1. Introduction
Arctic sea ice cover impacts and responds to the local atmospheric features and large-scale variability associated with the Arctic Oscillation (AO; Thompson and Wallace 1998) and North Atlantic Oscillation (NAO; e.g., Hurrell et al. 2003). Changes in surface turbulent heat fluxes and wind stress forcing resulting from sea ice–atmosphere interaction produce immediate and long-term effects on the Arctic climate system that extend into subsequent seasons. Wu and Zhang (2010) showed that variability in the 500-mb height field over the Arctic was correlated with sea ice concentration at 0-to-2-month lags (atmosphere leads ice) during all seasons except winter, when atmospheric patterns were correlated with sea ice concentration at lags between −1 and −4 months (ice leads atmosphere). The leading winter maximum covariance analysis modes of 500-mb height and sea ice concentration depicted ice loss in the marginal seas and a pattern resembling the negative phase of the AO, with positive height anomalies over the North Pole and negative height anomalies over the ocean basins. In a modeling framework, sea ice extent and concentration anomalies derived from observed winters with anomalously high and low ice over the Arctic produced atmospheric responses that resemble the NAO, although their magnitudes were weak relative to internal atmospheric variability (Alexander et al. 2004). The winter AO index is related to both winter sea ice motion and summer sea ice concentration (Rigor et al. 2002), where increased sea ice divergence and decreased ridging during the positive phase of wintertime AO contribute to thinning of Arctic sea ice during the following summer as a result of atmospheric warming from upward latent heat fluxes. In turn, increased upward latent heat fluxes resulting from anomalously low sea ice concentration over the Arctic in August and September decrease static stability, generating a barotropic atmospheric response and negative NAO-like pattern in winter (Jaiser et al. 2012). Reduced Arctic sea ice cover is also associated with decreased winter extratropical storm activity (Seierstad and Bader 2009), winter cold air outbreaks over the northern continents (Honda et al. 2009; Petoukhov and Semenov 2010; Yang and Christensen 2012), and enhanced snowfall over Europe, Eurasia, and portions of the United States (Liu et al. 2012).
The remote influence of sea ice on atmospheric variability may be attributed to stationary Rossby waves generated by surface turbulent heat flux anomalies (Honda et al. 1999). Yamamoto et al. (2006) proposed that Rossby waves originating from the Nordic Seas and Sea of Okhotsk could serve as a dampening mechanism for the NAO that is manifested in a lagged relationship between the positive phase of the NAO and the “seesaw” pattern of anomalously low sea ice concentration over the Nordic Seas and Sea of Okhotsk, and anomalously high sea ice concentration over the Labrador and Bering Seas throughout the winter. Ice removal over the Barents Sea may induce stationary Rossby waves that strengthen the Siberian high, leading to cold air outbreaks over Europe and Eurasia during winter (Honda et al. 2009; Petoukhov and Semenov 2010).
Sea ice–atmosphere interaction is particularly strong in the Atlantic basin during the cold season because of the enhanced variability of the NAO and storm track. The leading pattern of variability in winter Arctic SIC depicts positive and negative ice concentration anomalies east and west of Greenland, respectively, in association with the negative phase of the NAO (Deser et al. 2000). Similarly, Vinje (2001) showed that the December–March NAO index was negatively correlated with April sea ice extent east of Greenland and positively correlated with April ice extent in the Labrador Sea. The relationship between the sea ice dipole pattern and the NAO suggests a negative feedback (Magnusdottir et al. 2004; Strong et al. 2009; Strong and Magnusdottir 2010) that is primarily driven by sea ice anomalies over the Barents Sea (Magnusdottir et al. 2004): anomalously low Barents Sea ice cover is characteristic of the positive NAO, but induces the negative phase of the NAO toward the end of winter. Deser et al. (2004) decomposed the total modeled winter atmospheric response to negative ice extent anomalies over the Barents Sea and positive anomalies over the Labrador Sea into a direct baroclinic response driven by diabatic heating, and an indirect response defined by the projection onto the leading mode of variability in the control atmosphere resembling the AO/NAO. Results from a coupled modeling study by Koenigk et al. (2009) showed that anomalously high and low Barents Sea ice volume produced localized opposite-signed SLP and surface turbulent heat flux responses corresponding to a direct response. In addition, high Barents Sea ice volume in spring and winter generated large-scale negative NAO-like features consistent with an indirect response.
It is clear from observations and model results that sea ice in the Barents Sea region drives a substantial portion of the atmospheric variability on local and hemispheric scales. However, many of the aforementioned modeling studies have implemented sea ice anomalies that are unrealistically large, remain constant over time, or have opposite signs in the same experiment. Alexander et al. (2004) showed that more realistic sea ice boundary forcing did not generate a significant atmospheric response, but their ice forcing patterns had regionally varying signs. For example, at least one month in the high-SIC case contained negative ice anomalies over the Barents Sea region (see Fig. 1 in Alexander et al. 2004).
Here, we investigate the atmospheric response to sea ice anomalies over the Barents Sea during winter that are 1) exclusively positive or negative, 2) follow observed magnitudes, and 3) decay realistically over time. Daily, rather than monthly, ice boundary forcing is used to include the effects of synoptic-scale variations in sea ice (Dammann et al. 2013). In addition, we use a statistical analysis method involving phase-shifted complex Hilbert empirical orthogonal functions (Strong and Liptak 2012) to detect propagating features in the response patterns.
2. Data and methods
a. Model description
The standalone Community Atmosphere Model (CAM; Neale et al. 2010) and the Community Ice Code (CICE) with a slab ocean (Hunke and Lipscomb 2008) were used to produce continuous control runs and winter (December–February) experimental runs. The CAM and CICE are the atmosphere component and sea ice components of the Community Climate System Model/Community Earth System Model (CCSM/CESM; Gent et al. 2011). Version 4 of the CAM has 26 vertical levels, and was run on a 1.9° × 2.5° grid with a finite volume core (Lin 2004). Version 4 of CICE was run with five discrete ice thickness categories on a 1° displaced-pole grid with the pole centered over Greenland.
b. Experimental design
Default monthly climatological sea ice and SSTs from version 1 of the merged Hadley Centre sea ice and Sea Surface Temperature dataset (HadISST1) and version 2 of the National Oceanic and Atmospheric Administration (NOAA) weekly optimum interpolation (OI) SST data (Hurrell et al. 2008) were used to force a 100-yr continuous CAM control run (A100) initialized on 1 November. Output from A100 was used to produce a 100-yr continuous CICE control run (I100).
An index of daily area-weighted sea ice concentration (SIC) anomalies over the Barents Sea was derived from I100 output during winter (December–February) using the region spanning 70°–82°N and 20°–65°E (black box in Fig. 1). The index was defined as the area-weighted daily SIC anomalies relative to the climatological winter SIC computed from the I100 ensemble mean. The winter containing the most days with index values more than 1 standard deviation (1σ) above the climatological winter mean was selected as sea ice boundary forcing for the high-SIC CAM experiment (AHIGH-SIC, 55 days > 1σ). Likewise, the winter containing the most days with index values more than 1 standard deviation below the climatological winter mean was used to define the SIC boundary forcing for the low-SIC experiment (ALOW-SIC, 50 days < −1σ). The SIC anomalies over the Barents Sea were then superimposed on the winter daily climatological SIC computed from the daily I100 ensemble mean, and these boundary forcing datasets were used to create AHIGH-SIC and ALOW-SIC. The control run (ACLIM) was forced with the daily winter climatological SIC so that the sea ice boundary forcing was identical at all grid points in AHIGH-SIC, ALOW-SIC, and ACLIM except over the Barents Sea. Figure 1 shows the monthly mean values of the daily SIC anomalies for the AHIGH-SIC and ALOW-SIC experiments.
Monthly means (December, January, and February) of the daily SIC anomalies used to force CAM in the (a)–(c) high-SIC and (d)–(f) low-SIC experiments. The black box in (a) outlines the area used to define the Barents Sea SIC index.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
Monthly means (December, January, and February) of the daily SIC anomalies used to force CAM in the (a)–(c) high-SIC and (d)–(f) low-SIC experiments. The black box in (a) outlines the area used to define the Barents Sea SIC index.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
Monthly means (December, January, and February) of the daily SIC anomalies used to force CAM in the (a)–(c) high-SIC and (d)–(f) low-SIC experiments. The black box in (a) outlines the area used to define the Barents Sea SIC index.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
The AHIGH-SIC, ALOW-SIC, and ACLIM sea ice boundary conditions were regridded to the 1.9° × 2.5°CAM grid, and CAM was run for 100 winters initialized with 1 December conditions taken from each year of A100. It should be noted that the atmosphere was not in equilibrium with the sea ice at the time of initialization. We averaged AHIGH-SIC, ALOW-SIC, and ACLIM output across their respective ensemble members for each 6-hourly snapshot during winter, and then defined the positive response by subtracting the ACLIM ensemble mean from the AHIGH-SIC ensemble mean (denoted AHIGH-SIC − ACLIM and referred to as the “high-SIC response”), and likewise for the low-SIC response (denoted ALOW-SIC − ACLIM). The 6-hourly high-SIC and low-SIC response patterns were reduced to daily means and monthly means for different analyses. Statistical significance at the 95% confidence level was determined by bootstrapping the distributions of each anomaly 1000 times using resampling with replacement (e.g., Efron 1979).
Our sea ice boundary forcing datasets contain energy at daily time scales absent in prior studies using monthly mean forcing, but may still underestimate the observed variability in SIC for the following reasons: the monthly mean SSTs and SIC prescribed in A100 may produce atmospheric boundary conditions that dampen SIC variability in I100, and the SIC in I100 lacks the strong trends present in observations.
c. HEOF analysis
To determine the propagating patterns that accounted for the most variability in the SLP responses, the leading two Hilbert empirical orthogonal functions (HEOFs; e.g., Hannachi et al. 2007) were calculated from the covariance matrix of the December–February daily ensemble mean fields of SLP weighted by the square root of the cosine of the latitude over the Northern Hemisphere from 20° to 90°N for AHIGH-SIC − ACLIM and ALOW-SIC − ACLIM. Briefly, the computation of HEOFs follows the procedure used for standard EOFs, with the additional step of applying a Hilbert transform to the covariance matrix to “complexify” the data prior to performing a singular value decomposition. The resultant HEOFs and corresponding principal component time series are complex with associated phase angles. HEOF1 and HEOF2 denote the first and second HEOF spatial patterns, and the associated principal component time series were obtained by projecting the SLP data onto the HEOF coefficients. To clarify the relationship between the HEOFs and their temporal evolution in the two experiments, he phases of the HEOFs were shifted to maximize the correlation between the real parts of the HEOF time series of AHIGH-SIC − ACLIM and ALOW-SIC − ACLIM. Since the phase angle is arbitrary, shifting the phase does not impact the fundamental pattern or explained variance of the HEOF. A detailed explanation of the computation and application of the HEOF phase shift may be found in Strong and Liptak (2012).
3. Results
a. The low-SIC experiment
The low-SIC SLP response depicts a statistically significant localized negative anomaly over the Barents Sea during December (Fig. 2a). The response grows and switches sign, becoming a positive SLP anomaly over the pole in January (Fig. 2b), and then evolves into large-scale features resembling the negative phase of the NAO in February (Fig. 2c) with positive anomalies over the Greenland and Barents Seas and negative SLP anomalies over the North Atlantic region. The positive 500-mb geopotential height anomaly over the Barents Sea in December (Fig. 2d) indicates that the response is initially baroclinic. In January and February, significant 500-mb height anomalies are generally collocated with SLP anomalies of the same sign away from the localized sea ice forcing (Figs. 2e,f). Statistically significant positive (upward) surface turbulent heat flux anomalies (Figs. 3a–c) and anomalous wind stress convergence (Figs. 3d–f) are present over the Barents Sea in the vicinity of the ice anomaly (Figs. 1d–f), decreasing in magnitude and extent as the sea ice forcing weakens throughout the winter.
December, January, and February monthly mean responses of (a)–(c) SLP (mb) and (d)–(f) 500-mb geopotential height (m) for the low-SIC experiment. Solid curves denote positive values and dashed curves denote negative values. The contour intervals are 0.5 mb for SLP and 5 m for 500-mb height, and the zero contours are excluded. Positive and negative anomalies that are significant at the 95% confidence level are shaded.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
December, January, and February monthly mean responses of (a)–(c) SLP (mb) and (d)–(f) 500-mb geopotential height (m) for the low-SIC experiment. Solid curves denote positive values and dashed curves denote negative values. The contour intervals are 0.5 mb for SLP and 5 m for 500-mb height, and the zero contours are excluded. Positive and negative anomalies that are significant at the 95% confidence level are shaded.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
December, January, and February monthly mean responses of (a)–(c) SLP (mb) and (d)–(f) 500-mb geopotential height (m) for the low-SIC experiment. Solid curves denote positive values and dashed curves denote negative values. The contour intervals are 0.5 mb for SLP and 5 m for 500-mb height, and the zero contours are excluded. Positive and negative anomalies that are significant at the 95% confidence level are shaded.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
December, January, and February monthly mean responses of (a)–(c) surface sensible and latent heat flux (W m−2; positive upward) and (d)–(f) surface wind stress (N m−2) for the low-SIC experiment. Solid curves denote positive values and dashed curves denote negative values. The contour interval for the surface flux is 50 W m−2 and the zero contour is excluded. Shading in (a)–(c) indicates total heat flux anomalies that are significant at the 95% confidence level, and shading in (d)–(f) indicates the magnitude of the wind stress anomaly where at least one of its components is significant at the 95% confidence level.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
December, January, and February monthly mean responses of (a)–(c) surface sensible and latent heat flux (W m−2; positive upward) and (d)–(f) surface wind stress (N m−2) for the low-SIC experiment. Solid curves denote positive values and dashed curves denote negative values. The contour interval for the surface flux is 50 W m−2 and the zero contour is excluded. Shading in (a)–(c) indicates total heat flux anomalies that are significant at the 95% confidence level, and shading in (d)–(f) indicates the magnitude of the wind stress anomaly where at least one of its components is significant at the 95% confidence level.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
December, January, and February monthly mean responses of (a)–(c) surface sensible and latent heat flux (W m−2; positive upward) and (d)–(f) surface wind stress (N m−2) for the low-SIC experiment. Solid curves denote positive values and dashed curves denote negative values. The contour interval for the surface flux is 50 W m−2 and the zero contour is excluded. Shading in (a)–(c) indicates total heat flux anomalies that are significant at the 95% confidence level, and shading in (d)–(f) indicates the magnitude of the wind stress anomaly where at least one of its components is significant at the 95% confidence level.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
b. The high-SIC experiment
In December, the high-SIC SLP response shows localized positive anomalies over the Barents Sea (Fig. 4a). The response evolves into a dipole feature with a positive anomaly centered over northern Eurasia and a negative anomaly over the North Atlantic basin in January (Fig. 4b). In February, the negative NAO-like pattern is similar to the low-SIC response, but with the centers of action offset to the west (Fig. 4c). The 500-mb height response over the Barents Sea is not statistically significant in December (Fig. 4d), while the January response depicts a hemispheric wave-2-to-wave-3 pattern (Fig. 4e), and the February response resembles the negative phase of the NAO (Fig. 4f). Negative (downward) surface turbulent heat flux anomalies are present in all months (Figs. 5a–c) over the sea ice anomalies (Figs. 1a–c). The local wind stress response is strong in December (Fig. 5d) with divergence over the Barents Sea, and then becomes dominated by broad larger-scale circulation features in January and February (Figs. 5e,f).
As in Fig. 2, but for the high-SIC experiment.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
As in Fig. 2, but for the high-SIC experiment.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
As in Fig. 2, but for the high-SIC experiment.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
As in Fig. 3, but for the high-SIC experiment.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
As in Fig. 3, but for the high-SIC experiment.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
As in Fig. 3, but for the high-SIC experiment.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
c. Propagating variability associated with the SLP responses
The large-scale indirect responses of the atmosphere to high-SIC and low-SIC forcing appear remarkably similar and negative NAO–like in February, and HEOF analysis helps illuminate how they differ. HEOF1 of the high-SIC SLP response explains 43% of the variance, and its real part (Fig. 6a) shows a broad positive anomaly spanning the pole with centers over Scandinavia and the north coast of Eurasia, and negative anomalies over the midlatitudes. By definition, the imaginary part of HEOF1 (Fig. 6e) depicts the pattern in the real part phase-shifted by π/2, and shows positive anomalies over Eurasia and southeast of Greenland and a strong negative anomaly over the Pacific basin. HEOF1 of the low-SIC response (Figs. 6b,f) explains 29% of the variance, and is similar to the high-SIC response (Figs. 6a,e) with the centers of action displaced slightly equatorward. Incrementing the phases of the HEOFs (not shown) reveals a pattern resembling AO-like behavior in both responses (e.g., Figs. 6a,b) in which the signs and positions of the SLP anomalies oscillate between the polar and midlatitude regions.
Shown are (a),(b) the real part of HEOF1 of SLP, (c),(d) the real part of HEOF2 of SLP, (e),(f) the imaginary part of HEOF1 of SLP, and (g),(h) the imaginary part of HEOF2 of SLP, for AHIGH-SIC − ACLIM and ALOW-SIC − ACLIM, respectively. Solid curves denote positive values and dashed curves denote negative values. The contour interval is arbitrary, and the zero contour is excluded.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
Shown are (a),(b) the real part of HEOF1 of SLP, (c),(d) the real part of HEOF2 of SLP, (e),(f) the imaginary part of HEOF1 of SLP, and (g),(h) the imaginary part of HEOF2 of SLP, for AHIGH-SIC − ACLIM and ALOW-SIC − ACLIM, respectively. Solid curves denote positive values and dashed curves denote negative values. The contour interval is arbitrary, and the zero contour is excluded.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
Shown are (a),(b) the real part of HEOF1 of SLP, (c),(d) the real part of HEOF2 of SLP, (e),(f) the imaginary part of HEOF1 of SLP, and (g),(h) the imaginary part of HEOF2 of SLP, for AHIGH-SIC − ACLIM and ALOW-SIC − ACLIM, respectively. Solid curves denote positive values and dashed curves denote negative values. The contour interval is arbitrary, and the zero contour is excluded.
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
During the first two weeks in December, the magnitudes, or moduli, of both HEOF1 time series (Fig. 7a) are small, reflecting the weakness of the early-winter response at the spatial scale of HEOF1. The moduli of the high-SIC (Fig. 7a, red line) and low-SIC (Fig. 7a, blue line) responses increase between late December and mid-February. The similar phase angles of HEOF1 of the high-SIC (Fig. 7c, red circles) and low-SIC (Fig. 7c, blue circles) responses in December and February indicate that the anomaly patterns that project onto the leading HEOFs (Figs. 6a,b,e,f) are nearly collocated, while divergence in the phase angles during January indicates that the anomaly patterns are offset from one another. Additionally, the phase of HEOF1 of the high-SIC response completes approximately one cycle (i.e., shifts from its initial phase by 2π over the course of the winter), while the phase of the low-SIC response completes two cycles.
The modulus of the principal component time series for (a) HEOF1 and (b) HEOF2 of AHIGH-SIC − ACLIM (red lines) and ALOW-SIC − ACLIM (blue lines). The phase angle of the principal component time series for (c) HEOF1 and (d) HEOF2 of AHIGH-SIC − ACLIM (red circles) and ALOW-SIC − ACLIM (blue circles).
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
The modulus of the principal component time series for (a) HEOF1 and (b) HEOF2 of AHIGH-SIC − ACLIM (red lines) and ALOW-SIC − ACLIM (blue lines). The phase angle of the principal component time series for (c) HEOF1 and (d) HEOF2 of AHIGH-SIC − ACLIM (red circles) and ALOW-SIC − ACLIM (blue circles).
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
The modulus of the principal component time series for (a) HEOF1 and (b) HEOF2 of AHIGH-SIC − ACLIM (red lines) and ALOW-SIC − ACLIM (blue lines). The phase angle of the principal component time series for (c) HEOF1 and (d) HEOF2 of AHIGH-SIC − ACLIM (red circles) and ALOW-SIC − ACLIM (blue circles).
Citation: Journal of Climate 27, 2; 10.1175/JCLI-D-13-00186.1
The second HEOF (HEOF2) of the high-SIC SLP response accounts for 20% of the variance, and depicts zonally oriented wave-2 and wave-3 patterns centered over the northern ocean basins (Figs. 6c,g). HEOF2 of the low-SIC SLP response (Figs. 6d,h) explains 13% of the variance and shows an NAO-like pattern over the North Atlantic basin and a zonal wave train over the Pacific basin. In both cases, the HEOFS are strongest over the area south of approximately 70°N, indicating that the orthogonality constraint may limit the ability of the second HEOFs to fully capture physically realistic wave features.
The differences in the moduli (Fig. 7b) and phases (Fig. 7d) of HEOF2 in January and early February indicate contrast in the temporal evolution of the associated high-SIC and low-SIC anomaly patterns. As in HEOF1, the sign of the SIC anomaly affects the rate of phase change. The phase of HEOF2 of the high-SIC response (Fig. 7d, red circles) completes approximately three cycles per winter, while the phase of HEOF2 of the low-SIC response completes two cycles (Fig. 7d, blue circles).
4. Summary and conclusions
CAM was used to study the atmospheric response to daily high (AHIGH-SIC) and low (ALOW-SIC) sea ice concentration anomalies over the Barents Sea, and complex Hilbert empirical orthogonal function (HEOF) analysis was used to determine the propagating patterns that accounted for the most variability in the SLP responses. In December and January, low-SIC conditions over the Barents Sea locally generated upward turbulent heat flux anomalies that lowered SLP and induced wind stress convergence, while the opposite scenario occurred for the high-SIC conditions. The signs and magnitudes of the local SLP and surface heat flux responses were consistent with the winter responses to high and low Barents Sea ice volume in Koenigk et al. (2009). The surface heat flux and wind stress responses in February were weaker due to the small extent and magnitude of the ice forcing. During February, the large-scale low-SIC SLP and 500-mb height responses resembled the negative phase of the NAO in agreement with the large-scale atmospheric responses to negative Barents Sea ice anomalies in Magnusdottir et al. (2004), Deser et al. (2007), Strong and Magnusdottir (2010), and Seierstad and Bader (2009). Notably, the high-SIC SLP responses in February were remarkably similar to the corresponding low-SIC responses.
The development of the SLP and 500-mb height responses throughout the winter followed that of the atmospheric responses to winter (December–April) monthly varying negative sea ice extent anomalies over the Barents Sea and positive extent anomalies over the Labrador Sea in Deser et al. (2007), beginning in December as localized baroclinic responses centered over the ice anomalies and transitioning into hemispheric patterns resembling the negative AO/NAO by February. The December responses were similar to what Deser et al. (2004) referred to as the direct response of the atmosphere to diabatic forcing, while the February responses were similar to the indirect response (i.e., the AO/NAO). In our study, the generation of opposite-signed SLP responses by the high-SIC and low-SIC boundary forcing during December was consistent with the opposite-signed low-level direct responses to warm and cold SST anomalies southeast of Greenland in Deser et al. (2004). The high-SIC and low-SIC responses in February resembled the negative AO/NAO pattern present in the indirect response to negative SST and sea ice extent anomalies in Deser et al. (2004) and Deser et al. (2007).
HEOFs of ensemble mean high-SIC and low-SIC SLP responses indicated that the sign of the ice boundary forcing affected the spatiotemporal phase of propagating wave features, which is reflected in the displacement between the centers of action in the February high-SIC and low-SIC SLP and 500-mb height responses. The leading HEOF (HEOF1) of the SLP responses depicted variability associated with hemispheric waves resembling the AO. Transitions in the phases of the HEOF1 time series indicated shifts in the positions and signs of SLP anomalies over the pole and midlatitudes in both responses. HEOF1 of the high-SIC response completed one cycle during the winter, while HEOF1 of the low-SIC response completed two cycles. Determination of mechanisms by which changes in Barents Sea ice cover affect the phase of the large-scale circulation features is beyond the scope of this study. However, one possible explanation is that Barents Sea ice anomalies induce synoptic-scale wave energy flux anomalies that alter the progression and location of large-scale waves through shifts in the zonal wind as indicated by Sokolova et al. (2007). The overall similarity of the responses to opposite surface forcing motivates more detailed study of the underlying nonlinear atmospheric dynamics as in Deser et al. (2007) and Strong and Magnusdottir (2010).
HEOF2 showed wave-2 and wave-3 patterns over the Atlantic and Pacific basins that propagated east–west in the high-SIC response, and north–south in the low-SIC response. HEOF2 of the high-SIC response completed three cycles per winter and the phase of the low-SIC response completed two cycles; hence, the SLP anomaly pattern associated with HEOF2 cycled more quickly when ice was added to the Barents Sea than when ice was removed. However, possible effects of the orthogonality constraint on the physical interpretation of HEOF2 were noted.
The results here focus on the modification of the atmosphere by anomalous Barents Sea ice cover through changes in the surface turbulent heat fluxes. The atmosphere, in turn, influences the growth and retreat of sea ice via surface wind stress forcing and thermodynamic effects stemming in part from temperature advection. Research is underway to determine how the atmospheric responses to the Barents Sea ice anomalies feed back onto the sea ice by forcing the CICE model with atmospheric data and initial conditions derived from the high-SIC and low-SIC experiments analyzed here.
Acknowledgments
This research was supported by the National Science Foundation Arctic Sciences Division Grant 1022485. Provision of computer infrastructure by the Center for High Performance Computing at the University of Utah is gratefully acknowledged. The authors thank three anonymous reviewers for their insightful comments.
REFERENCES
Alexander, M. A., U. S. Bhatt, J. E. Walsh, M. S. Timlin, J. S. Miller, and J. D. Scott, 2004: The atmospheric response to realistic Arctic sea ice anomalies in an AGCM during winter. J. Climate, 17, 890–905.
Dammann, D. O., U. S. Bhatt, P. L. Langen, J. R. Krieger, and X. Zhang, 2013: Impact of daily Arctic sea ice variability in CAM3.0 during fall and winter. J. Climate, 26, 1939–1955.
Deser, C., J. E. Walsh, and M. S. Timlin, 2000: Arctic sea ice variability in the context of recent atmospheric circulation trends. J. Climate, 13, 617–633.
Deser, C., G. Magnusdottir, R. Saravanan, and A. S. Phillips, 2004: The effects of North Atlantic SST and sea ice anomalies on the winter circulation in CCM3. Part II: Direct and indirect components of the response. J. Climate, 17, 877–889.
Deser, C., R. A. Tomas, and S. Peng, 2007: The transient atmospheric circulation response to North Atlantic SST and sea ice anomalies. J. Climate, 20, 4751–4767.
Efron, B., 1979: Bootstrap methods: Another look at the jackknife. Ann. Stat., 7, 1–26.
Gent, P. R., and Coauthors, 2011: The Community Climate System Model version 4. J. Climate, 24, 4973–4991.
Hannachi, A., I. T. Jolliffe, and D. B. Stephenson, 2007: Empirical orthogonal functions and related techniques in atmospheric science: A review. Int. J. Climatol., 27, 1119–1152, doi:10.1002/joc.1499.
Honda, M., K. Yamazaki, H. Nakamura, and K. Takeuchi, 1999: Dynamic and thermodynamic characteristics of atmospheric response to anomalous sea-ice extent in the Sea of Okhotsk. J. Climate, 12, 3347–3358.
Honda, M., J. Inoue, and S. Yamane, 2009: Influence of low arctic sea-ice minima on anomalously cold Eurasian winters. Geophys. Res. Lett.,36, L08707, doi:10.1029/2008GL037079.
Hunke, E. C., and W. H. Lipscomb, 2008: CICE: The Los Alamos sea ice model user’s manual, version 4.1. Los Alamos National Laboratory Tech. Rep. LA-CC-06-012, 76 pp. [Available online at http://oceans11.lanl.gov/trac/CICE/.]
Hurrell, J. W., Y. Kushnir, G. Ottersen, and M. Visbeck, 2003: The North Atlantic Oscillation: Climate Significance and Environmental Impact.Geophys. Monogr., Vol. 134, Amer. Geophys. Union, 279 pp.
Hurrell, J. W., J. J. Hack, D. Shea, J. M. Caron, and J. Rosinski, 2008: A new sea surface temperature and sea ice boundary dataset for the Community Atmosphere Model. J. Climate, 21, 5145–5153.
Jaiser, R., K. Dethloff, D. Handorf, A. Rinke, and J. Cohen, 2012: Impact of sea ice cover changes on the Northern Hemisphere atmospheric winter circulation. Tellus, 64A, 11595, doi:10.3402/tellusa.v64i0.11595.
Koenigk, T., U. Mikolajewicz, J. H. Jungclaus, and A. Kroll, 2009: Sea ice in the Barents Sea: Seasonal to interannual variability and climate feedbacks in a global coupled model. Climate Dyn., 32, 1119–1138.
Lin, S.-J., 2004: A “vertically Lagrangian” finite-volume dynamical core for global models. Mon. Wea. Rev., 132, 2293–2307.
Liu, J., J. A. Curry, H. Wang, M. Song, and R. M. Horton, 2012: Impact of declining Arctic sea ice on winter snowfall. Proc. Natl. Acad. Sci. USA,109, 4074–4079, doi:10.1073/pnas.1114910109.
Magnusdottir, G., C. Deser, and R. Saravanan, 2004: The effects of North Atlantic SST and sea ice anomalies on the winter circulation in CCM3. Part I: Main features and storm track characteristics of the response. J. Climate, 17, 857–876.
Neale, R. B., and Coauthors, 2010: Description of the NCAR Community Atmosphere Model (CAM 4.0). NCAR Tech. Rep. NCAR/TN-485+STR, 212 pp.
Petoukhov, V. and V. A. Semenov, 2010: A link between reduced Barents-Kara sea ice and cold winter extremes over northern continents. J. Geophys. Res.,115, D21111, doi:10.1029/2009JD013568.
Rigor, I. G., J. M. Wallace, and R. L. Colony, 2002: Response of sea ice to the Arctic Oscillation. J. Climate, 15, 2648–2663.
Seierstad, I., and J. Bader, 2009: Impact of a projected future Arctic sea ice reduction on extratropical storminess and the NAO. Climate Dyn., 33, 937–943, doi:10.1007/s00382-008-0463-x.
Sokolova, E., K. Dethloff, A. Rinke, and A. Benkel, 2007: Planetary and synoptic scale adjustment of the Arctic atmosphere to sea ice cover changes. Geophys. Res. Lett.,34, L17816, doi:10.1029/2007GL030218.
Strong, C., and G. Magnusdottir, 2010: Dependence of NAO variability on coupling with sea ice. Climate Dyn., 36, 1681–1689, doi:10.1007/s00382-010-0752-z.
Strong, C., and J. Liptak, 2012: Propagating atmospheric patterns associated with winter Midwest precipitation. J. Hydrometeor., 13, 1371–1382.
Strong, C., G. Magnusdottir, and H. Stern, 2009: Observed feedback between winter sea ice and the North Atlantic Oscillation. J. Climate, 22, 6021–6032.
Thompson, D. W. J., and J. M. Wallace, 1998: The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., 25, 1297–1300.
Vinje, T., 2001: Anomalies and trends in sea-ice extent and atmospheric circulation in the Nordic Seas during the period 1864–1998. J. Climate, 14, 255–267.
Wu, Q., and X. Zhang, 2010: Observed forcing-feedback processes between Northern Hemisphere atmospheric circulation and Arctic sea ice coverage. J. Geophys. Res.,115, D14119, doi:10.1029/2009JD013574.
Yamamoto, K., Y. Tachibana, M. Honda, and J. Ukita, 2006: Intra-seasonal relationship between the Northern Hemisphere sea ice variability and the North Atlantic Oscillation. Geophys. Res. Lett.,33, L14711, doi:10.1029/2006GL026286.
Yang, S., and J. H. Christensen, 2012: Arctic sea ice reduction and European cold winters in CMIP5 climate change experiments. Geophys. Res. Lett., 39, L20707, doi:10.1029/2012GL053338.
