV in the several miles from shore, where we were confident that McMurdo Sound polynya where a thin layer of grease ice did not pose a magnetic interference from the igneous rocks at McMurdo problem. Station would be minimal. The calibration process, which is part of the vehicle’s software and was modified specifically attitudes. At each orientation, we confirmed that the vehicle for this deployment, required that the glider be aimed at was able to calculate an accurate true heading based on the several points of the compass and pitched and rolled 30° in measured magnetic field and the applied calibration. To each direction. During this process, the true heading from convert magnetic to true, the glider used declination values the satellite compass and the magnetic heading from the from the standard geomagnetic model onboard its GPS vehicle’s onboard magnetic compass were logged receiver. continuously, resulting in a calibration table that the vehicle could use for navigation. Once the process was complete, 3) Ice cover calibration coefficients were loaded into the vehicle’s Working with autonomous vehicles in polar regions memory and accuracy checked by again orienting the vehicle requires careful planning to avoid hazards associated with in multiple directions and establishing multiple pitch and roll both sea ice and icebergs. Because the gliders require access to the surface for both navigation and communication, it is C. Deployment plan A standard launch procedure includes an initial, shallow (~100m) dive with rapid return to the surface to ensure that all systems are functional. Consequently, we diligently searched for a suitable large opening in the ice which was accessible by helicopters which are not permitted to fly over open water due to safety and contractual constraints. Ideally, we hoped to deploy the gliders on the northeast side of Ross Island, near Cape Crozier (Fig. 2). A reconnaissance flight to this area on November 15, however, revealed very little open water near the solid, shore-fast ice and generally poor sea ice conditions near what open water was present. Alternatives were considered including lowering the vehicles from the cliff formed by the Ross Ice shelf roughly Figure 4. Glider 503 was launched into the Ross Sea through a breathing 50m above the Ross Sea, launching in the McMurdo Sound hole that was created and maintained by minke whales. polyna and making a long transit around Cape Bird and Beafort Island into the Ross Sea, or possibly cancelling the imperative that under-ice periods be minimized. Also, while launch altogether. the gliders do travel very slowly (~0.2m/sec) and collisions with ice are not expected to cause damage, it is possible for D. Deployments the vehicle to become entrapped or even crushed by heavy Following the reconnaissance flight and numerous planned our deployments to minimize under-ice transits and discussions with scientists and pilots who had overflown the to spend as much time in the polynyas as possible. When area, we elected to deploy the first glider (SN 502) in the under-ice operation was unavoidable, for example at launch McMurdo Sound Polynya, roughly 50km from the McMurdo at the ice-edge or when transiting pack ice, the glider relied Station. This site offered excellent, solid, shore-fast ice from on ice avoidance algorithms developed for APL-UW which to launch, and the aircraft pilots reported a large area Seaglider deployments in Davis Strait. of open water adjacent to the sea ice. Upon arrival at this site, we discovered that the “open” water was actually

Figure 5. For each dive, a complete set of engineering data was transmitted, allowing the pilot to fine tune all parameters. to the polynya and because a minke whale breathing hole had been spotted by one of the helicopter pilots when flying over this area. We used this opening to launch the glider (Fig. 4) in spite of misgivings regarding the relatively strong southerly current that promised to drag the glider under the shore-fast ice. This glider failed to gain access to satellite contact for 36 hours and, when it did, its location was to the east and immediately adjacent to the Ross Ice shelf

III. RESULTS

A. Glider trajectories During the missions, each glider provided complete engineering data (Fig. 5) so that the pilot at UW/APL could monitor its progress and fine tune all of its operational parameters. These data included the voltage levels of both propulsion and science battery packs so that we could ensure that ongoing demands would not exceed capacity (Fig. 6). Figure 6. Science batteries performed as expected, showing gradual As noted in this figure, the batteries performed well but some decline, but the propulsion batteries exhibited some behavior that is under of the behavior remains under consideration because the analysis. voltage levels cannot be entirely explained by the loads placed on them. Because of the large area of pack ice above Beaufort Island (Fig. 2) and the fast ice bridge south of it, we elected to keep glider 502 in the McMurdo polynya for more than a week with the expectation that conditions would improve. On 12/5/2011, the ice area was beginning to thin so we aimed glider 502 north and then, on 12/10/2011, gave it a waypoint in the Ross Sea polynya, causing it to transit under/beneath the pack ice. As expected, contact was lost periodically during the 40km under ice transit but on 14 December, it emerged in the polynya. From there, we assigned it the task of making east-west transects of the polynya study area, making a total of 701 dives and covering 1342km by the time it was recovered on 1/20/2011. As seen in Fig. 1, these transects were not entirely linear due to some rather impressive tidal currents encountered at the eastern terminus of the transect. When glider 503 emerged on 11/30/2010, it reported that it had completed 23 dives to depths up to 415m but the closest it was able to get to the surface was about 40m depth. Figure 7. . Temperature plots from dive 20 illustrate glider 503’s Most of the dives ended at about 90m when “no vertical excursion under the Ross Ice Shelf and its contact with water at sub- velocity” was detected, indicating that it had been swept freezing temperatures. under the Ross Ice Shelf. Also, during several of these dives, temperatures as low as -1.96 C (Fig. 7) were measured covered with a very thin layer of “grease” ice. We elected to which is below the surface freezing point of seawater (-1.90 deploy the glider in spite of this minimal ice cover but used a C) at the measured salinity, and thus is evidence of water that small ROV (VideoRay Pro-4) to tow the glider as far as had been in contact with ice well below the ocean surface possible from the ice edge to minimize the likelihood that it due to the depression of the freezing point of seawater with would be drawn under the ice by prevailing currents (Fig. 3). pressure (~ 7.53e-4 deg C/decibar). We have no way of The glider left the surface at around midnight on 11/22/2010 knowing how far under this floating glacier the glider was and was not able to reach the surface again for nearly 36 carried but we were impressed that it was finally able to hours. overcome the currents and emerge from this situation. Once The second glider was deployed at Cape Crozier on the clear of the ice, we assigned glider 503 the task of making an SW edge of the Ross Sea polynya just before midnight on hourglass shaped survey over a focal point at 76° 30'S, 11/29/2010. The site was selected because of its proximity 176°E. This glider covered 1,671 km in 923 dives during the 63 days it was in the water. Figure 8. Results from Glider 502 illustrate the stark contrast in chlorophyll fluoresence between the waters to the east (a) and west (b) of Beaufort Island.

signal, this is more likely to be due to advective processes B. Data but the final interpretation will be based on model results as In spite of the severe conditions, the gliders functioned well as a thorough examination of these and other data. well, and, with the exception of the optical sensor pack on 503 which failed after 55 dives, produced excellent data. A IV. FUTURE PLANS detailed discussion of the findings is beyond the scope of this paper and will be presented elsewhere, along with the results This project demonstrated the utility of using of models that will help interpret the data. However, a few autonomous gliders to study biogeochemical processes in highlights are readily apparent. First and as expected, there environments as extreme as the Ross Sea. It is clear that was a large amount of spatial variation in the amount of navigation by magnetic compass with GPS fixes can be chlorophyll fluorescence in the water. As shown in Fig. 8, sufficient for the vehicles to complete assigned tracks. As the water on the western side of Beaufort Island (early in the expected, however, logistic considerations for deployment transect) was almost devoid of any signal while that in the and recovery will be expected to limit glider operations. Ross Sea polynya (last part of the transect) exhibited large This includes the challenge of transporting the equipment signals. Second, temperatures in the surface waters warmed and team to the ice edge, arranging for a vessel to retrieve the dramatically during the duration of the study, creating glider at the end of the mission, safety issues associated with considerable stratification in the polynya (Fig. 9). This, in all operations near this dangerously cold water, and the turn, stabilizes the water column and would be expected to impacts of weather and ice conditions on all activities. As enhance primary productivity. Finally, the salinity signal in glider technology continues to improve, however, we can the water column exhibits surprising spatial variations all expect to see enhancements in autonomous navigation, through the water column. In contrast to the temperature perhaps associated with long baseline acoustics, that will

Figure 9 Data from glider 503 show the development of a warm, somewhat less saline layer at the surface as the season progressed from 12/6 (a) to 1/28 (b) allow longer transects beneath the ice. We are pleased to be [6] Garrison, D.L., Gibson, A., Kunze, H., Gowing, M.M., Vickers, C.L., among the first to deploy gliders at these latitudes and we Mathot, S., Bayre, R.C., 2003. The Ross Sea Polynya Project: diatom- and Phaeocystis-dominated phytoplankton assemblages in the Ross look forward to seeing many others emulate this success so Sea, Antarctica, 1994 and 1995. In: DiTullio, G.,Dunbar, R. (Eds.), that a fleet of gliders will eventually patrol these forbidding Biogeochemistry of the Ross Sea. Ant. Res. Ser., 78. regions. AmericanGeophysical Union, Washington, DC, pp. 279–293. [7] Peloquin, J.A., Smith Jr., W.O., 2007. Phytoplankton blooms in the Ross Sea, Antarctica:interannual variability in magnitude, temporal CKNOWLEDGMENT A patterns, and composition. J. Geophys. Res. 112, C08013. We would like to thank the staff of Raytheon Polar doi:10.1029/2006JC003816 Services for their assistance in launching the gliders and the [8] Smith Jr., W.O., Marra, J., Hiscock, M.R., Barber, R.T., 2000. The officers and crew of the Nathaniel B. Palmer for their seasonal cycle of phytoplankton biomass and primary productivity in assistance in recovering the gliders. the Ross Sea, Antarctica. Deep-Sea Res. II 47, 3119–3140. [9] Smith Jr., W.O., Shields, A.R., Peloquin, J.A., Catalano, G., Tozzi, S., Dinniman, M.S., Asper, V.A., 2006. Biogeochemical budgets in the REFERENCES Ross Sea: variations among years. Deep-Sea Res. II 53, 815–833. [10] Smith Jr., W.O., Asper, V.L., 2001. The influence of phytoplankton assemblage composition on biogeochemical characteristics and cycles [1] Arrigo, K.R., and G. van Dijken (2003), Phytoplankton dynamics in the southern Ross Sea, Antarctica. Deep-Sea Res. I 48, 137–161. within 37 Antarctic coastal polynya systems, Geophys. Res. Letters, 108, doi:10.1029/2002JC001739. [11] Smith Jr., W.O., Comiso, J.C., 2008. The influence of sea ice on primary production in the Southern Ocean: a satellite perspective. J. [2] Arrigo, K. R., D.H. Robinson, D. L. Worthen, R. B. Dunbar,G. R. Geophys. Res. 113, C05S93. doi:10.1029/2007JC004251. DiTullio, M. van Woert andM.P. Lizotte. 1999. Phytoplankton community structure and the drawdown of nutrients andCO2 in the [12] Smith Jr., W.O., Gordon, L.I., 1997. Hyperproductivity of the Ross Southern Ocean. Science 283: 365-367.; Sea (Antarctica)polynya during austral spring. Geophys. Res. Lett. 24, 233–236. [3] Barber, D.G., Massom, R., 2007. The role of sea ice in Arctic and Antarctic polynyas. In:Smith Jr., W.O., Barber, D.G. (Eds.), [13] Smith Jr., W.O., Nelson, D.M., 1985. Phytoplankton bloom produced Polynyas: Windows into Polar Oceans. SanDiego, Elsevier, pp. 1–54. by a recedingice edge in the Ross Sea: spatial coherence with the density field. Science 227,163–166. [4] DiTullio, G.R., Smith Jr., W.O., 1996. Spatial patterns in phytoplankton biomass and pigment distributions in the Ross Sea. J. [14] Tremblay, J.-E., Smith Jr., W.O., 2007. Phytoplankton processes in Geophys. Res. 101, 18467–18478. polynyas. In: Smith Jr.,W.O., Barber, D.G. (Eds.), Polynyas: Windows to the World's Oceans. Amsterdam, Elsevier, pp. 239–270. [5] El-Sayed, S.Z., Biggs, D.C., Holm-Hansen, O., 1983. Phytoplankton standing crop, primary productivity, and near surface nitrogenous nutrient fields. Deep-Sea Res. 30, 871-886.