SEARCH

SEARCH BY CITATION

Keywords:

  • absolute dispersion;
  • ice beacon trajectories;
  • scaling laws;
  • sea ice motion;
  • southern Beaufort Sea

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[1] Sea ice motion is an important element in mass balance calculations, ice thermodynamic modeling, ice management plans for industry, and ecosystems studies. In the historical literature, sea ice motion in the Beaufort Sea was characterized by a predominantly anticyclonic motion during winter months, with episodic reversals to cyclonic activity during summer. However, recent studies have shown an increase in cyclonic activity throughout the annual cycle. In this paper we examine circulation in the Beaufort Sea based on the trajectories of 22 ice beacons launched in the Franklin Bay area during the International Polar Year - Circumpolar Flaw Lead (IPY-CFL) study during an over-wintering experiment in 2007–2008. Dispersion characteristics of ice motion show that absolute zonal dispersion follows a t2 scaling law characteristic of advection associated with Beaufort Gyre circulation, whereas absolute meridional dispersion follows a scaling law of t5/4 characteristic of floaters and dispersion in 2-D turbulence. Temporal autocorrelations of ice velocity fluctuations highlight definitive timescales with values of 1.2 (0.7) days in the zonal (meridional) direction. Near-Gaussian behavior is reflected in higher-order moments for ice velocity fluctuation probability density functions (pdfs). Non-Gaussian behavior for absolute displacement pdfs indicates spatial heterogeneity in the ice motion fields. Atmospheric forcing of sea ice is explored through analysis of daily North American Regional Reanalysis and in situ wind data, where it is shown that ice in the CFL study region travels with an average speed of approximately 0.2% and an average angle of 51.5° to the right of the surface winds during the 2007–2008 winter. The results from this analysis further demonstrate seasonality in ice drift to wind ratios and angles that corresponds to stress buoy data indicative of increases in internal ice stress and connectivity due to consolidation of the seasonal ice zone to the coast and perennial ice pack during winter in the Beaufort Sea region.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[2] Sea ice motion in the Beaufort Sea is characterized in winter by anticyclonic circulation associated with a predominant sea level pressure (SLP) high in the region. Previous studies have described the Arctic climate in the context of the Beaufort Gyre and two large-scale circulation regimes in accordance with variations in a predominant SLP high: the anticyclonic circulation regime (ACCR) and the cyclonic circulation regime (CCR) [Proshutinsky and Johnson, 1997; Proshutinsky et al., 2002]. During the ACCR, a SLP high predominates resulting in enhanced ice growth and accumulation due to convergence and ridging, with an increase in warm water flux to the Arctic and migration of cyclones to the polar region [Proshutinsky et al., 2002]. By contrast during the CCR, a SLP low predominates and increased melt due to warming and loss of ice associated with Ekman divergence occurs. Reversals in the Beaufort Gyre occur in summer due to SLP lows, as highlighted in modeling studies by Preller and Posey [1989]. Previous studies have also captured large-scale circulation and reversals of the Beaufort Gyre based on regional computation of relative vorticity spatially averaged over the Beaufort Sea region [Ledrew et al., 1991; Lukovich and Barber, 2006]. Ice motion in this instance was computed based on an Eulerian interpretation derived from National Snow and Ice Data Centre (NSIDC) ice motion data [Fowler, 2003]. Also of interest are smaller-scale properties of ice motion based on in situ observations and a Lagrangian interpretation that follows particle paths in the Beaufort Sea region.

[3] Recent observations have highlighted a predominance of rotten, or decayed multiyear ice in the Beaufort Sea region [Barber et al., 2009], while recent studies have also demonstrated an increase in Beaufort sea ice gyre reversals, and thus ice divergence, throughout the annual cycle. Both an increasingly fractured ice cover and ice divergence will give rise to smaller-scale dynamical phenomena and heterogeneity in the ice pack that will affect the movement of ice floes and entrainment of nutrients and contaminants as regions that are typically ice covered become ice free. An understanding of sea ice motion and ice-atmosphere interactions is essential to the identification of navigable shipping and transportation routes; the assessment of regions appropriate for renewable energy extraction and development within seasonally ice-free regions; sea ice prediction and forecasting; the identification of ecologically sensitive and nutrient-rich regions; nutrient and contaminant transport, and the livelihoods of those living in northern communities.

[4] Ice beacon trajectories have been used to monitor ice motion for over 20 years and various techniques to monitor ice motion for over one hundred [Colony and Thorndike, 1984, 1985]. Quantitative estimates of mean sea ice motion in the Arctic were first conducted based on an optimal linear estimation approach for a 90 year record of ice beacon trajectories in the Arctic, and a model for ice dynamics in the context of Brownian (diffusive) motion [Colony and Thorndike, 1984, 1985]. A Lagrangian integral (characteristic) timescale of 5 days and more rapid growth in displacement variance (dispersion) than that encountered by a diffusive regime for the total mean sea ice field signaled the existence of long-range correlations not explained by Brownian motion, while also demonstrating the role of a Lagrangian statistical approach in sea ice predictions and in understanding pollutant transport by sea ice [Colony and Thorndike, 1985]. Earlier studies have also investigated backward trajectories to monitor transport and entrainment of sediments and contaminants by ice, and the age of sediment-laden ice relative to its origin [Pfirman et al., 1997].

[5] Recent studies have further illustrated the role of Lagrangian statistics and dispersion, namely absolute (single-particle) and relative (two-particle) dispersion, and velocity fluctuations in quantifying ice motion in the Arctic [Rampal et al., 2008, 2009a]. A single-particle statistical analysis of the International Arctic Buoy Program (IABP) ice beacon data set from 1979 to 2001 for flow partitioned into a mean and fluctuating component with spatial and temporal averaging scales of 400 km and 5.5 months in winter resulted in an integral timescale of 1.5 days and a linear time dependence in fluctuating variance (zonal and meridional combined) characteristic of diffusion [Rampal et al., 2009a], in a manner consistent with turbulent diffusion theory [Taylor, 1922]. Departures from the turbulent diffusion theory were however encountered in non-Gaussian behavior and exponential decay in ice velocity distributions, and in signatures of intermittency in sea ice velocity, thought to be an artifact of deformation and straining effects in sea ice.

[6] Lagrangian statistical analyses based on ocean drifters and subsurface floaters for ocean currents also illustrate similarities and deviations from turbulent diffusion theory [Bracco et al., 2000; Zhang et al., 2001; Martins et al., 2002]. Non-Gaussian behavior similar to that found for sea ice motion by Rampal et al. [2009a] was discovered in studies of subsurface floaters in the North Atlantic and equatorial Atlantic [Bracco et al., 2000]. Non-Gaussian behavior for fluctuating current velocities in the zonal and meridional direction was attributed to energetic events associated with organized structure in the flow such as jets and vortices. It was further noted that departures from Gaussian behavior, or Gaussianity, may be due to both geographical inhomogeneities (addressed through spatial binning) and distinct dynamical regimes (or what is referred to by Bracco et al. [2000] as dynamic inhomogeneities such as vortices in background turbulent flow and addressed through temporal binning). Here it was also shown that ocean circulation and exponential tails in the probability density functions (pdfs) for zonal and meridional fluctuating current velocities in the North Atlantic and equatorial Atlantic are captured by models of decaying barotropic turbulence. By contrast, Lagrangian statistics computed for isopycnal float data in the Newfoundland basin were found to be well represented by turbulent diffusion theory with Lagrangian integral timescales of 1.5–2.5 days, and Gaussian current velocity fluctuation distributions [Zhang et al., 2001]. Similarly, examination of near-surface circulation in the eastern North Atlantic determined from drifters resulted in Lagrangian integral timescales of 3.7–5.5 days in the zonal direction and 3.6–4.6 days in the meridional direction, while also obeying Taylor's theory of linear scaling for displacement variance (dispersion) in the long time limit [Martins et al., 2002].

[7] As is now well known, the record minimum in summertime sea ice extent in the Arctic was observed on 14 September 2007. Investigation of dynamic forcing mechanisms responsible for the record reduction in ice cover in 2007 showed that a persistent SLP high over the Beaufort Sea region (BSR) during the preceding summer (June, July, and August 2007) resulted in anticyclonic circulation in ice motion and favored increased advection from the Pacific to the Atlantic sector [Kwok, 2008]. Anticyclonic wind anomalies over the BSR in the summer of 2007 were also shown to induce Ekman ice drift toward the central Arctic (or convergence due to anticyclonic circulation and concomitant upwelling) in a manner consistent with evidence from analysis of daily Ekman transport from 1978 to 2003 for enhanced Ekman transport in the presence of strong anticyclonic winds and ice velocities [Yang, 2006], and resulting in increased ice edge retreat in the Beaufort Sea region in particular [Ogi et al., 2008]. Here it was also shown that sea ice response to atmospheric forcing is governed by free ice drift conditions and balance of the Coriolis force, wind stress on the surface of the ice, and water stress on the bottom of the ice; parcels of ice travel exhibit Ekman transport and travel to the right of surface winds. Internal stresses however disrupt the balance of forces so that ice drift is no longer governed by Ekman turning, resulting in ice motion that is parallel to the surface geostrophic winds. Interactions between the ocean, sea ice and atmosphere have also been documented in the context of Ekman transport. Evidence for accelerated sea ice drift and wind stress since the 1950s due to a poleward migration of storms, in addition to increased ice deformation rates since 1978, with maximum deformation rates for winter and summer of 2007, suggests increased fracturing, a thinner ice cover and correspondence between sea ice deformation and retreat [Hakkinen et al., 2008; Rampal et al., 2009b]. Numerous studies have noted a reduction in sea ice thickness, with the most dramatic decline occurring in marginal ice zones and coastal regions [Deser and Teng, 2008], while recent studies have highlighted a significant decline in multiyear ice [Rothrock et al., 1999; Lindsay and Zhang, 2005; Comiso, 2006; Maslanik et al., 2007; Haas et al., 2008; Barber et al., 2009]. Single-particle and Lagrangian dispersion statistics may provide the tool with which to capture changes in Arctic sea ice motion that accompany a thermodynamically susceptible and increasingly mobile ice pack.

[8] The purpose of the present investigation is to quantify regional scale sea ice dynamics in the International Polar Year - Circumpolar Flaw Lead (IPY-CFL) study region [Barber et al., 2010a] using a Lagrangian single-particle statistical analysis of ice beacon trajectories launched during the CFL study. Noteworthy in this field experiment is the location of the beacon drift trajectories in the seasonal ice zone, at significantly lower latitudes relative to the archived IABP drift trajectories that are located primarily in the perennial ice zone. Examined in particular are scaling laws to identify coherent spatial and temporal features in the flow associated with ice motion in the seasonal ice zone, for the purposes of validation in sea ice modeling studies. We also seek to determine whether ice beacon data obtained during the CFL study following the record minimum in sea ice extent in September 2007 yields results similar to the exponential decay observed in ice velocity fluctuation probability density functions for ice beacon trajectories from 1979 to 2001 [Rampal et al., 2009a], and to therefore determine the applicability of the advection-diffusion equation in modeling ice motion in an increasingly mobile and fractured ice cover. In order to achieve these objectives, we address the following research questions:

[9] 1. What is the nature of sea ice motion in the CFL study region? Can scaling laws for single-particle dispersion be determined? How are coherent features in ice motion captured by single-particle statistics (ice beacon trajectories and absolute dispersion; section 3.1)?

[10] 2. What are the spatial and temporal characteristics of ice motion in this region (temporal and spatial properties; section 3.2)?

[11] 3. How is sea ice motion influenced by atmospheric (surface wind) forcing? What was the nature of sea-ice-atmosphere coupling during the CFL study (atmospheric forcing; section 3.3)?

2. Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[12] Ice drift data were obtained from a total of 22 ice beacons launched and monitored in Franklin Bay, from 19 November 2007 to 22 May 2008 [Barber et al., 2010b]. A list of ice beacon IDs, start dates and locations and end date and locations is recorded in Table 1. Recorded parameters include latitude, longitude, speed and direction, with a sampling frequency of 2 h. Due to missing data for two ice beacons, 20 ice beacons were used in the present analysis (Figure 1). Zonal and meridional ice velocities were computed from the speed and direction of the velocities recorded for each ice beacon, and daily averages calculated.

image

Figure 1. Trajectories for ice beacons launched during CFL study, from 19 November 2007 to 22 May 2008. Smaller-scale features associated with meanders in the ice beacon trajectories are shown in the inset for the beacons indicated.

Download figure to PowerPoint

Table 1. Ice Beacon IDs, Start Dates and Locations, and End Dates and Locations
Beacon NumberBeacon IDStart DateStart LocationEnd DateEnd Location
11202225022 Dec 200771.6469°N, 127.379°W22 May 200873.4502°N, 163.019°W
21202333024 Nov 200773.9197°N, 127.398°W22 May 200875.3655°N, 171.258°W
3120242207 Dec 200771.2511°N, 124.936°W7 Apr 200872.2978°N, 153.73°W
4120253305 Jan 200871.5051°N, 125.397°W18 Mar 200871.6027°N, 134.709°W
51252060015 Jan 200871.4320°N, 125.037°W22 May 200871.9556°N, 148.967°W
61252061025 Feb 200870.9886°N, 123.920°W22 May 200870.7743°N, 138.88°W
71252559017 Apr 200870.7110°N, 122.116°W22 May 200871.7345°N, 129.033°W
8126150005 Apr 200871.0772°N, 123.724°W13 Apr 200871.3616°N, 125.358°W
9126187703 May 200870.8604°N, 125.041°W22 May 200872.1541°N, 127.883°W
101281332024 Nov 200773.8287°N, 127.447°W22 May 200875.4763°N, 171.157°W
111281333018 Feb 200871.3233°N, 124.521°W22 May 200871.3862°N, 143.474°W
121281431026 Feb 200871.0109°N, 123.422°W22 May 200870.7726°N, 134.402°W
131281732018 Mar 200871.4103°N, 120.573°W22 May 200871.1512°N, 120.845°W
141281830019 Nov 200771.2622°N, 121.735°W22 May 200872.2022°N, 150.915°W
15128193006 Mar 200871.0592°N, 123.478°W22 May 200870.7536°N, 138.255°W
161281931014 Apr 200871.2050°N, 122.031°W22 May 200871.2050°N, 122.031°W
171291292024 Nov 200873.8351°N, 127.437°W13 Apr 200873.7237°N, 162.774°W
181291593015 Jan 200871.3813°N, 124.307°W26 Jan 200871.4356°N, 124.598°W
191291792022 Dec 200771.4258°N, 125.408°W22 May 200872.0708°N, 150.74°W
201291892022 Dec 200771.6573°N, 127.372°W3 Feb 200870.6976°N, 142.254°W

[13] As noted in section 1, the concept of absolute, or single-particle displacement, is used in the present analysis to characterize ice motion in the zonal and meridional directions. Single-particle displacement is defined as

  • equation image

where xi depicts the zonal or meridional location in degrees or radians, and monitors the presence or absence of a single root-mean-square velocity in advecting flow. Absolute dispersion is defined as [Taylor, 1922]

  • equation image

and indicates the presence of coherent (organized) structure in the flow. Temporal scaling laws for single-particle dispersion provide a diagnostic for dispersion and identify distinct dispersion regimes according to a characteristic timescale known as the Lagrangian integral timescale, TL, expressed as

  • equation image

where R(τ) is the correlation coefficient and is defined as

  • equation image

where V is the difference between the Lagrangian speed and the ensemble average. The importance of the correlation coefficient exists in its monitoring of the memory of initial velocities. Moreover, the Lagrangian integral timescale monitors topological features of the flow and partitions turbulence into three temporal regimes. For short times (t ≪ TL), dispersion is characterized by a ballistic regime such that

  • equation image

while for long times (t ≫ TL), dispersion is distinguished by a diffusive regime, and

  • equation image

where particle paths follow random walks in a manner similar to Brownian motion [Provenzale, 1999]. For intermediate times, dispersion is quantified by the scaling law

  • equation image

where dispersion in elliptic (rotational) and hyperbolic (shear and stretching) regimes is characterized by slopes of 5/3 and 5/4, respectively [Elhmaïdi et al., 1993]. An anomalous intermediate time regime of A2 ∼ t5/4 is also observed for subsurface ocean floats, and in vortex-dominated barotropic and quasigeostrophic turbulence [Rupolo et al., 1996; Bracco et al., 2004; Koszalka et al., 2009]. Scaling exponents in the present analysis were determined from the slope of the log-log plots for single particle dispersion as a function of time elapsed. Of particular interest is the correspondence between the eddy diffusivity and absolute dispersion, which is defined as [Babiano et al., 1990]

  • equation image

According to this relation, eddy diffusivities are defined only when dispersion is described by a linear scaling law characteristic of the diffusive regime and Brownian motion. In the absence of a linear scaling regime characteristic of diffusion, a constant eddy diffusivity is not defined.

[14] Spatial and temporal properties were investigated by determining Lagrangian integral timescales, which were computed in the present study by numerically integrating the temporal autocorrelation function for fluctuating velocity components to the time of zero crossing. Probability density functions of ice velocities and their fluctuations were also determined in a manner similar to Rampal et al. [2009a], Zhang et al. [2001], and Bracco et al. [2000] in order to minimize contributions from zonal advection due to large-scale anticyclonic circulation of the Beaufort Gyre. Departures from Gaussianity were examined using higher-order moments of pdfs, namely the skewness and flatness factors. Skewness, the third-order moment of pdfs, is defined as

  • equation image

for X = x(t) − x(0) the displacement and σ = sqrt(A2), and monitors asymmetry in the flow. Flatness is the fourth-order moment of a distribution and is similarly defined as

  • equation image

and monitors spatial heterogeneity. Fourth-order moments are also monitored by kurtosis, K = F − 3; flatness is computed in the present analysis. S = 0 and F = 3 for Gaussian statistics. F > 3 (F < 3) indicates a more (less) slanted tail than for the Gaussian distribution [Jensen et al., 2005].

[15] Time evolution in zonal and meridional single-particle displacement pdfs was also examined to distinguish spatial properties in ice motion in the zonal and meridional directions. Nonzero skewness indicates asymmetry in beacon displacement. However, non-Gaussian behavior and values of F greater than 3 for single-particle displacement distributions provide a signature of long-range correlations and large-scale intermittency in the flow [Majda and Kramer, 1999].

[16] In consideration of instrument bias error for the ice beacons launched during the IPY-CFL study, position accuracy values ranged from image = 2.5 to 5 m based on circular and spherical error probability associated with the GPS module, whereas temporal accuracy was on the order of nanoseconds and thus disregarded in the present study. Error propagation analysis for single-particle displacement and dispersion yield error estimates on the order of δD = equation image · image ∼ 0.003–0.01 km for the single-particle displacement and image = 2 · image · equation image · equation image or 0.02 times the square root of the single-particle dispersion results, with maximum values of 3 km in the zonal and 2 km in the meridional direction.

[17] Ice conditions during the IPY-CFL study are examined using 12.5 km resolution Advanced Microwave Scanning Radiometer–EOS (AMSR-E) daily sea ice concentrations. Monthly ice concentrations illustrate spatial variability in the sea ice concentration field in the southern Beaufort Sea from November 2007 to May 2008. Time series of daily sea ice concentrations located within a 0.2° (approximately 25 km or twice the resolution of the AMSR-E data) radius of the ice beacons and corresponding minimum and maximum values are compared with zonal and meridional sea ice motion to gain further insight into connections between sea ice motion and ice conditions in the seasonal ice zone.

[18] Large-scale and low-frequency circulation of sea ice is governed by oceanic phenomena; in the absence of high-frequency ocean data we focus on high-frequency atmospheric variability and ice-atmosphere interactions using both North American Regional Reanalysis (NARR) wind data and in situ measurements from the ship. Atmospheric forcing of sea ice motion was investigated using daily averages of surface zonal and meridional winds computed from 3-hourly NARR data for the region extending from 69°N to 74°N and 160°W to 120°W to encompass the CFL study area and Beaufort Sea region. In light of recent concerns about the quality of NARR data at high latitudes, in situ data for wind speed and direction collected from an RM Young 05103 anemometer located on the Amundsen were also used in the atmospheric forcing analysis (see Barber et al. [2010a] for further details on meteorological equipment used during the IPY-CFL study). Examined in particular was the ratio of ice velocities to NARR mean velocities; positive (negative) ratio values significantly less than one indicated wind-driven ice events with comparable (opposite) orientation, while positive (negative) ratio values significantly greater than one indicated vanishing winds and ice events resulting from alternative forcing mechanisms. Ratios of higher-order moments illustrated deviations in ice and velocity distributions as a function of time (days during the November 2007 to May 2008 time interval). In particular, ratios of standard deviations for zonal and meridional ice velocities and NARR zonal and meridional winds provided a signature of relative variability in ice-atmosphere coupling as a function of time; ratios of skewness provided a signature of ice motion direction relative to surface wind flow direction; ratios of flatness highlighted relative occurrences in ice-atmosphere coupling of extreme events. Due to limited spatial information from the in situ ship-based measurements, only NARR winds were considered when examining higher-order moments of pdfs for winds. Ice-atmosphere coupling and Ekman transport were further examined through investigation of stick vector plots for NARR surface winds and ice velocities.

3. Results and Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

3.1. Ice Beacon Trajectories and Absolute Dispersion

[19] Ice beacon trajectories launched during the IPY-CFL study beginning in November 2007 and ending in May 2008 demonstrate both large- and small-scale features in ice motion (Figure 1). Noteworthy is the large-scale circulation and westward advection associated with anticyclonic circulation of the Beaufort Gyre. Small-scale variability is evident in loops and meanders in each of the trajectories. As an example of ice beacon paths, beacon 10 (Table 1) was launched just off the west coast of Banks Island where it began a southward descent until it began westward transport that was maintained until the end of the study period roughly 900 km north of the northeast corner of Russia in the Arctic Ocean. This corresponds to beacon 10 values for speeds in the zonal and meridional directions shown in Figure 2, with negative zonal winds and slightly positive meridional winds due to the gradual northward movement of the beacon after it passes Mackenzie Bay. This pattern of southward movement along the western edge of Canada's Arctic and then westward displacement close to the northern continental coast is characteristic of an arctic drift model presented by Pfirman et al. [1997]. The time taken to travel from Banks Island to the Bering Strait is also consistent with the 6 month interval Pfirman et al. [1997] attributed to ice beacons launched in this region between 1979 and 1994. In addition, most beacons stay within 250 km and are constrained to the continental coast until they travel northward over Bering Strait, which may perhaps be partially understood through investigation of sea ice extent from November 2007 to May 2008, in addition to spatial patterns of ocean currents and surface winds in this region.

image

Figure 2. Average zonal and meridional components of ice beacon motion during the CFL study.

Download figure to PowerPoint

[20] Average zonal and meridional ice motion speeds for each of the ice beacons highlight predominantly westward advection in the zonal direction and small-scale motion in the meridional direction (Figure 2). The zonal component of ice motion averaged from November 2007 to May 2008 exhibits negative values characteristic of westward advection for nineteen of the twenty ice beacons, with zonal speeds ranging from −0.8 to −0.1 m s−1. By contrast, meridional variability is captured in negative and positive average speeds characteristic of northerly and southerly flow, with meridional speeds that range from −45 to 35 cm s−1. Those beacons launched during winter and early spring exhibit vanishing mean zonal and meridional ice speeds (in particular, beacons 12618770, 12817320, and 12912930), while those launched during November and December show a strong westward and northerly component (beacons 12912920 and 12912930).

[21] The existence of a well-defined slope for zonal displacement provides a signature of advection in the zonal direction in the southern Beaufort Sea associated with mean anticyclonic circulation of the Beaufort Gyre during the 2007–2008 winter (Figure 3). The absence of a well-defined slope in the meridional direction indicates that meridional transport is not characterized by a single transport velocity, but rather a range of temporal and spatial scales that give rise to small-scale reversals evidenced in loops and meanders. In the zonal direction ice beacons travel a maximum distance of ∼1200 km, whereas in the meridional direction ice beacons travel distances of up to ∼150 km (southward displacement is on the order of 50 km, while northward displacement is on the order of 150 km).

image

Figure 3. Absolute (a) zonal and (b) meridional mean single-particle displacements for the 20 ice beacons launched during the CFL study.

Download figure to PowerPoint

[22] Properties of flow in the southern Beaufort Sea are further illustrated in distinct scaling laws for absolute zonal and meridional dispersion (Figure 4). In particular, zonal dispersion is depicted by t2 scaling, indicating flow dominated by advection. Meridional dispersion is depicted by t5/4 scaling which, as was noted in section 2, suggests that meridional transport is described by a hyperbolic (strain-dominated) domain (as opposed to an elliptic domain) in the southern Beaufort Sea during winter in a manner comparable to subsurface float observations and 2-D turbulence. Of interest is the existence of a well-defined power law despite the limited number of beacons used in the present analysis. Similar behavior, namely correspondence between scaling laws derived from 2-D turbulence and computed from limited observations, has however also been encountered in the context of atmospheric phenomena. Indeed, Morel and Larcheveque [1974] discovered eddy dispersion in the southern hemisphere based on balloon measurements to be well represented by two-dimensional turbulence models for spatial scales ranging from 100 to 1000 km, and used this data set to measure the eddy diffusion coefficient for large-scale motion in the lower stratosphere. Similarly, the aforementioned investigation of Lagrangian dispersion statistics for 31 drifters in the North Atlantic enabled computation of eddy diffusivities for the region [Martins et al., 2002]. Investigation of coherent features in the North Atlantic provided further illustration of the applicability of advection-diffusion equations for flow in the oceanic regime, and is explored further in section 3.2 in the context of temporal and spatial properties of ice motion.

image

Figure 4. Absolute (a) zonal dispersion, (b) meridional dispersion, and (c) zonal and meridional dispersion and scaling laws for 20 ice beacons launched during the CFL study.

Download figure to PowerPoint

3.2. Temporal and Spatial Properties

3.2.1. Temporal Properties

[23] Temporal autocorrelations for the zonal and meridional components of sea ice motion highlight dominant timescales associated with the total and fluctuating component as determined from the ice beacon data (Figure 5). Noteworthy is the absence of a definitive timescale for the total component (when the mean is not removed) in the zonal direction (Figure 5a); the autocorrelation function approaches but does not decay to zero and highlights persistent memory inherent in the system due to predominantly westward advection associated with anticyclonic circulation of the Beaufort Gyre during winter. By contrast, the autocorrelation function for the fluctuating zonal velocity decays to zero after approximately 10 days and equilibrates after 20 days. The distinction between the total and fluctuating velocity field is less apparent in the meridional direction (Figure 5b). Temporal autocorrelations decay to zero after approximately 10 days and equilibrate after 20 days, while a decline after approximately 7 days and equilibration following 10 days is observed for fluctuating meridional ice velocities.

image

Figure 5. Temporal autocorrelations for (a) zonal and (b) meridional components of sea ice motion.

Download figure to PowerPoint

[24] The more rapid decay for fluctuating meridional ice velocities, evidenced in Lagrangian integral timescales of 1.2 days in the zonal direction and 0.7 days in the meridional direction, indicates shorter memory in the meridional direction than in the zonal direction (Figures 5a and 5b). Lagrangian integral timescales determined from an analysis of 450 drifters as part of the IABP data set from 1979 to 2001 [Rampal et al., 2009a] found integral timescales to be on the order of 1.5 days. Rampal et al. [2009a] also noted that values for integral timescales determined from previous analyses of 28 drifters in the Arctic Basin [Colony and Thorndike, 1984] were on the order of 5 days, and attributed the distinction to varying responses to atmospheric and oceanic forcing mechanisms; the 1.5 day timescale was associated with oceanic phenomena and the 5 day timescale to synoptic atmospheric processes. The importance of accurately resolving the mean component of the flow to determine accurate timescales to ascertain the applicability of a turbulent diffusion description for fluctuating ice velocities was also emphasized.

[25] Results from this analysis indicate that a distinctive timescale does exist for the fluctuating zonal and meridional ice velocities with timescales on the order of 1.2 and 0.7 days, respectively, and suggest based on previous studies that ice motion in the zonal direction is governed by timescales associated with oceanic phenomena (namely circulation of the Beaufort Gyre), while ice motion in the meridional direction is governed by forcing mechanisms associated with higher-frequency phenomena. As previously noted, in the absence of high-frequency ocean data we examine the role of atmospheric forcing on ice motion in the zonal and meridional direction in section 3.3, where an assessment of time evolution in wind fluctuations illustrates enhanced ice-atmosphere coupling in early winter of 2007–2008, and reduction in ice to wind coupling in mid-January. This behavior may be attributed to combinations of atmospheric and oceanic forcing mechanisms and reduced response of the ice to forcing due to high ice concentrations and increasing internal ice stress associated with ice deformation and mechanical thickening of the ice cover.

[26] Analysis of probability density functions (pdfs) for sea ice fluctuating velocities during the CFL study illustrates near-Gaussian behavior with skewness factors of ∼−0.12 and −0.05, and flatness factors of ∼2.83 and 3.03 in the zonal and meridional directions, respectively (Figure 6). As previously noted, Gaussian statistics yield a skewness value of 0 and a flatness value of 3. Probability density functions for the total zonal and meridional velocities exhibit skewness values of ∼−0.84 and −0.39 and flatness factors of ∼11.42 and 23.2 in the zonal and meridional directions, respectively. Non-Gaussian behavior for zonal and meridional velocity pdfs could indicate organized flow due to shear, eddies or reversals in trajectories as a result of energetic infrequent events as described in the study of subsurface floaters by Bracco et al. [2000]. By contrast, near-Gaussian behavior for the pdfs of velocity fluctuations indicates ice flow in this region during winter may be approximated by fully developed turbulence, in contrast with recent studies by Rampal et al. [2009a]. The authors speculate that evidence for Gaussian pdfs in the zonal direction for beacons launched in 2007–2008 in contrast to non-Gaussian pdfs observed by Rampal et al. [2009a] for ice beacon trajectories launched between December 1978 and December 2001 could be a signature of increased mobility in the ice pack and attendant changes in the mechanical response of an increasingly fractured sea ice cover to oceanic and atmospheric forcing since 2001, particularly in light of the fact that the ice beacons were launched in the winter following the record reduction in sea ice extent in September 2007. Decreases in ice thickness in coastal regions [Deser and Teng, 2008] could also give rise to a reduction in internal stresses and deformation responsible for the non-Gaussian Lagrangian statistics of ice velocity fluctuations noted in dispersion studies by Rampal et al. [2009a]. Furthermore, ice beacon trajectories in the present analysis are located in the seasonal ice region and shear zone along the Beaufort coast, whereas ice beacon trajectories from the IABP archive investigated by Rampal et al. [2009a] are located primarily in the perennial ice zone. Differences in pdf behavior may therefore also be attributed to the development of internal stresses within the seasonal ice cover as sea ice consolidates and undergoes sustained winds and coastal interactions that give rise to fractures and leads [Richter-Menge et al., 2002a, 2002b]. Such differences in seasonal and perennial ice zone behavior have also been observed in two-particle statistical analyses of sea ice motion and diffusivity in the Beaufort and Bering Seas [Martin and Thorndike, 1985; Thorndike, 1986]. More recent studies illustrate increased ice production and deformation in the seasonal ice zone than in the perennial ice zone in the Arctic [Kwok, 2006], with implications for ice thickness, ridging, and thus sea ice motion and ice drift tracks in both ice regimes.

image

Figure 6. Probability density function of fluctuating zonal and meridional components for sea ice.

Download figure to PowerPoint

[27] In order to understand the relationship between sea ice conditions and sea ice motion in the seasonal ice zone, we examine sea ice concentrations in the vicinity of the ice beacons during the CFL study (Figure 7). Evolution in the seasonal ice zone comprising the CFL study region is illustrated in monthly sea ice concentration maps from November 2007 to May 2008 (Figure 7a). Regions of open water and reduced ice concentrations along the coastline in November and December 2008 are replaced by a continuous ice cover in January 2008, at which time ice-ice and ice-coast interactions govern the balance of forces acting on the seasonal ice cover. As regions of open water reappear in April and May 2008 the seasonal and perennial ice cover lose contact with the coastline, and the system returns to a state of semifree drift. Time series of daily ice concentrations within a 0.2° (approximately 25 km) radius show that ice beacons encounter consistently high ice concentrations (exceeding 95%) in the seasonal ice zone during fall and winter (Figure 7b), and lower concentrations (85%–95%) in May 2008. Near-Gaussian behavior for pdfs of velocity fluctuations despite consistently high sea ice concentrations in the CFL study region in winter suggests that sea ice cover in the seasonal ice zone is weak even at high concentrations. These results thus also suggest that seasonal ice differs mechanically from perennial ice in a manner consistent with previous studies demonstrating higher sea ice deformation and production in the seasonal than in the perennial ice zone during winter [Richter-Menge et al., 2002b; Kwok, 2006]. Consistently high ice concentrations also indicate that changes in the strength of the ice cover will be a consequence of mechanical thickening due to seasonal and perennial ice zone interactions with the coastline, in support of previous studies of ice stress in the western Arctic [Richter-Menge et al., 2002b], with implications for ice-atmosphere coupling.

image

Figure 7. Sea ice conditions in the southern Beaufort Sea. (a) Sea ice concentration maps from AMSR-E data, and (b) time series of daily mean, maximum, and minimum sea ice concentrations located within a 0.2° radius of the ice beacons (left y axis) and zonal and meridional sea ice motion (right y axis) from 19 November 2007 to 22 May 2008.

Download figure to PowerPoint

3.2.2. Spatial Properties

[28] Spatial properties of sea ice motion are illustrated in the time evolution of zonal and meridional absolute displacement pdfs at 20, 50, and 70 days (Figure 8). In the zonal direction, the majority of ice beacons traveled less than 50 km after 20 days. Two beacons (1 and 20), both launched on 22 December 2007 at approximately 71°N and 127°W, were displaced 400 km to the west of their origin. Beacon 14, launched on 19 November 2007 at approximately 71°N and 121°W was displaced 50 km east of its origin after 20 days, indicating a reversal in advection due to eddy or shearing effects associated with sea ice conditions. After 50 days, a dipolar configuration between ice beacons displaced approximately 200 km and those displaced 600 km, was established. In particular, two ice beacons, namely beacon 10, launched 24 November 2007 near 74°N and 127°W, and beacon 3 launched on 7 December 2007 near 71°N and 125°W exhibit maximum displacements not evident at 20 days of 600 km to the west of their origin, suggesting acceleration of both ice beacons. The maximum displacement for beacon 1 is sustained after 50 days. Similarly following 70 days, where beacons 10 and 3 also reached maximum zonal displacements approaching 800 km. Beacon 15, launched on 6 March 2008 near 71°N and 124°W, demonstrated acceleration following 50 days with a westward displacement of approximately 400 km from its launch. As expected, pdfs for zonal displacements broaden in time and demonstrate a negative skewness characteristic of westward advection associated with anticyclonic circulation of the Beaufort Gyre.

image

Figure 8. Histograms of (left) zonal absolute displacement and (right) meridional absolute displacement at t = 20, 50, and 70 days from launch.

Download figure to PowerPoint

[29] In the meridional direction, beacons 1 and 20 demonstrate southerly displacement from their origin after 20 days. However, a third beacon, launched on 3 May 2008 near 71°N and 125°W (beacon 9), demonstrated maximum poleward displacement on the order of 100 km, indicating significant meridional transport of ice in May. Three ice beacons, namely beacons 4, 5, and 14, were displaced approximately 50 km north. After 50 days, beacon 10 experienced maximum displacement of 200 km southward in a manner consistent with accelerated zonal displacement. After 70 days, the meridional displacement pdfs indicate a poleward displacement of ice beacons (positive displacement), most likely in response to easterly winds, depending on ice conditions. A reversal in beacon 14 occurs after 70 days, indicating poleward displacement and providing further evidence for a response to meridional wind anomalies. In contrast to zonal displacement pdfs, the meridional displacement pdfs are symmetric and show both poleward and equatorward displacement of ice beacons, characteristic of a Gaussian distribution and more diffusive regime. A retreat to a maximum in the distribution of displacements ranging from 0 to 50 km after 70 days indicates oscillation of the meridional ice motion about the origin.

[30] Higher-order moments of probability density functions for zonal and meridional displacements indicate large-scale intermittency through departures from Gaussianity (Figure 9). Of particular interest, as for the investigation of temporal properties, is the evolution in skewness and flatness factors that characterize behavior of the pdf tails. It should be noted that in the present investigation only days with a sample size greater than 10 beacons were used in calculating higher-order moments, and that time series reflect only those calculations including greater than ten beacons, found in this instance to be on the order of 90 days. Time series of the mean zonal displacement (Figure 9a) indicates a mean displacement of approximately 600 km after 90 days, as shown in Figure 3a, with instances of reversals to eastward flow as depicted in interruptions to the linear decline associated with mean flow due to westward advection. By contrast mean flow in the meridional direction is moderately constant, with oscillations indicating equatorward and poleward transport as is captured in Figure 3b. Time series for standard deviation of zonal and meridional displacements indicate extrema on the order of 250 and 100 km, respectively. Variability in both the zonal and meridional directions indicates the absence of a constant eddy diffusivity in both directions (see also Figure 4b, or the square of Figure 9b). Non-Gaussian values for the time-dependent skewness and flatness in the zonal and meridional directions suggests large-scale intermittency and long-range correlations in both the zonal and meridional directions. Separation pdfs are required to determine whether such behavior is a consequence of local or nonlocal interactions. We speculate that coherence in the zonal and meridional direction may be a signature of changes in internal ice stress and deformation as the seasonal ice zone consolidates with the perennial ice zone during winter, as described below, or oceanic eddy activity in this region; further examination, including evaluation of eddy kinetic energy, is required to verify this hypothesis, and is beyond the scope of the present investigation.

image

Figure 9. Time series for higher-order moments, namely (a) mean, (b) standard deviation, (c) skewness, and (d) flatness of pdfs for zonal and meridional absolute displacements. Dotted lines in Figures 9c and 9d indicate Gaussian skewness and flatness values of 0 and 3, respectively.

Download figure to PowerPoint

3.3. Atmospheric Forcing

[31] NARR wind fields for the Beaufort Sea region and the accompanying in situ wind measurements from the CCGS Amundsen illustrate daily evolution in the regional and local wind fields during the CFL study (Figure 10). Noteworthy from the stick vector plots for regional (NARR) winds is predominantly easterly and southeasterly flow in November 2007 and December 2007, characteristic of atmospheric flow following the quasi-stationary SLP high located over the Beaufort Sea region that contributed in prior months to the record minimum in sea ice extent in 2007. Increased variability in surface winds is observed in mid-January and February 2008 due to the dissipation in the SLP high. The return of the SLP high over the Beaufort Sea region in March 2008 is depicted by northeasterly winds that persist through to May 2008. Local (in situ) winds indicate intermittent northerly flow in the Amundsen Gulf in early December, mid-February, and early April, with strong southerly flow in March, in a manner consistent with wind-driven reversals in ice observed in the Amundsen Gulf [Barber et al., 2010a]. Increased correspondence with regional NARR winds is observed following March 2008. A correlation analysis of regional and in situ winds yields statistically significant values of r ∼ 0.6 in the zonal direction and r ∼ 0.2 in the meridional direction, suggesting that zonal variability is captured both by regional and local winds, whereas small-scale variability in the meridional direction depends on location and local phenomena.

image

Figure 10. Stick vector plot during the CFL study period (November 2007 to May 2008) in the Beaufort Sea region for scaled NARR winds (black), in situ (ship-based) winds (red), and ice velocities (blue).

Download figure to PowerPoint

[32] Comparison of daily NARR and in situ winds and sea ice motion shows that throughout most of the time interval considered, ice travels to the right of the surface winds, with exceptions between 7 January and 16 February, 17 March, 7 April, and 16 April (Figure 10); this may be a signature of thick, dense ice encountered by ice beacons which induces a force that opposes the free drift balance and results in the beacons traveling parallel to the geostrophic winds as described by Ogi et al. [2008]. Furthermore, correlations between NARR winds and ice motion yield values of r ∼ 0.4 and r ∼ 0.1 in the zonal and meridional directions, respectively, while correlations between in situ winds and ice motion yield values of r ∼ 0.5 and r ∼ 0.1 in the zonal and meridional directions, respectively.

[33] Previous studies have shown that ice travels at 2% the speed of surface winds, with angles ranging from 10° to 63° to the right of surface winds, depending on seasonal ice drift conditions [Thorndike and Colony, 1982; Fissel and Tang, 1991]. During the IPY-CFL study, ice travels with an average speed of approximately 0.2% and an average angle of 51.5° to the right of the surface winds (Figure 11). Noteworthy is increased ice mobility evident in larger ratios of ice to wind speed in addition to angle values exceeding 90° from mid-January to March, in a manner shown below to be consistent with the analysis of ratios for moments of ice and winds demonstrating a reduction in ice-atmosphere coupling (Figure 12), and reinforcing the assertion that forcing mechanisms other than surface winds contribute to sea ice motion events following mid-January. It is also interesting to note that prior to mid-January, ratios of ice to wind speeds are comparable to those found in summer in 1979 [Thorndike and Colony, 1982], while angles comparable to or less than 90° yield an average value of 45° until mid-January while the SLP high predominates over the CFL study region.

image

Figure 11. (left) Time series for ratio of vector norm for ice velocities to NARR winds (black line) and in situ (red line) winds, and (right) time series of angle (in °) between ice velocities and NARR winds (black line) and in situ winds (red crosses).

Download figure to PowerPoint

image

Figure 12. Ratio of zonal and meridional ice velocity moments to NARR and in situ wind moments. Shown in particular are the ratios of the (a) mean ice motion to mean NARR (solid) and in situ (dashed) winds, (b) standard deviation for ice and NARR winds, (c) skewness for ice and NARR winds, and (d) flatness for ice and NARR winds, for both the zonal (black) and meridional (red) components.

Download figure to PowerPoint

[34] Sea-ice-atmosphere coupling is further quantified in the evolution in higher-order moments of pdfs for ice motion and winds during the CFL study (Figure 12). Ratios of mean ice zonal velocities to NARR zonal winds indicate values less than unity from November 2007 until early January 2008, and following March, suggesting wind-driven ice motion (red line in Figure 12a). Ratio values significantly greater (less) than one near day 50 indicate reversals in orientation, and suggest ice motion events in the zonal direction driven by phenomena other than surface winds starting mid-January. Ratio values in the meridional direction also exhibit reversals and significant departures from unity following 17 January, near 16 April, and again near 6 May, suggesting that meridional ice motion is governed by surface winds from November to mid-January, as for the zonal ice velocity, with episodic ice motion events for the remainder of the CFL study period due to other forcing factors such as ocean currents, ice ridging and/or mesoscale eddies. Similar results are observed in ratios between mean ice motion and in situ winds (dashed line in Figure 12a), while significant differences are also observed following April due to the distance between the ship-based winds in the Amundsen Gulf and the zonally and meridionally displaced ice beacons. Ratios of standard deviations for ice and NARR winds show enhanced ice-atmosphere variability in the meridional direction, with significant differences following the beginning of February (Figure 12b). Ratios for skewness in the zonal direction show that ice beacon trajectories tend to travel opposite to surface winds in the southern Beaufort Sea following the end of March, while in the meridional direction significant departures are observed throughout the duration of the CFL study, with ice motion traveling in the same direction as surface winds (Figure 12c). Ice-atmosphere coupling determined from the ratios for flatness exhibits significant variability in extreme events in the zonal and meridional direction following mid-January (Figure 12d).

[35] Results from an investigation of the time evolution in ratios of ice and surface wind moments during the CFL study show that sea-ice-atmosphere coupling in the zonal direction is characterized by wind-driven events from November to mid-January, most likely associated with a dominant SLP high in the fall of 2007, and a breakdown in this correspondence from mid-January to March. Enhanced variability in ratios of ice and wind moments in the meridional direction suggests dominant contributions to ice motion from alternative external forcing mechanisms from mid-January to May, which may include ice deformation and ridging and oceanic forcing, the relative contributions of which are examined in a separate investigation of case studies for ice beacon reversals.

[36] It is interesting to note that seasonal evolution in ice to wind speed ratios and angles corresponds to seasonality in internal ice stress buoy measurements in the Beaufort Sea region [Richter-Menge et al., 2002a, 2002b]. Previous studies have shown the emergence of a continuous ice cover as sea ice consolidates and experiences sustained winds and ice-coast interactions that give rise to the propagation of ice stress away from the shore in the form of leads in the seasonal ice zone [Richter-Menge et al., 2002a]. Subsequent consolidation of the perennial and seasonal ice zone results in enhanced internal ice stress and ridging as ice stress propagates into the perennial ice zone, with coherence in ice motion and deformation over spatial scales of 200 km [Richter-Menge et al., 2002b; McNutt and Overland, 2003]. Similarly and as previously noted, consistently high ice concentrations encountered during the CFL study in winter (Figure 7b) indicates that changes in internal ice stress and strength in the seasonal ice zone during winter are associated with mechanical thickening, deformation and ice-coast interactions rather than spatial variability in ice concentration, with implications for ice-atmosphere interactions.

[37] Correspondence between the time series for angles between wind and ice drift and internal ice stress in winter [Richter-Menge et al., 2002a, Figure 1] suggests that the former provides a signature of connectivity within the Beaufort Sea region. In particular, angles between wind and ice drift exceeding 90° in mid-November correspond to intervals of increased internal stress due to ridging as the seasonal ice zone consolidates against the coast (see also Figure 7a), while subsequent enhanced ice-atmosphere coupling from November to mid-January suggests increased mobility within the seasonal ice zone. The underlying gradual increase in both the internal ice stress characteristic of emergence in a continuous ice cover associated with consolidation of the seasonal and perennial ice zones and angles between winds and ice drift occurs from November to March. Reduced ice-atmosphere coupling from mid-January to March, and a continued increase in angles between wind and ice drift exceeding 90°, for beacons located within 250 km of the coastline, indicate continued ice-coast interaction within the seasonal ice zone during the CFL study, in contrast to the interval of divergence encountered in the study by Richter-Menge et al. [2002a] for stress buoys located within the perennial ice zone located a significant distance from the coastline.

4. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[38] Ice beacon trajectories launched in November 2007 during the IPY-CFL study provided a Lagrangian interpretation of ice motion in the Beaufort Sea region, while also illustrating regional scale properties of ice circulation as captured in single-particle statistics. In consideration of our first research question, well-defined temporal scaling laws do exist for ice motion in the zonal and meridional directions during the 2007–2008 winter for the ensemble of twenty ice beacons. Absolute dispersion in the zonal direction was found to follow a t2 scaling law characteristic of advection associated with anticyclonic circulation of the Beaufort Gyre. Absolute dispersion in the meridional direction was however found to follow a scaling law of t5/4 characteristic of dispersion properties for floaters and in 2-D turbulence, providing a possible signature of mesoscale eddy formation in this region. Further research is required to verify this hypothesis.

[39] In consideration of our second research question, temporal autocorrelations indicate that a distinctive timescale does exist for the fluctuating zonal and meridional ice velocities with Lagrangian integral timescales on the order of 1.2 and 0.7 days, respectively. In addition, near-Gaussian behavior for pdfs of ice velocity fluctuations suggest that ice motion in the Beaufort Sea region in 2007–2008 may be approximated or modeled as a turbulent flow regime. This is in contrast to non-Gaussian behavior associated with ice velocity fluctuations for ice beacon trajectories from 1979 to 2001 [Rampal et al., 2009a], most likely due to differences associated with changes in internal ice stress in the seasonal as opposed to perennial ice zone as sea ice consolidates with the central pack and interacts with the coast resulting in the development of leads and ridges, in addition increased mobility associated with the record reduction in sea ice extent, and in the Beaufort Sea region in particular, in the fall of 2007 [Rampal et al., 2009b; Kwok, 2008; Ogi et al., 2008]. Probability density functions of single-particle displacements illustrate differences in the zonal and meridional directions evidenced in single-particle dispersion statistics: negative skewness characteristic of westward advection associated with anticyclonic circulation of the Beaufort Gyre in the zonal direction and a more symmetric distribution associated with poleward and equatorward displacements in the meridional direction. Departures from Gaussianity evidenced in the skewness and flatness factors indicate that ice motion is governed by spatial inhomogeneity in both the zonal and meridional direction that evolves in time.

[40] In consideration of our third research question, it is shown that sea ice travels with an average speed of approximately 0.2% and an average angle of 51.5° to the right of the surface winds during the 2007–2008 winter. Stick vector plots illustrate Ekman transport for much of the 2007–2008 winter with ice motion oriented to the right of surface winds, with exceptions that may be attributed to internal stresses due to ice deformation, ridging and fracturing of sea ice following the record reduction in sea ice extent in September, particularly in the Beaufort Sea region. Time evolution in the ratios between the moments of ice velocity and surface wind fluctuations shows that sea ice is governed by surface winds from November to mid-January most likely associated with a dominant SLP high in the fall of 2007, followed by a reduction in sea-ice-atmosphere coupling from mid-January to March due to other forcing factors such ocean currents, ice ridging and/or mesoscale eddies. Comparison with results from stress buoy data shows that seasonal evolution in ice-atmosphere interactions and angles between winds and ice drift corresponds to seasonality in internal ice stresses in the Beaufort Sea region and monitors the emergence of a continuous ice pack in winter.

[41] The results from this analysis have demonstrated the role of single-particle dispersion statistics in identifying coherent properties in the flow, and have highlighted distinct scaling laws for ice motion in both the zonal and meridional directions. Moreover, these results provide a Lagrangian interpretation of ice motion as a complement to the Eulerian description of large-scale motion in the Beaufort Sea region using the concept of relative vorticity. Meridional t5/4 scaling suggests the existence of a regime characteristic of dispersion in 2-D turbulence and subsurface float observations. The results from this analysis have implications for our understanding of ice dispersion, hydrocarbon industry related ice management on structures, and mechanisms responsible for nutrient transport and exchange in the Beaufort Sea within an increasingly mobile ice cover.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[42] The authors would like to thank W. Chan for processing the North American Regional Reanalysis (NARR) data provided by NCEP, B. Else for processing the ship-based data, and the crew of the NGCC Amundsen. AMSR-E data are produced by Remote Sensing Systems and sponsored by the NASA Earth Science MEaSUREs DISCOVER Project and the AMSR-E Science Team. Data are available at http://www.remss.com/. Thanks also go to the editors, Cathy Geiger, and an anonymous reviewer for recommendations that led to the improvement of this manuscript. Funding for this study was provided by the International Polar Year - Circumpolar Flaw Lead (IPY-CFL) study, the Canadian Networks of Centres of Excellence (NCE) program, and Canada Research Chairs (CRC) grant (D. G. Barber).

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
jgrc12087-sup-0001-t01.txtplain text document2KTab-delimited Table 1.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.