OceTHE OFFICIALa MAGAZINEn ogOF THE OCEANOGRAPHYra SOCIETYphy

CITATION Lee, C.M., J. Thomson, and the Marginal Ice Zone and Arctic Sea State Teams. 2017. An autonomous approach to observing the seasonal ice zone in the western Arctic. Oceanography 30(2):56–68, https://doi.org/10.5670/oceanog.2017.222.

DOI https://doi.org/10.5670/oceanog.2017.222

COPYRIGHT This article has been published in Oceanography, Volume 30, Number 2, a quarterly journal of The Oceanography Society. Copyright 2017 by The Oceanography Society. All rights reserved.

USAGE Permission is granted to copy this article for use in teaching and research. Republication, systematic reproduction, or collective redistribution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of The Oceanography Society. Send all correspondence to: [email protected] or The Oceanography Society, PO Box 1931, Rockville, MD 20849-1931, USA.

DOWNLOADED FROM HTTP://TOS.ORG/OCEANOGRAPHY SPECIAL ISSUE ON AUTONOMOUS AND LAGRANGIAN PLATFORMS AND SENSORS (ALPS)

An Autonomous Approach to Observing the Seasonal Ice Zone in the Western Arctic

By Craig M. Lee, Jim Thomson, and the Marginal Ice Zone and Arctic Sea State Teams

56 Oceanography | Vol.30, No.2 ABSTRACT. The Marginal Ice Zone and Arctic Sea State programs examined the MIZ is a region of complex atmosphere- processes that govern evolution of the rapidly changing seasonal ice zone in the Beaufort ice-ocean dynamics that varies with sea Sea. Autonomous platforms operating from the ice and within the water column ice properties and distance from the ice collected measurements across the atmosphere-ice-ocean system and provided the edge (Figure 2; e.g., Morison et al., 1987). persistence to sample continuously through the springtime retreat and autumn advance Additionally, the northward retreat of of sea ice. Autonomous platforms also allowed operational modalities that reduced the sea ice exposes an increasing expanse of field programs’ logistical requirements. Observations indicate that thermodynamics, open water south of the ice edge, eventu- especially the radiative balances of the ice-albedo feedback, govern the seasonal cycle ally providing sufficient fetch for the gen- of sea ice, with the role of surface waves confined to specific events. Continuous eration of long-period, large-amplitude sampling from winter into autumn also reveals the imprint of winter ice conditions and waves (e.g., Thomson and Rogers, 2014). fracturing on summertime floe size distribution. These programs demonstrate effective Such waves are capable of propagating use of integrated systems of autonomous platforms for persistent, multiscale Arctic north and penetrating into the pack to observing. Networks of autonomous systems are well suited to capturing the vast scales effect mechanical breakup of floes, greatly of variability inherent in the Arctic system. accelerating melt. Accurate characteriza- tion of MIZ processes becomes increas- INTRODUCTION dramatic sea ice decline has occurred in ingly important as the Beaufort MIZ Dramatic changes in summertime Arctic the Beaufort Sea and the Canada Basin gains prominence. sea ice motivated two process studies that (Figure 1a; Shimada et al., 2006), result- In addition to the increasing areal relied on recent advances in autonomous ing in the loss of thick, multiyear ice extent of open water, the duration of the observing to collect atmosphere, ice, and (e.g., Maslanik et al., 2007) and the north- open water season is also increasing. This ocean measurements across the necessary ward retreat of the summertime ice edge is particularly notable in the autumn; span of temporal and spatial scales. The from the shelf into the deep basin. the freeze-up in much of the Beaufort Arctic is warming at over twice the rate Climate models have successfully cap- and Chukchi Sea region is now one full observed at lower latitudes (Overland tured the overall trend in summertime sea month delayed from historic timing et al., 2016), with pronounced impacts ice extent; however, they under-predict (Thomson et al., 2016). This delay pro- on the timing and extent of sea ice. Arctic the observed rate of decline (Figure 1c; vides more opportunity for the ocean Ocean sea ice follows a seasonal cycle dic- Jeffries et al., 2013). Observed summer- to receive heat directly from solar radi- tated by incoming solar radiation, with time minimum sea ice extent varies sig- ation, as well as more opportunity for sea ice advancing southward in autumn, nificantly year to year, but the underlying autumn storms to mix that heat far below as insolation drops with the approaching rapid decline is just within the one stan- the canonical depths of the near-surface Arctic night, and retreating northward in dard deviation bound of predictions gen- temperature maximum (Jackson et al., spring as insolation increases. This sea- erated by an ensemble of climate models. 2010). The delay in ice formation thus sonality has changed in recent decades, This suggests that simulations may fail feeds back to affect the type of ice that with a trend toward greater ice retreat to properly represent the processes and is formed and the persistence of that ice each summer and a smaller, but signif- feedbacks that govern sea ice evolution. into the next seasonal cycle. For example, icant, trend of decreasing wintertime Limited understanding of the processes increasing surface wave activity is more maximum sea ice extent. The timing of that govern sea ice evolution in the mar- likely to produce pancake ice, which until these extrema has also shifted (Perovich ginal ice zone (MIZ) may contribute to the recently was rarely observed in the Arctic et al., 2016). Summertime minimum inability of models to reproduce the steep (Thomson et al., 2017). The formation of Arctic sea ice extent has been in decline decline in sea ice. Summertime open- pancake ice can actually accelerate the for nearly 40 years (Perovich et al., 2012), ing of the Beaufort and Chukchi Seas has autumn ice advance, but the ability to which, along with a reduction in thick- amplified the extent and influence of the forecast this process is extremely limited. ness, has led to an overall decrease in sea seasonal MIZ, the region of fractional ice Rapid changes in sea ice can have pro- ice volume (e.g., Kwok and Rothrock, cover that forms the transition between found impacts on human subsistence and 2009; Schweiger et al., 2011). The most open water and pack ice (Figure 1a). The commercial activities. Arctic ecosystems

Oceanography | June 2017 57 responding to changing sea ice affect the persistent sampling over many months that govern sea ice evolution during the timing and availability of subsistence while resolving scales of kilometers and highly dynamic period that spans melt- hunting. Declining sea ice has also gener- hours. Relative to conventional sampling out to freeze-up. Alongside their scientific ated increased interest in activities such as (Figure 3), autonomous observations can objectives, these programs also focused transpolar shipping, resource extraction, span seasonal cycles with much greater on advancing methodologies for autono- and tourism, with implications for safety temporal coverage. mous observing, developing and demon- and national security. Coastal commu- Motivated by an overall need to improve strating new approaches for conduct- nities suffer from accelerated erosion as predictability in the western Arctic on both ing sustained observations in ice-covered sea ice retreats northward, leaving shore- operational and climate time scales, the environments. Here, we report on the lines exposed to increased summertime US Office of Naval Research (ONR) sup- approaches and early combined findings wave activity. Improved sea ice predic- ported two large, integrated observational of both programs. tions are needed to inform planning and programs focused on the Beaufort Sea. In formulate responses to these challenging 2014, the Marginal Ice Zone (http://apl. AUTONOMOUS PLATFORMS developments. uw.edu/miz) program employed a large AND EXPERIMENT DESIGN Recent advances in autonomous plat- array of autonomous platforms to study The dynamics associated with summer- forms are providing new perspectives on the seasonal ice retreat. In 2015, the Sea time sea ice retreat and autumn advance the processes that govern Arctic sea ice State (http://apl.uw.edu/arcticseastate) pose significant observational challenges. evolution because they capture spatial and program used a research vessel and auton- Coincident measurements of the atmo- temporal scales that previously had been omous platforms to study the seasonal ice sphere, sea ice, and ocean must resolve challenging to sample. Robotic instru- advance. Although conducted in different a broad range of spatial and temporal ments that operate from the sea ice and years, taken together the two programs scales to document how the balance of in the water column have demonstrated provide a novel picture of the processes processes evolves in response to chang- ing surface forcing, ice cover, and upper- ocean stratification. During the transition a September 1980 b September 2007 periods, rapid shifts in sea ice extent and properties demand mobile approaches that are capable of tracking the ice edge as it moves. Continuous measurements that span months to years are needed to illu- minate the feedbacks that unfold at longer time scales, such as the sequestration of summertime solar heating and its influ- ence on the timing and spatial variability of freeze-up (Timmermans et al., 2014). Previous and ongoing Arctic research programs, such as the Beaufort Gyre 12 c NSIDC Observing System (e.g., Proshuntinsky Mean 10 Median et al., 2009a, 2009b) and the International ±1 SD Arctic Buoy Programme (http://iabp.apl. 8 washington.edu), have demonstrated the 6 value of autonomous sampling in this complex region. 4 The seasonal distribution of ice thick-

Sea-Ice Extent (10 km) 2 ness measurements in the western Arctic reflects the challenges of making mea- 0 surements in this difficult environment 1900 1920 1940 1960 1980 2000 2020 2040 2060 2080 2100 Year (Figure 3). The two largest concentrations FIGURE 1. September sea ice cover in 1980 (a) and in 2007 (b). The red outline shows the differ- of measurements center on late sum- ence in ice coverage, which is most notable in the western Arctic. Credit: NSIDC (c) Time series of mer, when maximal open water offers the sea ice area at the September minimum each year. The black line is a satellite data product from the best access for ships, and spring, the US National Snow and Ice Data Center. The yellow and blue lines mark the mean and median of September sea ice minimum predictions from an ensemble of climate models, with gray shading when the ice pack and operating condi- marking the one standard deviation bounds. From Jeffries et al. (2013) tions allow researchers to access the ice

58 Oceanography | Vol.30, No.2 using aircraft. During the transition peri- and lightweight logistics provided by sustain continuous sampling that extends ods, which embody the most dramatic autonomous platforms. Used in concert, from before the start of melt-out to the changes, observations have been fewer ice-based platforms along with mobile period of freeze-up and ice advance. This due to the challenges associated with col- instruments operating independently of sustained, highly resolved four-dimen- lecting measurements when both air- the ice can collect collocated time series sional sampling captures a region of the craft and ships struggle to provide access. of the atmosphere, ice, and ocean with spatial-temporal spectrum that previously Near-complete ice cover and inhospitable an expansive spatial footprint, resolving had been difficult to access, but is required operating conditions have limited win- scales of kilometers and hours. Robotic to quantify the multiscale processes and tertime data collection to only a handful instruments possess the endurance to feedbacks governing sea ice. of observations. The presence of sea ice presents chal- lenges beyond those associated with RETREAT logistics. Modern autonomous plat- forms rely on satellite services for geo- location (GPS) and telemetry (Iridium). Ice Divergence Storms Instruments mounted on the sea ice uti- and Convergence lize these services, but those that operate independently of the ice, such as floats Wind Surface Waves and gliders, must rely on arrays of acous- tic beacons for geolocation, and they can Melt Ponds Basal and communicate only irregularly through Lateral Melt Basal and leads and other ice-free areas. The tran- Lateral Melt sition periods (melt-out and freeze-up) Mixing present challenges to instruments that Entrainment rely on ice for a platform, making con- (Warming) tinuous operation in the seasonal ice Mixed Layer zone difficult. Significant effort has been invested to engineer platforms capa- ble of surviving these transitions. Lastly, the hazards of ice rafting, severe weather, extreme cold, and wildlife can take a toll on untended sensors mounted on ADVANCE the sea ice. Marginal Ice Zone and Sea State pro- Storms gram science objectives required contin- Wind uous, high-resolution observations of the Heat Loss Heat Loss atmosphere, sea ice, and upper ocean that Qnet < 0 Qnet < 0 Cooling Cooling Surface spanned the two sparsely sampled transi- Waves tion periods. To meet this challenge, both programs integrated the assets, interests, Mechanical Breakup and expertise of large international teams of investigators and drew upon autono- mous approaches. Such broad collabora- Entrainment tion reflects both the scope of the science (Warming) Mixed Layer and the need for a multifaceted observa- tional approach.

Observational Approach and Experiment Design The Marginal Ice Zone and Sea State pro- FIGURE 2. Atmosphere, ice, and upper-ocean processes governing (a) summertime sea ice grams adopted measurement strategies retreat and (b) autumn sea ice advance. Image credit: Kim Reading, Applied Physics Lab, that capitalize on the persistence, mobility, University of Washington

Oceanography | June 2017 59 The intrinsic nature of ice-based plat- providing the opportunity for widespread of autonomous platforms to sample each forms places sampling strategies into deployments of numerous, lightweight weather event, and then adapted the a Lagrangian framework: an ice-based ice-based robotic instruments. The MIZ plans as the event unfolded. As a result, instrument drifts with its host floe, thus program thus adopted an approach that the autonomous deployments during Sea providing an excellent reference frame for relied primarily on autonomous plat- State were generally much shorter than documenting its evolution. For observing forms deployed by aircraft. This strat- those in the MIZ experiment. The focus ocean variability, this framework resem- egy allowed autonomous platforms to in Sea State was more specifically on bles sampling from a slow-moving ship. begin sampling in spring, well before the using autonomy for dense spatial cover- Spatial sampling of the upper ocean start of melt, and operate continuously age, rather than for temporal endurance. around the drifting assets is thus required through the autumn freeze-up. to fully resolve the processes that govern In contrast, the Sea State program Components of an Integrated interactions between the upper ocean and focused on the period of advancing sea Atmosphere-Ice-Ocean Observing the sea ice. On a larger scale, the ability to ice, and thus began sampling around the System deploy many autonomous instruments, time of maximum open water extent. This The Marginal Ice Zone and Sea State along with gliders, autonomous surface allowed broad access for ship-based sam- programs employed observational sys- and underwater vehicles, and ships that pling, but limited the utility of ice-based tems (Figure 4) composed of autono- provide mobility, allowed sampling to instruments. Sea State thus adopted an mous platforms, ships, and moorings maintain focus on the rapidly translat- observing strategy centered on a heavily selected for their complementary capa- ing MIZ—the transition region between instrumented research vessel, with auton- bilities. Multiple platforms operating in open water and pack ice thought to have omous assets used to expand the foot- concert were typically required to span the greatest dynamic variability. print of shipboard operations and to pro- the range of variables (atmosphere, ice, MIZ experimental design exploited vide measurements (e.g., ice draft) that ocean) and scales required to resolve tar- operational modalities provided by could not be made from the ship. get processes. All platforms except moor- autonomous platforms to overcome the With a ship as the central asset during ings featured two-way telemetry, allowing logistical constraints imposed by sea the Sea State experiment, it was possible them to upload their data and download ice. Expansive ice cover at the start of to target specific events and conditions new commands. This setup mitigated risk the sampling program made ship-based in real time. R/V Sikuliaq operated with by ensuring data return regardless of the sampling problematic, but favored air- a rolling three-day plan, updated daily in fate of the platforms, and allowed the sci- craft operations. Fixed-wing and rotary response to weather forecasts and satellite ence teams to adjust sampling plans in aircraft can land on the springtime ice remote sensing. The team aboard devel- response to observed variability. to deliver personnel and equipment, oped plans to deploy and maintain arrays ICE-BASED DRIFTING AUTONOMOUS PLATFORMS 25,000 Autonomous instruments were deployed 22,079 on sea ice, and thus sampled in a drift-

20,000 ing reference frame that allowed their 18,071 measurements to document the evolu- tion of the host floes. Systems of com- 15,000 plementary ice-based platforms cur- rently provide the only approach capable

10,000 of collecting contemporaneous, collo- cated measurements of the atmosphere, 6,868 Number of Observations sea ice, and upper ocean. Ice-based plat- 4,929 5,000 4,517 forms that depend entirely on ice for flo- 2,662 2,022 tation do not survive melt-out, while oth- 226 334 675 555 481 ers, such as Ice-Tethered Profilers, that 0 have been engineered to survive melt- April May June July March August out and freeze-up can span these transi- January February October September November December tions, albeit with increased risk. During FIGURE 3. Number of sea ice thickness measurements, by month, in the western Arctic. These the MIZ and Sea State programs, these include on-ice (~in situ) sea ice thickness measurements taken by augers, cores, and surface elec- tromagnetic radiation compiled from field experiments conducted from the 1890s (Fram) through systems were deployed in regions of rel- 2011. Source: Benjamin Holt, NASA/JPL­Caltech atively solid ice cover.

60 Oceanography | Vol.30, No.2 IMB SWIFT (snow and ice) (surface waves) Wave Buoy (waves in ice)

Seaglider (upper ocean, surveys)

ITP-V (upper ocean, profiles) UAS (ice mapping) AOFB (upper ocean, mixing)

AWS (meteorological)

Wave Buoy (waves in ice) Acoustic Navigation Buoy (underwater geolocation)

AUV (under ice)

FIGURE 4. Autonomous platforms used in the field programs. SWIFT = Surface Wave Instrument Float with Tracking drifter. IMB = Ice Mass Balance Buoy. AWS = Autonomous Weather Station. ITP-V = Ice-Tethered Profiler with Velocity. UAS = Unmanned Aerial System. AOFB = Autonomous Ocean Flux Buoy. AUV = Autonomous Underwater Vehicle.

Ice Mass Balance Buoys (IMBs; Jackson Ice-Tethered Profilers (ITPs; Krishfield (using a three-dimensional sonic aneo- et al., 2013) measured profiles of tem- et al., 2008) provided high spatial (1 m) mometer) at 2 m above the ice, tempera- perature and thermal response at finely and temporal resolution profiles of tem- ture from the ice surface to 4.5 m depth spaced vertical intervals through the perature, salinity, and velocity (Cole et al., (using a thermistor string), and surface air, snow, ice, and upper water column. 2015) from near the ice-ocean interface to waves (Gallaher et al., 2017). Thermal characteristics were inter- 250 m depth (at three-hour intervals) and preted to identify the interfaces between 750 m depth (once per day), and direct Automated Weather Stations (AWSs) air, snow, ice, and water, and to quan- estimates of the turbulent vertical fluxes composed of commercial, off-the-shelf tify surface and basal sea ice melt rates of heat, salt, and momentum within the instruments measured wind velocity, (Polashenski et al., 2011). ocean mixed layer (Cole et al., 2015). humidity, air temperature, surface pres- sure, solar radiation, and floe rotation. Wave Buoys (Doble et al., 2017) installed Autonomous Ocean Flux Buoys (AOFBs) Untended atmospheric measurements on the ice used accelerometers and tilt- measured profiles of mixed layer cur- are extremely difficult to sustain in the meters to measure spectral surface wave rents; vertical turbulent fluxes of heat, Arctic, and data return from these sta- properties. These instruments quantified salt, and momentum near the top of the tions was limited. surface wave activity that penetrated into ocean mixed layer; shortwave radiative ice-covered waters. fluxes; and air-ice momentum transfer

Oceanography | June 2017 61 Acoustic Navigation Buoys provided different regimes and providing spatial REMOTE SENSING geolocation for autonomous platforms context for drifting assets. Both programs included extensive collec- operating beneath the ice. A network tion of synthetic aperture radar (3 m to of 900 Hz, broadband sound sources Jaguar Autonomous Underwater 50 m resolution) and National Technical employed highly accurate clocks to Vehicle (Woods Hole Oceanographic Means (visible, 1 m resolution) sea ice broadcast on a fixed schedule, with each Institution, Deep Submergence Lab) was imagery targeted on the in situ assets element transmitting six times per day used to conduct under-ice surveys, com- (http://www.apl.washington.edu/miz). (Freitag et al., 2015). Autonomous plat- plementing observations collected by the Time series of open water fraction and forms calculated range from each source ship. Measurements included ice draft floe size distribution, critical for interpret- based on travel time and source loca- as well as upper-ocean properties. The ing the in situ observations, were derived tion, which was transmitted as part resulting ice draft maps are merged with from the collection of images. Real-time of the signal. Position was then esti- aerial surveys to create maps and distri- remotely sensed sea ice extent was also mated through multilateration from butions of ice thickness. used for operational decision-making for multiple receptions. ships and mobile autonomous platforms. Autonomous Surface Vehicles (Wave The Sea State program also employed OCEAN-BASED DRIFTING Gliders; e.g., Lenain and Melville, 2014) local remote sensing, including ship- AUTONOMOUS PLATFORMS maintained a persistent presence in the board radar, video, and LiDAR, as well as Surface-drifting platforms targeted open open water immediately south of the video from UAS. water and the MIZ, collecting comple- MIZ, collecting meteorological and mentary measurements in the regimes near-surface measurements just south of SITUATIONAL AWARENESS AND that were challenging for ice-based the ice edge. DECISION-MAKING instrumentation. Real-time situational awareness was Unmanned Aerial Systems (UAS) such required to target satellite image acqui- Surface Wave Instrument Float with as the quad-rotor (DJI Phantom 3, a sitions on drifting ice-based assets and Tracking (SWIFT) Drifters (Thomson, commercial, off-the-shelf system) and mobile vehicles during the Marginal 2012) characterized surface waves and fixed wing systems were flown to map Ice Zone program, and to inform ship- turbulence in both open water and partial ice and surface conditions near the ship based surveys conducted during the Sea ice cover. SWIFTs measured wind speed, and autonomous in situ arrays during State program. During the MIZ program, wave height, wave directional spectra, air the Sea State campaign. Video and still real-time platform position data were temperature, sea surface temperature, images collected during each flight used to model drift and predict instru- surface currents, and dissipation by tur- were post-processed for georectifica- ment positions at the time of proposed bulence while drifting at the sea surface. tion and estimation of three-dimensional image collections. A dedicated targeting digital elevation models using struc- team then used these predictions to keep MOBILE AUTONOMOUS PLATFORMS ture-from-motion techniques. remote-sensing collections centered on Mobile platforms were directed to follow MIZ instruments. This facilitated acqui- the shifting ice edge to characterize spa- MOORINGS sition of time series of images that doc- tial variability around other drifting ele- The Marginal Ice Zone program aug- ument sea ice evolution around MIZ ments of the system. mented existing surface moorings, program assets. The Sea State program deployed as elements of other pro- relied on rapid processing and curation Seagliders (Eriksen et al., 2001) are grams, to obtain extended time series (to optimize use of R/V Sikuliaq’s limited long-endurance, buoyancy-driven under- of surface wave properties at sites in the Internet bandwidth) of satellite imagery water vehicles that profile between the Beaufort Sea. and weather forecasts to inform observa- surface (or ice-ocean interface) and tional strategies. A dedicated shore-based 1,000 m depth while moving at a horizon- SHIPS team ensured timely delivery of useful tal speed of ~0.25 m s–1. Gliders collected During Sea State, a shipboard suite of products to the seagoing effort. profiles of temperature, salinity, dissolved instruments on R/V Sikuliaq that was oxygen, chlorophyll fluorescence, opti- designed to survey the air-ice-ocean sys- Putting it All Together— cal backscatter, and spectral downwelling tem included an underway CTD sensor The MIZ and Sea State Programs irradiance. The mobility provided by glid- towed from the stern, a meteorological MIZ ers was used to collect high-resolution​ flux package mounted on the bow, and The MIZ sampling strategy featured a sections from open water, through the a Sea Ice Measurement System (SIMS) large array of ice-based instruments, MIZ, and into the pack, spanning the deployed on a boom over the side. including acoustic navigation sources to

62 Oceanography | Vol.30, No.2 support autonomous platforms working the course of a single week using a pair track and satellite images of the ice condi- under the ice (Figure 5). Aircraft were of Twin Otter fixed-wing aircraft and a tions as the autumn progressed. Although used to deploy the array in early spring Bell 412 helicopter operating out of Sachs the approach was more traditional (based 2014 along a 300 km-long line extending Harbour, Canada. Overnight stays on on a ship), arrays of autonomous plat- northward in the eastern Beaufort Sea. the ice involved minimal personnel sup- forms were central to the work, both as The array was designed to quantify ice ported by light, mountaineering-style a way to achieve greater spatial coverage and snow thickness, surface wave proper- camps, minimizing the equipment and and as a way to avoid the influence of the ties within the MIZ, sea ice deformation, fuel required to execute the deployments. ship in the measurements. upper-ocean water properties, currents Ship operations were limited to a deploy- A central science theme for the Sea and turbulence, and meteorological vari- ment cruise conducted from a small State expedition was the interaction of ables as a function of distance from the boat (R/V Ukpik), autumn recovery of surface waves and ice; this signal is intrin- ice edge. A 300 km-long span was cho- Seagliders and SWIFTs from a short cruise sically spatial, and thus distributed arrays sen to ensure continuous measurements aboard R/V Norseman 2, and instrument of autonomous platforms provided a dis- from the MIZ northward throughout the deployments and measurements far to tinct advantage over a single vessel in melt season, accounting for the expected the north conducted in collaboration observing this process. A total of seven westward drift, deformation, and melt- with KOPRI aboard IBRV Araon. wave-ice experiments were conducted out (and resulting instrument loss) from during the expedition, each including the the south. The actual drift involved con- SEA STATE deployment of up to 16 Wave Buoys and siderably more rotation than had been The Sea State field program consisted of a SWIFTs. The ship surveyed both during predicted from modeling and analy- 42-day expedition on R/V Sikuliaq in the and in between these wave experiments, ses of historical drift data, with the ice- autumn of 2015. Figure 7 shows the ship using the surface flux system to map the based array into an east-west orienta- tion as it swept westward through the Ice Edge (August) 5 km Beaufort Sea (Figure 6). Drifting and mobile platforms sampled 400 km 5 km Acoustic Navigation Source within the matrix of the ice-based array. A AOFB small boat (R/V Ukpik) operating out of AWAC Mooring AWS Prudhoe Bay was used to deploy drifting Ice Edge (July) 300 km IMB/Wave Buoy and mobile assets (SWIFTs, Seagliders, ITP-V and Wave Gliders) as soon as open water Polar Pro ling Float Seaglider Sections (2x) developed along the coast. SWIFTS and SWIFT Wave Gliders sampled the open water and 200 km Wave Glider the MIZ, while Seagliders occupied sec- Ice Edge (June) 100 km tions that spanned open water, the MIZ, Ice Pack ard and pack ice, providing spatial context to 100 km rthw ) –Sep bind measurements from the other assets. MIZ retreats(Jun no The MIZ program also included late summer sampling from the Korean Polar Research Institute (KOPRI) Ice Breaking Research Vessel (IBRV) Araon. Araon extended the northward reach of the Open Water Barrow array by deploying a cluster of ice-based platforms at 78°N and conducted inten- sive sampling of sea ice, melt ponds, and ALASKA Prudhoe biological and biogeochemical variability Bay N from a brief, collaborative, KOPRI-MIZ (Not to Scale) program ice camp staged from the ship. FIGURE 5. Idealized Marginal Ice Zone (MIZ) program observing array configuration. Note the Due to its focus on small, robotic markers indicating various instrument separations (drawing is not to scale). The northernmost clus- platforms, the MIZ program was able ter (400 km) was deployed by IBRV Araon. All other ice-based instruments were deployed using to operate with a relatively light logis- aircraft. Ice-based instruments melt out from the south as the MIZ retreats northward. Blue, dou- ble-ended arrows mark glider sections that follow the northward retreat of the sea ice to remain tical footprint. An array of over 60 ice- centered on the MIZ. Solid (dashed) light blue lines mark notional positions of the ice edge in June based instruments was deployed over (July and August) relative to the observing array. From Lee et al. (2012)

Oceanography | June 2017 63 cooling at the surface, and the underway mediates exchanges of heat and momen- influences of ocean heat and surface CTD to observe the ocean response. The tum between atmosphere and ocean that waves. The relative importance of these ship also supported 12 ice stations that increase as the sea ice retreats. In the processes appears to shift throughout collected detailed maps of ice floes from MIZ, the large range of sea ice coverage the seasonal cycle. Thermodynamics are above (with UAS) and from below (with and properties and the strong lateral gra- clearly dominant in the quiescent con- autonomous underwater vehicles), while dients associated with the transitions lead ditions of early spring. Once melting is more conventional thickness samples to large variations in processes that gov- well underway and the ocean is exposed (hand drills) were collected by personnel ern these exchanges. In addition, surface to direct forcing by the wind, the system on the ice. waves, incident from open water, propa- becomes much more energetic. Steele gate into the MIZ. et al. (2015) show the importance of A SEASONAL CHRONOLOGY Preliminary findings from these pro- chronology in the seasonal ice cycle, and OF ATMOSPHERE-ICE-OCEAN grams exploit the persistence pro- Zhang et al. (2016) report on ice floe size INTERACTIONS vided by autonomous sampling to con- distribution throughout this cycle. Here, Sea ice is part of a tightly coupled air-ice- firm the dominance of thermodynamics we provide a brief overview of the key ocean system where there are multiple and radiative balances in controlling processes targeted during the field cam- feedbacks and interdependencies. Sea ice ice evolution, and also point to strong paigns (Figure 2).

May 16, 2014 Jun 22, 2014 ITP 76°N AOFB 76°N AWS Wavebuoy IMB Navigation Source Seaglider 74°N 74°N SWIFT

72°N 72°N

Sachs Harbour Sachs Harbour

70°N 70°N

Wainwright Wainwright Prudhoe Bay Prudhoe Bay 160°W 150°W 140°W 130°W 160°W 150°W 140°W 130°W Aug 27, 2014 Sep 25, 2014 76°N 76°N

74°N 74°N

72°N 72°N

Sachs Harbour Sachs Harbour

70°N 70°N

Wainwright Wainwright Prudhoe Bay Prudhoe Bay 160°W 150°W 140°W 130°W 160°W 150°W 140°W 130°W

FIGURE 6. Snapshots of the progression of autonomous platform positions and RADARSAT-2 synthetic aperture radar ice images during the MIZ exper- iment. Light colors denote sea ice, while blacks and dark grays indicate open water. Colored markers indicate instrument positions at the time of the image. The May 16, 2014, image was taken roughly two months after MIZ deployment, before the array accelerated westward. By June 22, ice-based instruments had drifted westward and rotated into an east-west orientation. RADARSAT-2 Data and Product MacDonald, Dettwiler, and Associates Ltd., All Rights Reserved. Figure credit: Luc Rainville, Applied Physics Lab, University of Washington, and the National Ice Center

64 Oceanography | Vol.30, No.2 Ice Retreat Processes SURFACE MELT introduce heat into the upper ocean that Ice cover modulates penetration of solar In spring, increased local insolation drives is then available to drive basal and lateral radiation and isolates the upper ocean snow and surface ice melt. Melting causes melting. Buoyancy from melt pond drain- from direct wind forcing, but increas- ponds to form on the surface of the sea ice age also helps insulate the sea ice from ing open water within the MIZ and the (Figure 8a), decreasing surface albedo and heat stored below. Increasingly mobile proximity of large expanses of open water initiating a strong feedback mechanism sea ice and regions of open water provide immediately to the south permit more where reduced albedo increases absorp- more efficient transfer of momentum direct connection with the atmosphere. tion of solar radiation, driving more melt from the atmosphere to the upper ocean. Strong open water swell and surface wave and thus further reductions in albedo. This transfer of momentum can lead to activity attenuate as they enter the MIZ. Gallaher et al. (2017) use AOFB data to generation of near-inertial motions and Likewise, internal waves, submesoscale show that the eventual drainage of these eddies that can enhance mixing and lat- eddies, and mixing weaken with increas- melt ponds provides a significant source eral stirring, competing with the damp- ing ice cover. Small-scale wind stress curl of buoyant water to maintain strong ening effects of elevated stratification. associated with ice to open-water tran- near-surface stratification. This inhibits This represents another positive feed- sitions and variations in ice roughness vertical mixing and thus helps generate back, where ice melt leads to elevated may induce intense secondary circula- and preserve the near-surface tempera- mixing and more ice melt, modulated by tions that drive rapid vertical exchange. ture maximum throughout much of the the negative feedback of stabilizing melt Enhanced mixing and vertical exchange open water season. water. These processes can bring recently can entrain heat stored below the mixed warmed waters into contact with the base layer, increasing basal melting of sea ice BASAL AND LATERAL MELT of the remaining sea ice, driving basal within the MIZ. In ice-covered regions, As spring progresses, expansion of open melting. This basal melting often occurs local radiative solar warming leads to water areas and melt pond coverage simultaneously with the ice breaking up direct ablation of sea ice and some bot- tom melt from the radiation penetrating weakly into the ice-covered upper ocean. October 3, 2015 October 10, 2015 Open water regions within and out- side of the MIZ allow increased radiative upper-ocean warming and, through lat- eral advection, accelerated ice melt. These processes are expected to amplify vari- ance at short spatial and temporal scales across the MIZ.

WINTER CONDITIONING Persistent observations of the atmo- sphere-ice-ocean system provided by autonomous platforms, combined with October 20, 2015 October 27, 2015 time series of remotely sensed synthetic aperture radar and visible sea ice imag- ery, allowed Hwang et al. (2017) to docu- ment episodic, storm-driven, wintertime sea ice breakup events, and to identify the influence of wintertime sea ice com- position and fracturing on floe size dis- tribution in the subsequent summer. Illuminating the connections between winter sea ice conditions and summer evolution required the sustained observa- tions across winter-to-summer transition FIGURE 7. Ship track (red) and RADARSAT-2 synthetic aperture radar ice images during the Sea provided by autonomous platforms. State experiment. The Alaskan coast and Point Barrow are visible in the lower portion of each image. The ice advanced through the experiment. RADARSAT-2 Data and Product MacDonald, Dettwiler, and Associates Ltd., All Rights Reserved. Figure credit: Steve Roberts, University of Alaska Fairbanks, and the National Ice Center

Oceanography | June 2017 65 to form a marginal ice zone (Figure 8b). summer progresses because the open autumn, this heat is either removed (via Gallaher et al. (2016) use AOFB data water fetch available for wave generation mixing to the surface) or trapped (via sta- to show that ice divergence in the MIZ increases (Smith and Thomson, 2016). ble stratification). allows for enhanced ocean-to-ice turbu- This is another potential feedback mech- lent heat flux and increased basal melting anism, because the potential for waves to GREASE AND FRAZIL ICE as the spring season progresses. This MIZ enhance ice retreat will create more fetch As ice crystals form, they float near the is still thermodynamically driven, mod- for even larger waves to form. Wang et al. ocean surface in a slurry called frazil ice, ulated by the mechanics of winds and (2016) used remote sensing and autono- and are advected by winds and waves waves moving the sea ice. mous wave buoy data to show that floe (Lange et al., 1989). Zippel and Thomson size distribution eventually correlates (2016) use SWIFT data to show strong WAVES AND STORMS with surface wave conditions as the suppression of turbulence in the pres- As spring turns to summer, later stages season progresses. ence of grease ice (a very thin layer of fra- of ice retreat can be enhanced by storms, zil crystals clumped together at the sea which generate waves in open water that Ice Advance Processes surface, making it look like an oil slick), propagate into the MIZ. The storms also The ocean surface heat budget controls which may decouple the surface from the provide wind stress to drive ocean mix- ice formation. In the autumn, local inso- wind and wave forcing (Figure 8d). If the ing, which can enhance the basal melt- lation diminishes and cold air moves wind and the waves are strong enough, ing already occurring. Figure 8c shows over the ocean. As the surface cools, the the ice crystals coalesce and begin to an example of open water and remnant upper ocean also cools. There is a known form floes. The next stages of ice forma- brash ice (accumulations of floating ice near-surface temperature maximum, tion are thus distinct between calm and made up of fragments not more than which is formed by solar heating over storm conditions. 2 m across) in the MIZ near the north- open water areas each summer (Jackson ern coast of Alaska in August 2014. et al., 2010). The near-surface tempera- NILAS ICE This process can be challenging to cap- ture maximum layer can contain suffi- When heat loss is dramatic and condi- ture with observing platforms because cient heat to delay ice formation in the tions are calm, thin sheets of new ice, it is episodic and often localized. Wind autumn, provided there is enough mix- termed nilas, can rapidly cover large and wave processes become more likely ing to bring this heat to the surface where areas (Figure 8e). This ice type is most to dominate the ice retreat processes as the strong cooling is occurring. By late prevalent during periods of off-ice wind.

(a) Melt Ponds (b) Basal Melt (c) Brash and Open Water

(d) Grease and Frazil Ice (e) Nilas Ice Formation (f) Pancake Ice Formation

FIGURE 8. Surface images throughout the seasonal ice cycle. (a) Photo of spring melt ponds on the surface of the sea ice taken from an autonomous Wave Buoy during the MIZ 2104 deployment. (b) Photo looking upward from an autonomous underwater vehicle, with the lower hull of a SWIFT buoy vis- ible. (c) Open water and brash ice, with a SWIFT buoy. (d) Grease and frazil ice, with bands of open water. (e) Nilas ice sheets. (f) Pancake ice in waves, with a Wave Buoy in the distance. Photo Credits: (a) Jeremy Wilkinson and Martin Doble, (b) Ted Maksym, (c) Jim Thomson, (d) Steve Ackley, (e) Jim Thomson, (f) Martin Doble

66 Oceanography | Vol.30, No.2 Several nilas ice events were observed observatories to demonstrate that multi- oceanographic research. IEEE Journal of Oceanic Engineering 26:424–436, https://doi.org/10.1109/​ during the Sea State program, includ- scale Arctic observing can be done effec- 48.972073. ing one when a region approximately tively and successfully using coordinated Freitag, L., K. Ball, J. Partan, P. Koski, and S. Singh. 2 2015. Long range acoustic communications and 5,000 km was frozen in a single day. autonomous platforms. Ships and aircraft navigation in the Arctic. In Proceedings of the are still required as part of these opera- Oceans’15 MTS/IEEE Washington Conference. PANCAKE ICE October 19–22, 2015, Washington, DC, tions, but they need not be the central https://doi.org/10.23919/OCEANS.2015.7401956. Waves are responsible for the formation platforms for data collection. Distributed Gallaher, S.G., T.P. Stanton, W.J. Shaw, S.T. Cole, J.M. Toole, J.P. Wilkinson, T. Maksym, and of pancake ice, which occurs when frazil networks of autonomous platforms are B. Hwang. 2016. Evolution of a Canada Basin ice is continually agitated by wave orbital far better suited to capturing the vast ice-ocean boundary layer and mixed layer across a developing thermodynamically forced motions (Figure 8f). During an October scales of variability present across the marginal ice zone. Journal of Geophysical 2015 storm observed on the Sea State Arctic system. Research 121:6,223–6,250, https://doi.org/​ 10.1002/2016JC011778. cruise, the ice edge retreated, but then As the Arctic continues to change, Gallaher, S.G., T.P. Stanton, W.J. Shaw, S.-H. Kang, rapidly advanced again immediately fol- observational approaches will be forced J.-H. Kim, and K.-H. Cho. 2017. Field observa- tions and results of a 1-D boundary layer model lowing the event, because the air tempera- to continue to adapt. It may be that for developing near-surface temperature max- tures were well below freezing. Rogers springtime ice camps supported by air- ima in the Western Arctic. Elementa: Science of the Anthropocene 5:11, https://doi.org/10.1525/ et al. (2016) use data from the array of 16 craft are no longer viable because of weak elementa.195. wave buoys and SWIFTs to explore the ice, or that ice-based autonomous assets Hwang, B., J. Wilkinson, E. Maksym, H.C. Graber, A. Schweiger, C. Horvat, D.K. Perovich, damping of waves by pancake ice during melt out long before intended end-of- A.E. Arntsen, T.P. Stanton, J. Ren, and P. Wadhams. this storm; the work also presents the mission. The answers to these challenges 2017. Winter-to-summer transition of Arctic sea ice breakup and floe size distribution in the Beaufort state-of-the-art in wave forecasting in the will no doubt be more autonomy, with Sea. Elementa: Science of the Anthropocene 5:40, presence of ice. Though the effects of this better synthesis of forecast and satellite https://doi.org/10.1525/elementa.232. Jackson, J.M., E.C. Carmack, F.A. McLaughlin, autumn event were dramatic, as the sea- products to help make decisions on the S.E. Allen, and R.G. Ingram. 2010. Identification, son continues, the storms have less and best use of assets in near-real time. The characterization, and change of the near­surface temperature maximum in the Canada Basin, 1993– less effect, because there is very little open next ONR-supported effort, Stratified 2008. Journal of Geophysical Research 115, water left to generate surface waves and Ocean Dynamics in the Arctic (SODA, C05021, https://doi.org/10.1029/2009JC005265. Jackson, K., J. Wilkinson, T. Maksym, D. Meldrum, little direct air-ocean heat exchange. www.apl.washington.edu/soda), will J. Beckers, C. Haas, and D. MacKenzie. study an entire annual cycle of the region 2013. A novel low-cost sea ice mass balance buoy. Journal of Atmospheric and Oceanic CONCLUSIONS and will continue to advance autonomous Technology 30:2,676–2,688, https://doi.org/10.1175/ The seasonal/marginal ice zone of the capabilities, including sustained under- JTECH-D-13-00058.1. Jeffries, M.O., J.E. Overland, and D.K. Perovich. western Arctic Ocean is expanding. Key ice observations with profiling floats and 2013. The Arctic shifts to a new normal. Physics processes coupling the atmosphere, ice, Seagliders throughout the winter. Today 66:35–40, https://doi.org/10.1063/PT.3.2147. Krishfield, R., J. Toole, and M.-L. Timmermans. and ocean have been observed in two Data analysis from the MIZ and Sea 2008. Automated ice-tethered profilers for sea- recent ONR-funded field programs: MIZ State programs is ongoing. Journal arti- water observations under pack ice in all sea- sons. Journal of Atmospheric and Oceanic (2014) and Sea State (2015). These pro- cles detailing the results are forth- Technology 25:2,019–2,105, https://doi.org/​ grams made extensive use of autono- coming in special issues of Elementa: 10.1175/2008JTECHO587.1. Kwok, R., and D.A. Rothrock. 2009. Decline in Arctic mous platforms to investigate the sea- Science of the Anthropocene and sea ice thickness from submarine and ICESat sonal retreat and advance of sea ice. the Journal of Geophysical Research – records: 1958–2008. Geophysical Research Letters 36, L15501, https://doi.org/10.1029/​ Relative to a starting hypothesis that sur- Oceans, respectively. 2009GL039035. face waves have an increasing role in the Lange, M.A., S.F. Ackley, P. Wadhams, REFERENCES G.S. Dieckmann, and H. Eicken. 1989. Development seasonal cycle of the western Arctic, we Cole, S.T., F.T. Thwaites, R.A. Krishfield, and J.M. Toole. of sea ice in the Weddell Sea, Antarctica. Annals find that this effect is, at present, con- 2015. Processing of velocity observations from Ice- of Glaciology 12:92–96, https://doi.org/10.1017/ Tethered Profilers. In Proceedings of the Oceans’15 S0260305500007023. fined to specific events in late summer MTS/IEEE Washington Conference. October 19–22, Lee, C.M., S. Cole, M. Doble, L. Freitag, P. Hwang, and early autumn. Much of the seasonal 2015, Washington, DC, https://doi.org/10.23919/ S. Jayne, M. Jeffries, R. Krishfield, T. Maksym, OCEANS.2015.7401887. W. Maslowski, and others. 2012. Marginal Ice Zone cycle appears to be controlled by ther- Doble, M.J., J.P. Wilkinson, L. Valcic, J. Robst, A. Tait, (MIZ) Program: Science and Experiment Plan. modynamics, especially surface radi- M. Preston, J.-R. Bidlot, B. Hwang, T. Maksym, and Technical Report APL-UW 1201, Applied Physics P. Wadhams. 2017. Robust wavebuoys for the mar- Laboratory, University of Washington, Seattle, ative balances and the well-known ginal ice zone: Experiences from a large persistent September 2012, 48 pp. ice-albedo feedback. array in the Beaufort Sea. Elementa: Science of Lenain, L., and W.K. Melville. 2014. Autonomous the Anthropocene 5:47, https://doi.org/10.1525/ surface vehicle measurements of the Observing the many and coupled pro- elementa.233. ocean’s response to Tropical Cyclone cesses of sea ice evolution requires per- Eriksen, C.C., T.J. Osse, R.D. Light, T. Wen, Freda. Journal of Atmospheric and Oceanic T.W. Lehman, P.L. Sabin, J.W. Ballard, Technology 31:2,169–2,190, https://doi.org/10.1175/ sistence, especially for event-driven pro- and A.M. Chodi. 2001. Seaglider: A long- JTECH-D-14-00012.1. cesses. These programs built on previous range autonomous underwater vehicle for deployments of autonomous ice-based

Oceanography | June 2017 67 Maslanik, J.A., C. Fowler, J. Stroeve, S. Drobot, Thomson, J., Y. Fan, S. Stammerjohn, J. Stopa, Meylan, Fabien Montiel, William Perrie, Ola Persson, H.J. Zwally, D. Yi, and W.J. Emery. 2007. A younger, W.E. Rogers, F. Girard-Ardhuin, F. Ardhuin, H. Shen, Erick Rogers, Hayley Shen, Hui Shen, Vernon Squire, thinner Arctic ice cover: Increased potential W. Perrie, H. Shen, and others. 2016. Emerging Sharon Stammerjohn, Justin Stopa, Peter Sutherland, for rapid, extensive sea ice loss. Geophysical trends in the sea state of the Beaufort and Chukchi Jim Thomson, Peter Wadhams, and Timothy Williams. Research Letters 34, L24501, https://doi.org/​ Seas. Ocean Modelling 105:1–12, https://doi.org/​ 10.1029/​2007GL032043. 10.1016/j.ocemod.2016.02.009. ARTICLE CITATION Morison, J.H., M.G. McPhee, and G.A. Maykut. 1987. Thomson, J., and E. Rogers. 2014. Swell and sea Lee, C.M., J. Thomson, and the Marginal Ice Zone Boundary-layer, upper ocean, and ice observa- in the emerging Arctic Ocean. Geophysical and Arctic Sea State Teams. 2017. An autonomous tions in the Greenland Sea marginal ice-zone. Research Letters 41:3,136–3,140, https://doi.org/​ approach to observing the seasonal ice zone in Journal of Geophysical Research 92:6,987–7,011, 10.1002/2014GL059983. the western Arctic. Oceanography 30(2):56–68, https://doi.org/10.1029/JC092iC07p06987. Timmermans, M.-L., A. Proshutinsky, E. Golubeva, https://doi.org/10.5670/oceanog.2017.222. Overland, J., E. Hanna, I. Hanssen-Bauer, S.-J. Kim, J.M. Jackson, R. Krishfield, M. McCall, G. Platov, J. Walsh, M. Wang, U.S. Bhatt, and R.L. Thoman. J. Toole, W. Williams, T. Kikuchi, and S. Nishino. 2016. Surface air temperature. State of the Climate 2014. Mechanisms of Pacific Summer Water vari- in 2015, Bulletin of the American Meteorological ability in the Arctic’s Central Canada Basin. Society 97:S132–S134 Journal of Geophysical Research 119:7,523–75,48, Perovich, D.K., W. Meier, J. Maslanik, and J. Richter- https://doi.org/10.1002/2014JC010273. Menge. 2012. Sea ice cover. State of the Climate Wang, Y., B. Holt, W.E. Rogers, J. Thomson, and in 2011, Bulletin of the American Meteorological H.H. Shen. 2016. Wind and wave influences on sea Society 93:S140–S142. ice floe size and leads in the Beaufort and Chukchi Perovich, D., W. Meier, M. Tschudi, S. Farrell, Seas during the summer-fall transition 2014. S. Gerland, and S. Hendricks. 2016. Sea ice Journal of Geophysical Research 121:1,502–1,525, cover. State of the Climate in 2015, Bulletin of the https://doi.org/10.1002/2015JC011349. American Meteorological Society 97:S134–S135. Zhang, J., H. Stern, B. Hwang, A. Schweiger, Polashenski, C., D. Perovich, J. Richter-Menge, and M. Steele, M. Stark, and H.C. Graber. 2016. B. Elder. 2011. Seasonal ice-mass balance buoys: Modeling the seasonal evolution of the Arctic sea Adapting tools to the changing Arctic. Annals of ice floe size distribution. Elementa: Science of Glaciology 52(57):18–26. the Anthropocene 4:126, http://doi.org/10.12952/​ Proshutinsky, A., R. Krishfield, and D. Barber. 2009a. journal.elementa.000126. Preface to special section on Beaufort Gyre Zippel, S., and J. Thomson. 2016. Air-sea interactions Climate System Exploration Studies: Documenting in the marginal ice zone. Elementa: Science of the key parameters to understand environmental vari- Anthropocene 4:95, http://doi.org/10.12952/journal. ability. Journal of Geophysical Research, 114, elementa.000095. C00A08, https://doi.org/10.1029/2008JC005162. Proshutinsky, A., R. Krishfield, M.-L. Timmermans, ACKNOWLEDGMENTS J. Toole, E. Carmack, F. McLaughlin, W.J. Williams, The results presented here are the product of the S. Zimmermann, M. Itoh, and K. Shimada. 2009b. dedication and hard work of the Marginal Ice Zone Beaufort Gyre freshwater reservoir: State and Sea State Science teams. We thank the Office of and variability from observations. Journal of Naval Research Arctic and Global Prediction Program, Geophysical Research, 114, C00A10, https://doi.org/​ and program managers Martin Jeffries and Scott 10.1029/2008JC005104. Harper, for their continued support of both Arctic Rogers, E., J. Thomson, H. Shen, M. Doble, research and the development of new technologies S. Cheng, and P. Wadhams. 2016. Dissipation and approaches for collecting the necessary observa- of wind waves by pancake and frazil ice in the tions. The Korean Polar Research Institute has been autumn Beaufort Sea. Journal of Geophysical an exceptional, and generous, scientific collabora- Research 121:7,991–8,007, https://doi.org/​ tor—we thank Sung-Ho Kang and the Captain and 10.1002/2016JC012251. crew of IBRV Araon. We also thank the US National Schweiger, A., R. Lindsay, J. Zhang, M. Steele, Ice Center for their timely and sustained support H. Stern, and R. Kwok. 2011. Uncertainty in mod- through both projects, and the Captains and crews of eled Arctic sea ice volume. Journal of Geophysical R/Vs Ukpik, Norseman II, and Sikuliaq. The Office of Research 116, C00D06, https://doi.org/​10.1029/​ Naval Research supported preparation of this paper 2011JC007084. under grant N00014-12-1-0180 (Lee) and N00014-13-1- Shimada, K., T. Kamoshida, M. Itoh, S. Nishino, 0284 (Thomson). E. Carmack, F. McLaughlin, S. Zimmermann, and A. Proshutinsky. 2006. Pacific Ocean inflow: Influence on catastrophic reduction of AUTHORS sea ice cover in the Arctic Ocean. Geophysical Craig M. Lee ([email protected]) is Senior Research Letters 33, L08605, https://doi.org/​ Principal Oceanographer and Professor, and 10.1029/2005GL025624. Jim Thomson is Principal Oceanographer, both Smith, M., and J. Thomson. 2016. Scaling obser- at Applied Physics Lab, University of Washington, vations of surface waves in the Beaufort Sea. Seattle, WA, USA. Elementa: Science of the Anthropocene 4:97, The Marginal Ice Zone Team: Kyoung-Ho Cho, http://doi.org/10.12952/journal.elementa.000097. Sylvia Cole, Martin Doble, Lee Freitag, Hans Graber, Steele, M., S. Dickinson, J. Zhang, and R. Lindsay. Byongjun Hwang, Steve Jayne, Sung-Ho Kang, 2015. Seasonal ice loss in the Beaufort Sea: Joo-Hong Kim, Rick Krishfield, Craig Lee, Ted Toward synchrony and prediction. Journal Maksym, Wieslaw Maslowski, Breckner Owens, of Geophysical Research 120:1,118–1,132, Mary Jane Perry, Pam Posey, Luc Rainville, Jackie https://doi.org/10.1002/2014JC010247. Richter-Menge, Andrew Roberts, Axel Schweiger, Thomson, J. 2012. Wave breaking dis- William Shaw, Timothy Stanton, Michael Steele, Jim sipation observed with “SWIFT” drift- Thomson, Mary-Louise Timmermans, John Toole, ers. Journal of Atmospheric and Oceanic Peter Wadhams, Jeremy Wilkinson, Eun Jin Yang, Technology 29:1,866–1,882, https://doi.org/10.1175/ and Jinlun Zhang. JTECH-D-12-00018.1. Thomson, J., S. Ackley, H.H. Shen, and W.E. Rogers. The Arctic Sea State Team: Stephen Ackley, Fabrice 2017. The balance of ice, waves, and winds in the Ardhuin, Fanny Girard Ardhuin, Alexander Babanin, Arctic autumn. Eos 98, https://doi.org/10.1029/​ Tripp Collins, Martin Doble, Christopher Fairall, 2017EO066029. Johannes Gemmrich, Hans Graber, Peter Guest, Benjamin Holt, Susanne Lehner, Ted Maksym, Michael

68 Oceanography | Vol.30, No.2