• 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 , 890905.

    • Search Google Scholar
    • Export Citation
  • Bhatt, U. S., M. A. Alexander, C. Deser, J. E. Walsh, J. S. Miller, M. Timlin, J. D. Scott, and R. Tomas, 2008: The atmospheric response to realistic reduced summer Arctic sea ice anomalies. Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications, Geophys. Monogr., Vol. 180, Amer. Geophys. Union, 91–110.

    • Search Google Scholar
    • Export Citation
  • Boé, J., A. Hall, and X. Qu, 2009: Current GCMs’ unrealistic negative feedback in the Arctic. J. Climate, 22 , 46824695.

  • Boville, B. A., P. J. Rasch, J. J. Hack, and J. R. McCaa, 2006: Representation of clouds and precipitation processes in the Community Atmosphere Model Version 3 (CAM3). J. Climate, 19 , 21842198.

    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., and D. S. Battisti, 2000: An interpretation of the results from atmospheric general circulation models forced by the time history of the observed sea surface temperature distribution. Geophys. Res. Lett., 27 , 767770.

    • Search Google Scholar
    • Export Citation
  • Collins, W. D., and Coauthors, 2006a: The Community Climate System Model Version 3 (CCSM3). J. Climate, 19 , 21222143.

  • Collins, W. D., and Coauthors, 2006b: The formulation and atmospheric simulation of the Community Atmosphere Model Version 3 (CAM3). J. Climate, 19 , 21442161.

    • Search Google Scholar
    • Export Citation
  • Comiso, J. C., C. L. Parkinson, R. Gersten, and L. Stock, 2008: Accelerated decline in the Arctic sea ice cover. Geophys. Res. Lett., 35 , L01703. doi:10.1029/2007GL031972.

    • Search Google Scholar
    • Export Citation
  • Deser, C., and H. Teng, 2008: Recent trends in Arctic sea ice and the evolving role of atmospheric circulation forcing, 1979–2007. Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications, Geophys. Monogr., Vol. 180, Amer. Geophys. Union, 7–26.

    • Search Google Scholar
    • Export Citation
  • 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 , 617633.

    • Search Google Scholar
    • Export Citation
  • 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 , 877889.

    • Search Google Scholar
    • Export Citation
  • 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 , 47514767.

    • Search Google Scholar
    • Export Citation
  • Gerdes, R., 2006: Atmospheric response to changes in Arctic sea ice thickness. Geophys. Res. Lett., 33 , L18709. doi:10.1029/2006GL027146.

    • Search Google Scholar
    • Export Citation
  • Hack, J. J., J. M. Caron, G. Danabasoglu, K. W. Oleson, C. Bitz, and J. Truesdale, 2006: CCSM–CAM3 climate simulation sensitivity to changes in horizontal resolution. J. Climate, 19 , 22672289.

    • Search Google Scholar
    • Export Citation
  • Hall, N. M. J., J. Derome, and H. Lin, 2001: The extratropical signal generated by a midlatitude SST anomaly. Part I: Sensitivity at equilibrium. J. Climate, 14 , 20352053.

    • Search Google Scholar
    • Export Citation
  • Holland, M. M., C. M. Bitz, and B. Tremblay, 2006: Future abrupt reductions in the summer Arctic sea ice. Geophys. Res. Lett., 33 , L23503. doi:10.1029/2006GL028024.

    • Search Google Scholar
    • Export Citation
  • Holton, J. R., 2004: An Introduction to Dynamic Meteorology. Academic Press, 535 pp.

  • 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 , 33473358.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., and D. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38 , 11791196.

    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation: Regional temperatures and precipitation. Science, 269 , 676679.

    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., J. J. Hack, A. S. Phillips, J. Caron, and J. Yin, 2006: The Dynamical Simulation of the Community Atmosphere Model Version 3 (CAM3). J. Climate, 19 , 21622183.

    • Search Google Scholar
    • Export Citation
  • Kvamstø, N. G., P. Skeie, and D. B. Stephenson, 2004: Impact of Labrador sea-ice extent on the North Atlantic Oscillation. Int. J. Climatol., 24 , 603612.

    • Search Google Scholar
    • Export Citation
  • Lau, N-C., and E. O. Holopainen, 1984: Transient eddy forcing of the time-mean flow as identified by geopotential tendencies. J. Atmos. Sci., 41 , 313328.

    • Search Google Scholar
    • Export Citation
  • Lawrence, D. M., A. G. Slater, R. A. Tomas, M. M. Holland, and C. Deser, 2008: Accelerated Arctic land warming and permafrost degradation during rapid sea ice loss. Geophys. Res. Lett., 35 , L11506. doi:10.1029/2008GL033985.

    • Search Google Scholar
    • Export Citation
  • 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 , 857876.

    • Search Google Scholar
    • Export Citation
  • Meehl, G. A., and Coauthors, 2006: Climate change projections for the twenty-first century and climate change commitment in the CCSM3. J. Climate, 19 , 25972616.

    • Search Google Scholar
    • Export Citation
  • Peng, S. L., W. A. Robinson, and M. P. Hoerling, 1997: The modeled atmospheric response to midlatitude SST anomalies and its dependence on background circulation states. J. Climate, 10 , 971987.

    • Search Google Scholar
    • Export Citation
  • Post, E., and M. C. Forchhammer, 2008: Climate change reduces reproductive success of an Arctic herbivore through trophic mismatch. Philos. Trans. Roy. Soc. London, B363 , 23672373.

    • Search Google Scholar
    • Export Citation
  • Seierstad, I. A., and J. Bader, 2009: Impact of a projected future Arctic sea ice reduction on extratropical storminess and the NAO. Climate Dyn., 33 , 937943. doi:10.1007/s00382-008-0463-x.

    • Search Google Scholar
    • Export Citation
  • Serreze, M. C., M. M. Holland, and J. Stroeve, 2007: Perspectives on the Arctic’s shrinking sea-ice cover. Science, 315 , 15331536.

  • Serreze, M. C., A. P. Barrett, J. C. Stroeve, D. N. Kindig, and M. M. Holland, 2008: The emergence of surface-based Arctic amplification. Cryosphere Discuss., 2 , 601622.

    • Search Google Scholar
    • Export Citation
  • Sewall, J. O., 2005: Precipitation shifts over western North America as a result of declning Arctic sea ice cover, the coupled system response. Earth Interactions, 9 .[Available online at http://EarthInteractions.org].

    • Search Google Scholar
    • Export Citation
  • Singarayer, J. S., J. L. Bamber, and P. J. Valdes, 2006: Twenty-first-century climate impacts from a declining Arctic sea ice cover. J. Climate, 19 , 11091125.

    • Search Google Scholar
    • Export Citation
  • Solomon, S., D. Qin, M. Manning, M. Marquis, K. Averyt, M. M. B. Tignor, H. L. Miller Jr., and Z. Chen, Eds.,. 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

    • Search Google Scholar
    • Export Citation
  • Stroeve, J., M. M. Holland, W. Meier, T. Scambos, and M. C. Serreze, 2007: Arctic sea ice decline: Faster than forecast. Geophys. Res. Lett., 34 , L09501. doi:10.1029/2007GL029703.

    • Search Google Scholar
    • Export Citation
  • Vavrus, S., and D. Waliser, 2008: An improved parameterization for simulating Arctic cloud amount in the CCSM3 climate model. J. Climate, 21 , 56735687.

    • Search Google Scholar
    • Export Citation
  • Wallace, J. M., 2000: North Atlantic Oscillation/annular mode: Two paradigms–one phenomenon. Quart. J. Roy. Meteor. Soc., 126 , 791805.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Bimonthly distributions of Arctic (a) sea ice concentration (%) and (b) sea ice thickness (m) during 1980–99 and 2080–99 from CCSM3.

  • View in gallery

    Bimonthly Arctic (top row) sea ice thickness (ΔSIT; m) and (second row) concentration (ΔSIC; %) differences (2080–99 minus 1980–99) from CCSM3. Bimonthly (third row) turbulent energy flux (ΔSH + LH), (fourth row) longwave radiative flux (ΔLW), and (bottom row) shortwave radiative flux (ΔSW) responses to sea ice cover changes. Fluxes (W m−2) are positive upward.

  • View in gallery

    Seasonal cycle of the turbulent energy flux (Wm−2; ΔSH + LH; thin solid curve), longwave radiative flux (ΔLW; dotted curve), and shortwave radiative flux (ΔSW; dashed curve) responses area averaged over the Arctic Ocean. The net surface energy flux response is given by the thick solid curve, and the SIC changes (%) are indicated by the gray bars (note the inverted scale). Fluxes are positive upward.

  • View in gallery

    Seasonal cycles of air temperature (°C; dotted curve) and precipitation (mm day−1; dashed curve) responses area averaged over (a) the Arctic Ocean and (b) the high-latitude continents (65°–80°N; 60°–300°E). The solid curve in both (a) and (b) shows the sum of the turbulent and longwave fluxes area averaged over the Arctic Ocean (W m−2). SIC changes are indicated by the gray bars (scale as in Fig. 3, not shown).

  • View in gallery

    Bimonthly responses of net surface energy flux (ΔQnet; W m−2), terrestrial air temperature (ΔAir T; °C), terrestrial snow depth (ΔSnow; cm liquid water equivalent), and terrestrial precipitation (ΔPrecip; mm day−1). Color scales are given at the bottom of the figure. Thick black contours on the air temperature panels outline regions with a low-level temperature inversion (T850hPa − T1000hPa > 0°C) during 1980–99. Precipitation responses significant at the 5% confidence level are outlined with black contours.

  • View in gallery

    Vertical profiles of atmospheric temperature during 1980–99 (dashed curves) and 2080–99 (solid curves) over (left) the Arctic Ocean and (center) the high-latitude (65°–80°N) continents in December. (right) The 2080–99 minus 1980–99 differences over the Arctic Ocean (thick curve) and high-latitude continents (thin curve).

  • View in gallery

    Geographical distributions of the strength of the December low-level inversion (T850hPa − T1000hPa > 0°C) during (left) 1980–99 and (right) 2080–99. Values <1.5°C not shown.

  • View in gallery

    Seasonal cycles of T850hPa − T1000hPa during 1980–99 (dark colors) and 2080–99 (light colors) over the (top) Arctic Ocean (blue) and (middle) high-latitude continents (red). (bottom) The 2080–99 minus 1980–99 differences for the Arctic Ocean (blue) and high-latitude continents (red).

  • View in gallery

    Accumulated cold season (October–March) precipitation and March snow depth responses (cm). The thick black and gray contours on the snow panel denote the −10° and 0°C air temperature isotherms, respectively, during 2080–99.

  • View in gallery

    Seasonal cycles of precipitation (mm day−1; dashed curve), accumulated precipitation (cm; dotted curve), and snow depth (cm liquid water equivalent; solid curve) responses for (a) northern Canada and (b) Siberia.

  • View in gallery

    Heat budget response (°C day−1) for the (left) Arctic Ocean and (right) high-latitude (65°–80°N) continents. Submonthly transient temperature advection (solid black curve); monthly mean temperature advection (dashed black curve); condensational heating (dashed purple curve); vertical and horizontal diffusion (green curve); longwave radiation (solid purple curve). The time-tendency and shortwave radiation terms are negligible (not shown). The sum of all the terms (e.g., the heat budget residual) is given by the orange curve.

  • View in gallery

    Bimonthly geopotential height responses at 1000 and 500 hPa. The contour interval is 10 m, with positive (negative) values in red (blue) and the zero contours omitted. Shading indicates values that exceed the 5% confidence level based on a two-sided Student’s t test.

  • View in gallery

    Vertical structures of the geopotential height (m; contours) and temperature (°C; color shading) responses along 90°E in (left) November–December and (right) January–February.

  • View in gallery

    (a) Bimonthly terrestrial air temperature responses (°C) for (top) CCSM3 and (bottom) CCSM3 minus CAM3. (b) Seasonal cycle of the high-latitude (65°–80°N) terrestrial air temperature response in CCSM3 (dashed), CAM3 (solid), and their difference (dotted). (c) Vertical profiles of the December high-latitude (65°–80°N) terrestrial air temperature response in CCSM3 (dashed), CAM3 (solid), and their difference (dotted).

  • View in gallery

    Accumulated cold season (October–March) precipitation and March snow depth responses (cm) in CCSM3. The thick black and gray contours on the snow panel denote the −10° and 0°C air temperature isotherms, respectively, during 2080–99.

  • View in gallery

    Bimonthly SLP responses for CAM3 and CCSM3. The contour interval is 1 hPa, with positive (negative) values in red (blue) and the zero contours omitted. Shading indicates values that exceed the 5% confidence level based on a two-sided Student’s t test.

  • View in gallery

    As in Fig. 3 but for CCSM3. The net surface energy flux response curve is omitted.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 849 789 52
PDF Downloads 587 541 31

The Seasonal Atmospheric Response to Projected Arctic Sea Ice Loss in the Late Twenty-First Century

View More View Less
  • 1 National Center for Atmospheric Research,* Boulder, Colorado
  • | 2 NOAA/Earth System Research Laboratory, Boulder, Colorado
  • | 3 National Center for Atmospheric Research,* Boulder, Colorado
© Get Permissions
Full access

Abstract

The authors investigate the atmospheric response to projected Arctic sea ice loss at the end of the twenty-first century using an atmospheric general circulation model (GCM) coupled to a land surface model. The response was obtained from two 60-yr integrations: one with a repeating seasonal cycle of specified sea ice conditions for the late twentieth century (1980–99) and one with that of sea ice conditions for the late twenty-first century (2080–99). In both integrations, a repeating seasonal cycle of SSTs for 1980–99 was prescribed to isolate the impact of projected future sea ice loss. Note that greenhouse gas concentrations remained fixed at 1980–99 levels in both sets of experiments. The twentieth- and twenty-first-century sea ice (and SST) conditions were obtained from ensemble mean integrations of a coupled GCM under historical forcing and Special Report on Emissions Scenarios (SRES) A1B scenario forcing, respectively.

The loss of Arctic sea ice is greatest in summer and fall, yet the response of the net surface energy budget over the Arctic Ocean is largest in winter. Air temperature and precipitation responses also maximize in winter, both over the Arctic Ocean and over the adjacent high-latitude continents. Snow depths increase over Siberia and northern Canada because of the enhanced winter precipitation. Atmospheric warming over the high-latitude continents is mainly confined to the boundary layer (below ∼850 hPa) and to regions with a strong low-level temperature inversion. Enhanced warm air advection by submonthly transient motions is the primary mechanism for the terrestrial warming. A significant large-scale atmospheric circulation response is found during winter, with a baroclinic (equivalent barotropic) vertical structure over the Arctic in November–December (January–March). This response resembles the negative phase of the North Atlantic Oscillation in February only. Comparison with the fully coupled model reveals that Arctic sea ice loss accounts for most of the seasonal, spatial, and vertical structure of the high-latitude warming response to greenhouse gas forcing at the end of the twenty-first century.

Corresponding author address: Dr. Clara Deser, Climate and Global Dynamics Division, NCAR, P.O. Box 3000, Boulder CO 80307. Email: cdeser@ucar.edu

Abstract

The authors investigate the atmospheric response to projected Arctic sea ice loss at the end of the twenty-first century using an atmospheric general circulation model (GCM) coupled to a land surface model. The response was obtained from two 60-yr integrations: one with a repeating seasonal cycle of specified sea ice conditions for the late twentieth century (1980–99) and one with that of sea ice conditions for the late twenty-first century (2080–99). In both integrations, a repeating seasonal cycle of SSTs for 1980–99 was prescribed to isolate the impact of projected future sea ice loss. Note that greenhouse gas concentrations remained fixed at 1980–99 levels in both sets of experiments. The twentieth- and twenty-first-century sea ice (and SST) conditions were obtained from ensemble mean integrations of a coupled GCM under historical forcing and Special Report on Emissions Scenarios (SRES) A1B scenario forcing, respectively.

The loss of Arctic sea ice is greatest in summer and fall, yet the response of the net surface energy budget over the Arctic Ocean is largest in winter. Air temperature and precipitation responses also maximize in winter, both over the Arctic Ocean and over the adjacent high-latitude continents. Snow depths increase over Siberia and northern Canada because of the enhanced winter precipitation. Atmospheric warming over the high-latitude continents is mainly confined to the boundary layer (below ∼850 hPa) and to regions with a strong low-level temperature inversion. Enhanced warm air advection by submonthly transient motions is the primary mechanism for the terrestrial warming. A significant large-scale atmospheric circulation response is found during winter, with a baroclinic (equivalent barotropic) vertical structure over the Arctic in November–December (January–March). This response resembles the negative phase of the North Atlantic Oscillation in February only. Comparison with the fully coupled model reveals that Arctic sea ice loss accounts for most of the seasonal, spatial, and vertical structure of the high-latitude warming response to greenhouse gas forcing at the end of the twenty-first century.

Corresponding author address: Dr. Clara Deser, Climate and Global Dynamics Division, NCAR, P.O. Box 3000, Boulder CO 80307. Email: cdeser@ucar.edu

1. Introduction

Arctic sea ice extent has declined over the past several decades, with the largest rate of retreat (∼−10% decade−1) in late summer (Serreze et al. 2007; Comiso et al. 2008; Deser and Teng 2008; among others). The rate of decline has accelerated substantially in the past decade and now outpaces that simulated by most climate models in response to increasing greenhouse gas (GHG) concentrations (Stroeve et al. 2007). The record losses of perennial Arctic sea ice in both 2007 and 2008 highlight the ongoing trajectory toward ice-free summers, a state that climate models project may be realized within 15–50 yr (Holland et al. 2006).

A seasonally ice-free Arctic Ocean is expected to have widespread socioeconomic, ecological, and climatic consequences. For example, commercial shipping routes and energy resource development in the Arctic are likely to change, impacting native populations and habitat. Warming associated with Arctic sea ice loss may hasten permafrost degradation (Lawrence et al. 2008) and lead to trophic mismatch (Post and Forchhammer 2008). Atmospheric circulation patterns and accompanying precipitation and storm-track distributions over mid- and high latitudes may also be affected (Sewall 2005; Singarayer et al. 2006; Gerdes 2006; Seierstad and Bader 2009).

Identification of the climatic impacts of a seasonally ice-free Arctic Ocean from observational data alone is difficult because of a variety of factors, including 1) that the loss of perennial Arctic sea ice is not yet complete, 2) the confounding presence of climate variability due to internal atmospheric processes and/or forced by factors other than Arctic sea ice loss, and 3) that observational data contain a mixture of forcing and response. To circumvent these difficulties, we use an atmospheric modeling approach to address the atmospheric response to projected future Arctic sea ice loss. The model results may be useful for informing observational attribution studies of the climatic effects of Arctic sea ice retreat. For example, is the enhanced Arctic warming in autumn over the past few years an early signal of the emerging climate response to Arctic sea ice loss as proposed by Serreze et al. (2008)?

Many studies have used atmospheric general circulation models to investigate the effects of prescribed changes in Arctic sea ice cover upon the atmosphere. The specified sea ice extent or concentration conditions range from realistic values for winter (Alexander et al. 2004; Singarayer et al. 2006) and summer (Bhatt et al. 2008) to more idealized configurations (realistic spatial patterns with exaggerated amplitudes) within the Atlantic (Magnusdottir et al. 2004; Deser et al. 2004; Kvamstø et al. 2004) and Pacific (Honda et al. 1999) sectors in winter. Focusing on dynamical aspects, these studies reported a range of atmospheric circulation responses depending on the location and polarity of the sea ice changes as well as time of year. For the Atlantic sea ice cases, the winter circulation response was found to resemble the North Atlantic Oscillation (NAO; Hurrell 1995) or northern annular mode (NAM; Wallace 2000), the leading structure of internal atmospheric variability over the extratropical Northern Hemisphere. For the Pacific sea ice case, the winter circulation response consisted of a stationary Rossby wave train downstream of the Sea of Okhotsk and extending across the Pacific into North America. The summer circulation response to sea ice changes within the Arctic Ocean displayed a significant remote effect over the North Pacific.

In addition to present-day sea ice conditions, projected future changes in Arctic sea ice concentrations at the end of the twenty-first century have been prescribed as boundary conditions to atmospheric general circulation models. Singarayer et al. (2006) employed idealized scenarios of future Arctic sea ice loss, while Seierstad and Bader (2009) used Arctic sea ice loss projections taken from coupled model simulations driven by increasing GHG concentrations. The former study focused upon the wintertime (December–February average) responses in circulation, precipitation, and surface air temperature, while the latter was concerned mainly with the circulation response, contrasting December–February and March. Singarayer et al. (2006) reported reductions in sea level pressure over the Arctic, northeastern Canada, and the Bering Sea, accompanied by local increases in precipitation and air temperature. Seierstad and Bader (2009) showed a positive 500-hPa geopotential height response over the Atlantic sector of the Arctic in December–February and an amplified circulation response in March that resembles the negative polarity of the NAO/NAM. Because of a lack of information on the vertical structure of the circulation responses in the two studies, it is difficult to make a direct comparison between them.

The present study aims to provide a more comprehensive assessment of the atmospheric response to projected Arctic sea ice loss at the end of the twenty-first century than earlier studies, including aspects not previously addressed in detail such as seasonal dependence and vertical structure. For example, we document the full seasonal cycle of the three-dimensional circulation response to future Arctic sea ice loss. Another focus of our study is the terrestrial climate response to future Arctic sea ice loss, including effects on air temperature, precipitation, and snow depth. We also investigate the role of atmospheric boundary layer stability in determining the geographical distribution and vertical structure of the terrestrial climate response. In addition to terrestrial impacts, the seasonal cycle of the response of the Arctic Ocean surface energy balance is investigated. Finally, we compare the atmospheric response to future Arctic sea ice loss with the atmospheric response to GHG forcing in a fully coupled climate model in order to assess the relative role of Arctic sea ice loss in future climate change. Early results on the terrestrial air temperature response were presented in Lawrence et al. (2008).

We note that our atmospheric modeling approach is designed to isolate the direct impact of future sea ice loss upon the atmosphere without accounting for feedbacks from the oceans or other components of the climate system. As such, it may be used as a baseline for evaluating the impact of Arctic sea ice loss upon the atmosphere.

The paper is organized as follows. The model and experimental design are described in section 2. Results are presented in section 3. Summary and discussion are given in sections 4 and 5, respectively.

2. Model experiments

To address the impacts of projected future changes in Arctic sea ice cover upon the global atmospheric circulation and climate, we have conducted two experiments with the National Center for Atmospheric Research (NCAR) Community Atmospheric Model Version 3 (CAM3), an atmospheric general circulation model with a horizontal resolution of ∼1.4° latitude and 1.4° longitude (85-wave triangular spectral truncation; T85) and 26 vertical levels, coupled to the Community Land Model (CLM). CAM3 is the atmospheric component of one of the principal climate models used in the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (Solomon et al. 2007). The physical and numerical methods used in CAM3 are documented in Collins et al. (2006a) and references therein. The most important differences from the previous version of the NCAR atmospheric model are associated with changes to the parameterized physics package, notably the representation of cloud and precipitation processes (Boville et al. 2006) and the treatment of aerosols and radiative processes (Collins et al. 2006b).

The CAM3 “control” experiment consists of a 60-yr integration with a specified repeating seasonal cycle of SSTs and sea ice (concentration and thickness) for the period 1980–99, obtained from the 7-member ensemble mean of twentieth-century Community Climate System Model, version 3 (CCSM3) simulations at T85 resolution. CCSM3 is a fully coupled climate model comprised of CAM3, CLM, the Parallel Ocean Program model, and the Community Sea Ice Model version 4. The CCSM3 twentieth-century simulations are forced with observed estimates of time-varying atmospheric chemical composition (greenhouse gases, tropospheric and stratospheric ozone, and sulfate and volcanic aerosols) and solar output as described in Meehl et al. (2006), and are conducted without flux adjustments.

The CAM3 “perturbation” experiment consists of a 60-yr integration with a repeating seasonal cycle of Arctic sea ice (concentration and thickness) for the period 2080–99, taken from the 8-member ensemble mean of twenty-first-century CCSM3 simulations at T85 resolution under the Special Report on Emissions Scenarios (SRES) A1B greenhouse gas forcing scenario. For the CAM3 experiments with prescribed sea ice conditions for the late twenty-first century, SSTs are set to those in the control experiment so as to isolate the impact of the sea ice changes. At grid boxes where fractional sea ice cover in the late twentieth century is replaced by open water in the late twenty-first century, SSTs are set to the freezing point of seawater (∼−1.8°C). A two-sided Student’s t test is used to evaluate the statistical significance of the atmospheric changes between the 60-yr averages of the control and perturbation experiments.

Our experimental design is similar to the studies of Singarayer et al. (2006) and Seierstad and Bader (2009) except 1) we do not include any changes in SSTs, 2) we include changes in sea ice thickness (in addition to concentration), 3) our ensemble size is considerably larger, and 4) the horizontal resolution of our atmospheric GCM is approximately double.

As with all climate model experiments, model biases must be kept in mind when evaluating the results. CAM3 has been extensively compared against observations (see the special issue of the Journal of Climate, 2006, Vol. 19, No. 11). In particular, the dynamical and hydrological aspects have been evaluated in Hurrell et al. (2006) and Hack et al. (2006). These studies show that the model reproduces well the main features of the seasonal cycle of the extratropical atmospheric circulation, although the simulated sea level pressures are too high over midlatitudes and too low over high latitudes. In particular, the Icelandic low is too deep and extends too far over Eurasia and the Arctic Basin compared to observations. In addition, CAM3 overestimates the amount of low stratus clouds and precipitation over the Arctic Ocean in winter, a common bias among current-generation climate models (Vavrus and Waliser 2008). The strength of the mean low-level temperature inversion in winter over the Arctic and adjacent high-latitude continents is also overestimated in CAM3 as in many other models (Boé et al. 2009).

3. Results

a. Arctic sea ice concentration and thickness

The prescribed sea ice concentration (SIC) and sea ice thickness (SIT) distributions for the late-twentieth-century (1980–99) and late-twenty-first-century (2080–99) CAM3 experiments, taken from CCSM3, are shown in Fig. 1 as bimonthly averages (January–February, March–April, May–June, July–August, September–October, and November–December). The most dramatic loss of Arctic sea ice between the late twentieth and twenty-first centuries occurs during summer, with a nearly ice-free Arctic Ocean projected during August–October. Although largest in summer, the loss of sea ice occurs year-round as the ice edge retreats from the peripheral Arctic seas. The areal reduction in Arctic sea ice is accompanied by a thinning of the ice pack. SIT in the central Arctic Ocean decreases from 3–4 m to 0.5–1 m in winter and from 2.5–3.5 m to <0.5 m in summer. The late-twentieth-century SIC and SIT distributions are generally realistic compared to the available observations (Holland et al. 2006).

The bimonthly changes in SIT and SIC between the late twentieth and twenty-first centuries are shown in the top two rows of Fig. 2. The magnitude and pattern of sea ice thinning is relatively uniform throughout the year, with maximum values ∼2.5–3.5 m in the central Arctic Ocean. In contrast, the reductions in SIC are seasonally dependent, with the largest decreases (∼80%–90%) within the central Arctic Ocean in summer (September–October) and smaller decreases (∼50%–60%) within the marginal seas in winter.

b. Surface energy flux response

The changes in Arctic sea ice are communicated to the atmosphere via changes in the net surface energy fluxes. Figure 2 shows bimonthly differences between the late twentieth and twenty-first centuries in the surface turbulent energy fluxes (sensible plus latent) and the longwave and shortwave radiative fluxes. The turbulent energy flux response is considerably greater than, and exhibits a different seasonal dependence from, the radiative flux response. The former is largest (maximum amplitudes ∼60–90 W m−2) in winter whereas the latter is largest (maximum amplitudes ∼10–20 W m−2) in summer (shortwave) and late fall (longwave). The winter maximum of the turbulent energy flux response is due to the fact that this is the time of year when the air temperatures are coolest relative to the underlying surface (ice or open water).

The radiative flux responses (upward for longwave and downward for shortwave) are localized to the SIC changes, whereas the turbulent flux response exhibits a dipole structure with upward anomalies over the SIC losses and downward anomalies directly to the south (Fig. 2). The downward turbulent heat flux response directly to the south of the SIC changes may be understood as a result of the retreat of the ice edge and the accompanying region of maximum turbulent energy flux loss from the ocean to the atmosphere (Deser et al. 2000; Magnusdottir et al. 2004; Alexander et al. 2004). Although the magnitude of the downward turbulent heat flux response directly to the south of the SIC changes is overestimated in CAM3 because of the lack of an interactive ocean, the dipole-like structure remains evident even in the fully coupled CCSM3 model (not shown) and in observations (Deser et al. 2000).

A more detailed view of the seasonal cycle of the SIC changes and the surface energy flux response over the Arctic Ocean is given in Fig. 3. For each month, the data were area averaged over all grid points for which SIC during 1980–99 exceeds 50%. This criterion encompasses the full region of SIC losses while excluding most of the negative turbulent energy flux response along the equatorward margins of the ice edge. Consistent with Fig. 2, the largest SIC loss occurs in July–November (peaking in October), while the greatest net surface energy flux response occurs in October–February (peaking in November). The delay in the surface energy flux response relative to the maximum SIC loss has important implications for the timing of the atmospheric circulation and climate response.

Indeed, the seasonal cycles of the surface air temperature and precipitation responses over the Arctic Ocean and over the high-latitude continents (65°–80°N) are in phase with the seasonal cycle of the net surface energy flux response, not SIC (Figs. 4a,b). (Note that in Figs. 4a,b the shortwave radiative component is omitted from the net surface energy flux response in view of the fact that the enhanced downward shortwave radiation resulting from the reduced albedo associated with an ice-free Arctic Ocean does not warm the atmosphere directly but instead warms the ocean, an effect that has been suppressed in our experiments.) Over the Arctic Ocean, the maximum responses occur in November, with values of 17°C, 0.4 mm day−1, and 67 W m−2 for temperature, precipitation, and net surface energy flux, respectively (Fig. 4a). Over land, the maximum temperature and precipitation responses occur in November–December, with values of 7°C and 0.16 mm day−1, respectively (Fig. 4b). The close agreement among the seasonal cycles of the temperature, precipitation, and net surface energy flux responses indicate that the Arctic Ocean net surface energy flux response to sea ice loss exerts a strong local control on climate within the Arctic Ocean (Fig. 4a) and even an important remote control on climate over the adjacent continents (Fig. 4b).

c. Atmospheric temperature response

The bimonthly net surface energy flux (turbulent plus longwave radiation) response and the terrestrial air temperature, snow depth, and precipitation responses are shown in Fig. 5. Air temperature responses that exceed ∼0.5°C (first level of shading) and snow depth responses that exceed ∼0.75 cm liquid water equivalent in absolute value (second level of shading) are statistically significant at the 5% confidence level based on a two-sided Student’s t test. Precipitation responses significant at the 5% confidence level are outlined with thick black contours in the figure.

Consistent with Fig. 4b, the seasonal cycle of the terrestrial air temperature response follows that of the net surface energy flux response, with maximum warming in winter (November–December and January–February) and weaker warming in autumn (September–October) and spring (March–April). The terrestrial warming is largest in coastal regions adjacent to the Arctic Ocean, with the maximum temperature response over Siberia and northern Canada and Alaska, and penetrates approximately 1500 km inland.

The terrestrial surface air temperature responses in early (November–December) and mid-(January–February) winter are largely confined to regions with a mean boundary layer temperature inversion in the late twentieth century (marked by thick black contours on the bimonthly air temperature responses in Fig. 5; note that there is no inversion in the warm season May–June through September–October). Indeed, the vertical structures of the December atmospheric temperature responses over the Arctic Ocean and high-latitude (65°–80°N) continents are confined to below ∼800 hPa, with the warming amplifying toward the surface (maximum values of 6.5°C over land and 16°C over the ocean; Fig. 6). As a consequence of the vertical structure of the warming, the static stability of the boundary layer decreases from the late twentieth century to the late twenty-first century. Over the ocean, the 10°C inversion between 1000 and 900 hPa in the late twentieth century is completely eroded in the late twenty-first century. Over land, the capping inversion, while not completely gone in the twenty-first century, is only approximately 50% of that in the twentieth century (Fig. 6).

The geographical distributions of the strength of the December low-level inversion in the late twentieth and twenty-first centuries are shown in Fig. 7. The marine inversion, which exceeds 12°C over the central Arctic Ocean in the late twentieth century, disappears entirely in the late twenty-first century. The terrestrial inversion weakens and retreats from far eastern Siberia and the north slope of Alaska in the late twenty-first century compared to the late twentieth century. The inversion strength over the Canadian Archipelago decreases from ∼14°C in the late twentieth century to ∼6°C in the late twenty-first century.

The seasonal cycles of the strength of the boundary layer temperature inversion over the Arctic Ocean and high-latitude continents during the late twentieth and twenty-first centuries are shown in Fig. 8. As before, the Arctic Ocean region is defined on a monthly basis as the area with SIC >50% during 1980–99, and the terrestrial region is defined as 65°–80°N, 60°–300°E. In the late twentieth century, the Arctic marine (terrestrial) inversion exists from October through April (November through March), with maximum values ∼9°C in January. The wintertime Arctic marine inversion almost disappears entirely in the late twenty-first century, with only a weak vestige left in February–March (1°–2°C). The wintertime terrestrial inversion also diminishes in strength (maximum values ∼5°C) in the late twenty-first century, but not as dramatically as the marine inversion, and its onset is delayed to December. The largest changes in inversion strength between the twentieth and twenty-first centuries occur in November–December, with maximum values of ∼5°C over land and ∼15°C over the Arctic Ocean (Fig. 8, lower panel).

d. Terrestrial snow cover and precipitation responses

Despite the warming of the atmospheric boundary layer, winter snow depth increases over Siberia, northern Canada, and the northern slope of Alaska (Fig. 5). At the end of winter (March–April), snow depth has increased by ∼1.5–3 cm liquid water equivalent over Siberia and ∼1.0–1.5 cm liquid water equivalent over northern Canada and Alaska. Smaller decreases in snow depth occur to the south, primarily over western Russia and the Canadian Rockies. The patterns of snow depth change persist throughout the winter season, emerging in November–December and reaching their maximum amplitude in March–April.

Given the widespread terrestrial warming during the winter season, the increases in snow depth must be a result of enhanced precipitation. Indeed, precipitation increases in early winter (November–December) over Siberia, eastern Alaska, and northern Canada, and through midwinter (January–February) over Siberia (Fig. 5). Several factors may contribute to the precipitation increases, including enhanced water vapor content in the terrestrial boundary layer (due to increased moisture transport out of the Arctic by the submonthly transients; not shown), destabilization of the terrestrial boundary layer due to surface-intensified warming (recall Figs. 6 –8), and enhanced low-level convergence associated with decreases in sea level pressure (discussed in section 3f).

The role of precipitation in the snow depth response is further assessed by comparing the accumulated precipitation during the cold season (October–March) with March snow depth (Fig. 9). The spatial patterns of the two fields are similar, with areas of positive (negative) accumulated precipitation changes generally corresponding to regions of positive (negative) snow depth changes. There is also quantitative agreement between the magnitudes of the positive accumulated precipitation and snow depth responses. Note that the region of positive accumulated precipitation values over western Europe (excluding Scandinavia) does not correspond to increased snow depth because the air temperature is above 0°C (thick gray curve in Fig. 9, right).

More detail on the seasonal timing of the precipitation, accumulated precipitation, and snow depth responses over Siberia and northern Canada is provided in Fig. 10. The regions are defined using those grid boxes for which the snow depth response in March exceeds 2.5 cm liquid water equivalent. Over northern Canada (Fig. 10a), positive precipitation anomalies occur from October through February, with the largest increases in November–December. The resulting accumulated precipitation response tracks the snow depth response, with a gradual increase throughout the winter to maximum values in February. Similar results are found for Siberia, although there the maximum precipitation response increases occur in December–January and the maximum accumulated precipitation and snow depth responses occur in March (Fig. 10b).

e. Heat budget

What processes account for the winter (October–April) air temperature increases over the Arctic Ocean and adjacent continents? We evaluated the terms in the thermodynamic energy equation (e.g., Holton 2004) using daily model output. The resulting area-averaged heat budget response for the Arctic Ocean (defined as the region with SIC >50% during 1980–99) is shown in Fig. 11 (left panel). The near-surface warming (1000–950 hPa) is primarily due to vertical diffusion (acting to transmit the upward sensible and latent surface heat flux anomalies; green curve) while the warming in the upper part of the boundary layer (900–850 hPa) is attributable to condensational heating (dashed purple curve; Fig. 11, left). Horizontal temperature advection by both the monthly mean and submonthly transient circulation (dashed black and solid black curves) and longwave radiation (solid purple curve) act to cool the boundary layer (Fig. 11, left). Contributions to the heat budget by shortwave radiation and monthly mean temperature tendency are negligible (not shown). The residual term (e.g., the sum of all the terms in the heat budget, including shortwave radiation and temperature tendency) is small compared to the dominant terms in the balance (orange curve in Fig. 11, left).

A different balance of terms obtains for the adjacent continents (Fig. 11, right). The high-latitude (poleward of 65°N) terrestrial boundary layer is warmed by means of horizontal temperature advection by submonthly transient atmospheric motions (primarily their meridional component), condensational heating (except at 1000 hPa), and subgrid-scale horizontal and vertical diffusion. Of these, submonthly transient advection is the dominant mechanism for warming the lower portion of the boundary layer (below 900 hPa; maximum value ∼0.80°C day−1) while condensational heating makes a substantial contribution in the upper portion of the boundary layer (but cools the surface). The other terms in the heat budget act to cool the boundary layer: primarily horizontal temperature advection by the monthly mean atmospheric circulation and longwave radiation (shortwave radiation and the monthly mean temperature tendency terms are near zero; not shown). There is also a substantial residual in the heat budget that acts to cool the air at 950 and 1000 hPa; the origin of this residual is not known.

In summary, heat released from the Arctic Ocean under reduced sea ice conditions (primarily via turbulent sensible and latent heat fluxes) is communicated to the Arctic atmospheric boundary layer by vertical diffusion and to a lesser extent condensational heating. The resulting temperature increase is mixed out over the adjacent continents by submonthly transient atmospheric motions, causing a warming of the terrestrial atmospheric boundary layer. A similar picture holds for the moisture budget (not shown).

f. Atmospheric circulation response

Unlike the thermodynamic responses discussed above, the dynamical atmospheric response varies greatly from month to month (not shown). The bimonthly atmospheric circulation responses, depicted by geopotential height responses at 1000 and 500 hPa, are shown in Fig. 12. The circulation responses are weak (generally <10 m and not statistically significant) during the warm season (June–September), in accord with the small response of the net surface energy fluxes. Although the circulation responses are larger and statistically significant during the cold season (October–May), they exhibit considerable variation in pattern and amplitude. The response in November–December (and in each month individually; not shown) exhibits a baroclinic vertical structure over the Arctic consisting of negative values (−20 to −30 m) at 1000 hPa and positive (10–20 m) values at 500 hPa, and an equivalent barotropic (e.g., amplifying with height) ridge over central and eastern Russia and trough over the Bering Sea. Similar features are found in March–April with weaker amplitudes. A different circulation response is seen in midwinter (January–February), which resembles the negative polarity of the NAO (although this occurs mainly in February; not shown). In this season, the Arctic is dominated by an upper-level ridge response (maximum amplitude ∼50 m at 500 hPa) and negligible response at the surface accompanied by equivalent barotropic troughs over the Atlantic and northeast Pacific.

More detail on the vertical structure of the circulation responses is given in Fig. 13, which shows transects of the temperature and geopotential height changes along 90°E in early (November–December) and mid-(January–February) winter. In early winter, a shallow baroclinic geopotential height response with a nodal point near 925 hPa develops over the Arctic in association with the ice-induced near-surface warming. Farther south, the response consists of an equivalent barotropic ridge with maximum values ∼40 m at 250 hPa near 65°N. The Arctic baroclinic response is also evident in midwinter, but it competes with the equivalent barotropic ridge aloft that weakens the surface trough compared to that in early winter.

The shallow baroclinic atmospheric circulation response over the Arctic in early (and late) winter may be understood as a linear dynamical response to enhanced boundary layer heating induced by the underlying loss of sea ice (Hoskins and Karoly 1981). On the other hand, the equivalent barotropic component of the circulation response in midwinter (e.g., the NAO) and the ridge response over Eurasia in early and late winter represent a nonlinear dynamical response to enhanced boundary layer heating in which transient eddy momentum flux feedbacks associated with perturbations in the storm track play a dominant role (Lau and Holopainen 1984; Peng et al. 1997; Deser et al. 2007; among others). We conjecture that the lack of a surface circulation response over the Arctic in midwinter is due to the near cancellation between the competing effects of the linear and nonlinear dynamical components of the response. A quantitative analysis of the momentum balances of the circulation responses in CAM3 is beyond the scope of this paper.

Internal modes of atmospheric circulation variability have been shown to play a role in shaping the structure of the atmospheric response to different types of external forcing, for example SST changes, sea ice anomalies, or orbital variations (Peng et al. 1997; Deser et al. 2004; Hall et al. 2001; among others). In the case of our CAM3 experiments, however, there is little correspondence between the dominant patterns of internal circulation variability and the patterns of geopotential height response to Arctic sea ice loss, with the notable exception of the month of February (not shown).

g. Comparison with CCSM3

How much of the projected Northern Hemisphere climate response in the fully coupled climate model CCSM3 at the end of the twenty-first century is due to Arctic sea ice loss? To address this question, we compare the climate response to Arctic sea ice loss in the CAM3 experiments with the climate response to GHG forcing under the SRES A1B scenario in CCSM3. Figure 14a (top row) shows the CCSM3 bimonthly terrestrial air temperature response, formed by subtracting the period 1980–99 taken from the twentieth-century ensemble mean from the period 2080–99 taken from the twenty-first-century ensemble mean. The CCSM3 terrestrial air temperature response exhibits the expected poleward amplification in winter, with maximum values exceeding 10°C over Siberia and northern Canada in November–December. Weaker and more spatially uniform warming occurs in summer, with values ∼2°–5°C. Removing the sea ice–induced component from the CCSM3 response (e.g., subtracting the CAM3 response from the full CCSM3 response) eliminates most of the poleward amplification, resulting in a more spatially homogeneous pattern of winter warming over the continents. The main structure in the pattern of residual winter air temperature response is zonal rather than meridional, with ∼2°–4°C (1°C) less warming on the western side of Eurasia (North America) relative to the eastern side. This effect, which is most pronounced in early winter (November–December), may be due to advection of maritime air over Europe and western North America.

In addition to reducing the spatial inhomogeneity, removing the effect of Arctic sea ice loss also reduces the amplitude of the seasonal cycle of the terrestrial air temperature response in CCSM3. This aspect is documented further in Fig. 14b, which shows the seasonal cycles of the CCSM3, CAM3, and sea ice residual CCSM3 terrestrial air temperature responses averaged over the region 60°–300°E poleward of 65°N. The seasonal marches of the temperature responses are remarkably similar in CCSM3 and CAM3, with maximum warming in November–December (10°C in CCSM3 and 7°C in CAM3) and minimum warming in May–August (3°–4°C in CCSM3 and 0°–0.5°C in CAM3). The similarity of the amplitude and timing of the seasonal cycles in CCSM3 and CAM3 results in a relatively constant offset (2°–4°C) between the two. That is, the sea ice–residual component of the high-latitude terrestrial warming in CCSM3 is much less seasonally dependent than the full response, with maximum values ∼3°–4°C during July–February and minimum values ∼2°C in April–May.

Finally, the vertical structures of the high-latitude terrestrial air temperature responses in December for CCSM3 and CAM3 are compared in Fig. 14c. The CCSM3 profile is very similar to CAM3 except for a nearly constant offset of approximately 3°C. Thus, most of the enhanced warming within the boundary layer (∼6°C temperature increases at 1000 hPa relative to that at 800 hPa) results directly from Arctic sea ice loss. The height-invariant sea ice residual warming in CCSM3 is presumably due mainly to enhanced longwave radiation from increased GHG and water vapor concentrations; advection from lower latitudes may also play a role.

Figure 15 shows the accumulated cold season (October–March) precipitation and March snow depth responses in CCSM3. These may be compared with the CAM3 responses to Arctic sea ice loss shown in Fig. 9 (note the doubled range of the color bar scale in Fig. 15 compared to Fig. 9). The patterns of snow depth response in CCSM3 and CAM3 are similar, but the amplitudes are considerably greater in the coupled model compared to the atmospheric model. In particular, the negative (positive) snow depth response values are 4–5 (1.5–2) times larger in CCSM3 than in CAM3. Warmer air temperatures in CCSM3 relative to CAM3 account for the larger snow reductions, while higher accumulated winter precipitation amounts (associated with the warmer air temperatures) account for the greater snow increases. Both CCSM3 and CAM3 exhibit positive snow depth responses over eastern Russia and northern Canada where the accumulated precipitation responses are positive and surface air temperatures during 2080–99 are <−10°C (thick black contour on snow cover panel in Fig. 15). However, the conditions leading to the negative snow depth responses over western Eurasia and North America differ between the two models: in the case of CCSM3 warmer air temperatures control the decreases in snow depth despite the enhanced winter precipitation, whereas in CAM3 warmer temperatures and decreased precipitation lead to diminished snow cover.

The monthly SLP responses for CCSM3 and CAM3 are shown in Fig. 16. Perhaps the most striking aspect of this comparison is that CCSM3 exhibits a substantial SLP response in summer (May/June–September/October) that is almost entirely absent from CAM3. This summer response is quasi-annular in its spatial pattern, with a trough over the Arctic and a ridge over the North Pacific and Atlantic, and exhibits an equivalent barotropic vertical structure (not shown). This comparison indicates that Arctic sea ice loss does not directly drive the high-latitude Northern Hemisphere circulation response in the fully coupled model in summer. There is some similarity between the CAM3 and CCSM3 SLP responses in November–December over the Arctic and North Atlantic as well as in January and May (not shown).

4. Summary

We have documented the response of CAM3/CLM3, an atmospheric general circulation model coupled to a land surface model, to projected Arctic sea ice loss at the end of the twenty-first century. The response was obtained from two 60-yr integrations of the model: one with a repeating seasonal cycle of specified sea ice concentration and thickness conditions for the late twentieth century (1980–99) and one with sea ice conditions for the late twenty-first century (2080–99). In both integrations, the same repeating seasonal cycle of SSTs for the late twentieth century (1980–99) was specified in order to isolate the direct impact of projected future sea ice loss. Note that greenhouse gas concentrations remained fixed at 1980–99 levels in both sets of experiments. The twentieth- and twenty-first-century sea ice (and SST) conditions were obtained from ensemble mean integrations of the fully coupled climate model CCSM3 under historical forcing and SRES A1B scenario forcing, respectively.

An important finding of this study is the delayed response of the net surface energy budget over the Arctic Ocean to sea ice loss. Specifically, the loss of Arctic sea ice is greatest in summer and autumn (July–November, peaking in October) yet the response of the net surface energy flux is largest in winter (October–February, peaking in November). This is because the turbulent (sensible and latent) energy loss, the dominant term in the net surface energy budget, is greatest when the air temperatures are coolest relative to the underlying surface (ice or open water). The delay in the surface energy flux response relative to the maximum sea ice loss has important implications for the timing of the atmospheric response since the energy fluxes communicate the sea ice change to the atmosphere. Indeed, the seasonal cycle of the climate response to future Arctic sea ice loss was shown to follow that of the net surface energy flux rather than that of the sea ice (e.g., the response is greatest in October–February).

Another important finding is the impact of future Arctic sea ice loss on high-latitude terrestrial air temperatures, precipitation, and snow cover. The air temperature and precipitation responses are greatest in November–December over Siberia and northern Canada, with values ∼7°C and ∼0.16 mm day−1, respectively. As a result of enhanced winter precipitation (and despite the warmer air temperatures), snow depths over Siberia and northern Canada increase by ∼1 cm liquid water equivalent in late winter (February–April). The climatological air temperature inversion over the high-latitude continents in winter plays an important role in determining the geographical distribution and vertical structure of the air temperature response to Arctic sea ice loss. Specifically, the spatial extent of the winter air temperature response is confined to the region with a present-day inversion, and the vertical extent of the temperature response is confined to the boundary layer (e.g., below ∼850 hPa and amplifying toward the surface). As a consequence of the vertical structure of the terrestrial air temperature response, the static stability of the boundary layer decreases by ∼50% from the late twentieth century to the late twenty-first century.

The dominant process warming the atmospheric boundary layer over the high-latitude continents is horizontal heat advection by submonthly transient atmospheric motions. In other words, high-frequency wind fluctuations mix the air warmed over the Arctic Ocean because of enhanced sensible heat loss associated with reduced sea ice cover out over the high-latitude continents. Temperature advection by the monthly mean atmospheric circulation acts as a negative feedback, as do all of the other terms in the heat budget except for latent heat release during condensation (e.g., enhanced cloud formation). Similar results are found for the high-latitude terrestrial moisture budget.

The seasonal response of the atmospheric circulation to Arctic sea ice loss is also approximately in phase with the timing of the net surface energy flux response. In particular, significant circulation responses are found only during the cold season (October–April) with insignificant responses during summer. Within the cold season, the spatial and vertical structures of the response differ from month to month. In early winter (November–December), the response is baroclinic over the Arctic, with low pressure anomalies near the surface (maximum values ∼3–5 hPa) and high pressure anomalies aloft. The baroclinic response over the Arctic is also evident in midwinter (January–March), but it competes with an equivalent barotropic ridge aloft, resulting in near-zero surface pressure anomalies. In February, the response resembles the negative phase of the North Atlantic Oscillation, the dominant internal mode of winter circulation variability. In April, the surface trough over the Arctic is accompanied by low pressure anomalies aloft.

Comparison of the CAM3 results with the coupled model response to GHG forcing reveals that Arctic sea ice loss accounts for most of the seasonal, spatial, and vertical structure of the late-twenty-first-century high-latitude terrestrial air temperature change in CCSM3. The terrestrial warming associated with Arctic sea ice loss has implications for Arctic ecosystems and permafrost degradation (Lawrence et al. 2008). Arctic sea ice loss also accounts for much of CCSM3’s high-latitude SLP response during November–January but plays a negligible role in the atmospheric circulation response in other months. Finally, Arctic sea ice loss is not the dominant factor in the coupled model’s terrestrial snow cover and winter precipitation responses to GHG increases.

5. Discussion

A common criticism of AGCM experiments is that the specified boundary conditions are themselves a response to atmospheric conditions rather than a cause (e.g., Bretherton and Battisti 2000), rendering the experimental design inappropriate. This is not the case, however, for future Arctic sea ice loss. Indeed, the net surface turbulent energy flux response over the Arctic Ocean, the dominant mechanism whereby the ice loss is thermally communicated to the atmosphere, is similar between the fully coupled CCSM3 and the CAM3 experiments (Fig. 17). In both models, the maximum upward surface turbulent energy flux response occurs during November–January, with slightly weaker amplitude in CCSM3 (∼40 W m−2) than CAM3 (∼50 W m−2) due to the additional atmospheric warming from the direct radiative effect of increased GHG concentrations in the coupled model simulation (not shown). The spatial pattern of the net surface turbulent energy flux response is also very similar in CCSM3 and CAM3 (not shown). As expected, the surface longwave radiative flux response is opposite in sign between CCSM3 and CAM3 because of the lack of GHG concentration changes in the CAM3 experiments. The downward surface longwave radiative flux response in CCSM3, associated with the direct radiative effect of increased GHGs, is approximately constant over the course of the year at ∼15 W m−2 (Fig. 17).

Our experiments address only the direct impact of Arctic sea ice loss on the atmospheric circulation and climate and neglect the potential role of oceanic feedbacks. In particular, warming of the Arctic Ocean due to enhanced solar heating associated with sea ice loss may provide additional forcing to the overlying atmosphere, although Singarayer et al. (2006) has shown this effect to be small. In addition, warming of the high-latitude North Pacific and Atlantic Oceans due to enhanced downward turbulent energy fluxes as a result of anomalous warm air advection out of the Arctic may also alter the atmospheric circulation response through feedbacks with the midlatitude storm tracks (e.g., Peng et al. 1997; Deser et al. 2007). A follow-up study with an interactive ocean model will be conducted to address the role of oceanic feedbacks.

Magnusdottir et al. (2004) found a consistent negative NAO–NAM-like circulation response (e.g., an equivalent barotropic vertical structure with positive SLP anomalies over the Arctic and negative SLP anomalies in midlatitudes) to an idealized pattern of Arctic sea ice loss in the Atlantic sector in each of the winter months examined (December–March), with the strongest response in March. Their findings are generally consistent with those reported in Seierstad and Bader (2009) for an Arcticwide pattern of future Arctic sea ice loss. In this study, we find different circulation responses in early (November–December) and mid- (January–March) winter. The early winter responses are not NAO-like, and exhibit negative SLP values over the Arctic, the Bering Sea, and northeastern Canada, with a baroclinic structure in the vertical. This early winter response closely resembles that found by Singarayer et al. (2006) for the December–February average. The midwinter response exhibits negligible SLP change over the Arctic, with a strong negative NAO pattern in February only. Further research is needed to understand the reasons for the disparity in the winter circulation responses among different studies.

The lack of a pronounced summertime circulation response found in this study, although consistent with the small net surface energy flux response to the imposed sea ice loss, is at odds with the results of Bhatt et al. (2008). In that study, a statistically significant equivalent barotropic high pressure anomaly over the North Pacific (maximum SLP amplitude ∼2 hPa) was found in response to specified sea ice extent anomalies taken from observations in August 1995. Such a response is opposite to the 1-hPa low pressure anomaly obtained in this study. The sea ice extent anomaly imposed in the study of Bhatt et al. (2008) is confined to the coastal Arctic seas (Laptev, Kara, East Siberian, and Beaufort), and thus differs from the Arcticwide decrease in summer sea ice concentration used in our experiments. Also, sea ice thickness was kept constant at 2 m in Bhatt et al. (2008), which is different from the treatment of sea ice thickness in our study. In summary, it appears prudent to acknowledge that large uncertainties remain in quantifying the dynamical atmospheric response to future Arctic sea ice loss.

The low-level temperature inversion was shown to play a role in the surface thermal response over the Arctic Ocean and adjacent continental regions. Given that the strength of the present-day (late twentieth century) wintertime inversion is considerably overestimated in CAM3 compared to both the National Centers for Environmental Prediction (NCEP)–NCAR and European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis products (not shown but see Boé et al. 2009), this may inflate the magnitude of the wintertime surface temperature response to Arctic sea ice loss. However, the geographical distribution of the present-day inversion is well simulated in CAM3 compared to the reanalysis products (not shown), implying that the spatial extent of the surface thermal response to future Arctic sea ice loss may be realistic.

With the caveats noted above, the results shown here may serve as a guide to the direct impact of projected future Arctic sea ice loss upon climate and atmospheric circulation. Indeed, the emerging signal of enhanced autumn warming over the Arctic in the past ∼5 yr (Serreze et al. 2008) exhibits many similarities with the simulated response to Arctic sea ice loss documented in this study. It remains to be seen whether other aspects of the simulated response become detectable in the near future as Arctic sea ice continues to decline.

Acknowledgments

We thank the two anonymous reviewers for helpful comments and suggestions. We gratefully acknowledge support from the National Science Foundation’s Arctic System Science Program.

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 , 890905.

    • Search Google Scholar
    • Export Citation
  • Bhatt, U. S., M. A. Alexander, C. Deser, J. E. Walsh, J. S. Miller, M. Timlin, J. D. Scott, and R. Tomas, 2008: The atmospheric response to realistic reduced summer Arctic sea ice anomalies. Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications, Geophys. Monogr., Vol. 180, Amer. Geophys. Union, 91–110.

    • Search Google Scholar
    • Export Citation
  • Boé, J., A. Hall, and X. Qu, 2009: Current GCMs’ unrealistic negative feedback in the Arctic. J. Climate, 22 , 46824695.

  • Boville, B. A., P. J. Rasch, J. J. Hack, and J. R. McCaa, 2006: Representation of clouds and precipitation processes in the Community Atmosphere Model Version 3 (CAM3). J. Climate, 19 , 21842198.

    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., and D. S. Battisti, 2000: An interpretation of the results from atmospheric general circulation models forced by the time history of the observed sea surface temperature distribution. Geophys. Res. Lett., 27 , 767770.

    • Search Google Scholar
    • Export Citation
  • Collins, W. D., and Coauthors, 2006a: The Community Climate System Model Version 3 (CCSM3). J. Climate, 19 , 21222143.

  • Collins, W. D., and Coauthors, 2006b: The formulation and atmospheric simulation of the Community Atmosphere Model Version 3 (CAM3). J. Climate, 19 , 21442161.

    • Search Google Scholar
    • Export Citation
  • Comiso, J. C., C. L. Parkinson, R. Gersten, and L. Stock, 2008: Accelerated decline in the Arctic sea ice cover. Geophys. Res. Lett., 35 , L01703. doi:10.1029/2007GL031972.

    • Search Google Scholar
    • Export Citation
  • Deser, C., and H. Teng, 2008: Recent trends in Arctic sea ice and the evolving role of atmospheric circulation forcing, 1979–2007. Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications, Geophys. Monogr., Vol. 180, Amer. Geophys. Union, 7–26.

    • Search Google Scholar
    • Export Citation
  • 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 , 617633.

    • Search Google Scholar
    • Export Citation
  • 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 , 877889.

    • Search Google Scholar
    • Export Citation
  • 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 , 47514767.

    • Search Google Scholar
    • Export Citation
  • Gerdes, R., 2006: Atmospheric response to changes in Arctic sea ice thickness. Geophys. Res. Lett., 33 , L18709. doi:10.1029/2006GL027146.

    • Search Google Scholar
    • Export Citation
  • Hack, J. J., J. M. Caron, G. Danabasoglu, K. W. Oleson, C. Bitz, and J. Truesdale, 2006: CCSM–CAM3 climate simulation sensitivity to changes in horizontal resolution. J. Climate, 19 , 22672289.

    • Search Google Scholar
    • Export Citation
  • Hall, N. M. J., J. Derome, and H. Lin, 2001: The extratropical signal generated by a midlatitude SST anomaly. Part I: Sensitivity at equilibrium. J. Climate, 14 , 20352053.

    • Search Google Scholar
    • Export Citation
  • Holland, M. M., C. M. Bitz, and B. Tremblay, 2006: Future abrupt reductions in the summer Arctic sea ice. Geophys. Res. Lett., 33 , L23503. doi:10.1029/2006GL028024.

    • Search Google Scholar
    • Export Citation
  • Holton, J. R., 2004: An Introduction to Dynamic Meteorology. Academic Press, 535 pp.

  • 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 , 33473358.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., and D. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38 , 11791196.

    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation: Regional temperatures and precipitation. Science, 269 , 676679.

    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., J. J. Hack, A. S. Phillips, J. Caron, and J. Yin, 2006: The Dynamical Simulation of the Community Atmosphere Model Version 3 (CAM3). J. Climate, 19 , 21622183.

    • Search Google Scholar
    • Export Citation
  • Kvamstø, N. G., P. Skeie, and D. B. Stephenson, 2004: Impact of Labrador sea-ice extent on the North Atlantic Oscillation. Int. J. Climatol., 24 , 603612.

    • Search Google Scholar
    • Export Citation
  • Lau, N-C., and E. O. Holopainen, 1984: Transient eddy forcing of the time-mean flow as identified by geopotential tendencies. J. Atmos. Sci., 41 , 313328.

    • Search Google Scholar
    • Export Citation
  • Lawrence, D. M., A. G. Slater, R. A. Tomas, M. M. Holland, and C. Deser, 2008: Accelerated Arctic land warming and permafrost degradation during rapid sea ice loss. Geophys. Res. Lett., 35 , L11506. doi:10.1029/2008GL033985.

    • Search Google Scholar
    • Export Citation
  • 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 , 857876.

    • Search Google Scholar
    • Export Citation
  • Meehl, G. A., and Coauthors, 2006: Climate change projections for the twenty-first century and climate change commitment in the CCSM3. J. Climate, 19 , 25972616.

    • Search Google Scholar
    • Export Citation
  • Peng, S. L., W. A. Robinson, and M. P. Hoerling, 1997: The modeled atmospheric response to midlatitude SST anomalies and its dependence on background circulation states. J. Climate, 10 , 971987.

    • Search Google Scholar
    • Export Citation
  • Post, E., and M. C. Forchhammer, 2008: Climate change reduces reproductive success of an Arctic herbivore through trophic mismatch. Philos. Trans. Roy. Soc. London, B363 , 23672373.

    • Search Google Scholar
    • Export Citation
  • Seierstad, I. A., and J. Bader, 2009: Impact of a projected future Arctic sea ice reduction on extratropical storminess and the NAO. Climate Dyn., 33 , 937943. doi:10.1007/s00382-008-0463-x.

    • Search Google Scholar
    • Export Citation
  • Serreze, M. C., M. M. Holland, and J. Stroeve, 2007: Perspectives on the Arctic’s shrinking sea-ice cover. Science, 315 , 15331536.

  • Serreze, M. C., A. P. Barrett, J. C. Stroeve, D. N. Kindig, and M. M. Holland, 2008: The emergence of surface-based Arctic amplification. Cryosphere Discuss., 2 , 601622.

    • Search Google Scholar
    • Export Citation
  • Sewall, J. O., 2005: Precipitation shifts over western North America as a result of declning Arctic sea ice cover, the coupled system response. Earth Interactions, 9 .[Available online at http://EarthInteractions.org].

    • Search Google Scholar
    • Export Citation
  • Singarayer, J. S., J. L. Bamber, and P. J. Valdes, 2006: Twenty-first-century climate impacts from a declining Arctic sea ice cover. J. Climate, 19 , 11091125.

    • Search Google Scholar
    • Export Citation
  • Solomon, S., D. Qin, M. Manning, M. Marquis, K. Averyt, M. M. B. Tignor, H. L. Miller Jr., and Z. Chen, Eds.,. 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

    • Search Google Scholar
    • Export Citation
  • Stroeve, J., M. M. Holland, W. Meier, T. Scambos, and M. C. Serreze, 2007: Arctic sea ice decline: Faster than forecast. Geophys. Res. Lett., 34 , L09501. doi:10.1029/2007GL029703.

    • Search Google Scholar
    • Export Citation
  • Vavrus, S., and D. Waliser, 2008: An improved parameterization for simulating Arctic cloud amount in the CCSM3 climate model. J. Climate, 21 , 56735687.

    • Search Google Scholar
    • Export Citation
  • Wallace, J. M., 2000: North Atlantic Oscillation/annular mode: Two paradigms–one phenomenon. Quart. J. Roy. Meteor. Soc., 126 , 791805.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Bimonthly distributions of Arctic (a) sea ice concentration (%) and (b) sea ice thickness (m) during 1980–99 and 2080–99 from CCSM3.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 2.
Fig. 2.

Bimonthly Arctic (top row) sea ice thickness (ΔSIT; m) and (second row) concentration (ΔSIC; %) differences (2080–99 minus 1980–99) from CCSM3. Bimonthly (third row) turbulent energy flux (ΔSH + LH), (fourth row) longwave radiative flux (ΔLW), and (bottom row) shortwave radiative flux (ΔSW) responses to sea ice cover changes. Fluxes (W m−2) are positive upward.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 3.
Fig. 3.

Seasonal cycle of the turbulent energy flux (Wm−2; ΔSH + LH; thin solid curve), longwave radiative flux (ΔLW; dotted curve), and shortwave radiative flux (ΔSW; dashed curve) responses area averaged over the Arctic Ocean. The net surface energy flux response is given by the thick solid curve, and the SIC changes (%) are indicated by the gray bars (note the inverted scale). Fluxes are positive upward.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 4.
Fig. 4.

Seasonal cycles of air temperature (°C; dotted curve) and precipitation (mm day−1; dashed curve) responses area averaged over (a) the Arctic Ocean and (b) the high-latitude continents (65°–80°N; 60°–300°E). The solid curve in both (a) and (b) shows the sum of the turbulent and longwave fluxes area averaged over the Arctic Ocean (W m−2). SIC changes are indicated by the gray bars (scale as in Fig. 3, not shown).

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 5.
Fig. 5.

Bimonthly responses of net surface energy flux (ΔQnet; W m−2), terrestrial air temperature (ΔAir T; °C), terrestrial snow depth (ΔSnow; cm liquid water equivalent), and terrestrial precipitation (ΔPrecip; mm day−1). Color scales are given at the bottom of the figure. Thick black contours on the air temperature panels outline regions with a low-level temperature inversion (T850hPa − T1000hPa > 0°C) during 1980–99. Precipitation responses significant at the 5% confidence level are outlined with black contours.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 6.
Fig. 6.

Vertical profiles of atmospheric temperature during 1980–99 (dashed curves) and 2080–99 (solid curves) over (left) the Arctic Ocean and (center) the high-latitude (65°–80°N) continents in December. (right) The 2080–99 minus 1980–99 differences over the Arctic Ocean (thick curve) and high-latitude continents (thin curve).

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 7.
Fig. 7.

Geographical distributions of the strength of the December low-level inversion (T850hPa − T1000hPa > 0°C) during (left) 1980–99 and (right) 2080–99. Values <1.5°C not shown.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 8.
Fig. 8.

Seasonal cycles of T850hPa − T1000hPa during 1980–99 (dark colors) and 2080–99 (light colors) over the (top) Arctic Ocean (blue) and (middle) high-latitude continents (red). (bottom) The 2080–99 minus 1980–99 differences for the Arctic Ocean (blue) and high-latitude continents (red).

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 9.
Fig. 9.

Accumulated cold season (October–March) precipitation and March snow depth responses (cm). The thick black and gray contours on the snow panel denote the −10° and 0°C air temperature isotherms, respectively, during 2080–99.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 10.
Fig. 10.

Seasonal cycles of precipitation (mm day−1; dashed curve), accumulated precipitation (cm; dotted curve), and snow depth (cm liquid water equivalent; solid curve) responses for (a) northern Canada and (b) Siberia.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 11.
Fig. 11.

Heat budget response (°C day−1) for the (left) Arctic Ocean and (right) high-latitude (65°–80°N) continents. Submonthly transient temperature advection (solid black curve); monthly mean temperature advection (dashed black curve); condensational heating (dashed purple curve); vertical and horizontal diffusion (green curve); longwave radiation (solid purple curve). The time-tendency and shortwave radiation terms are negligible (not shown). The sum of all the terms (e.g., the heat budget residual) is given by the orange curve.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 12.
Fig. 12.

Bimonthly geopotential height responses at 1000 and 500 hPa. The contour interval is 10 m, with positive (negative) values in red (blue) and the zero contours omitted. Shading indicates values that exceed the 5% confidence level based on a two-sided Student’s t test.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 13.
Fig. 13.

Vertical structures of the geopotential height (m; contours) and temperature (°C; color shading) responses along 90°E in (left) November–December and (right) January–February.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 14.
Fig. 14.

(a) Bimonthly terrestrial air temperature responses (°C) for (top) CCSM3 and (bottom) CCSM3 minus CAM3. (b) Seasonal cycle of the high-latitude (65°–80°N) terrestrial air temperature response in CCSM3 (dashed), CAM3 (solid), and their difference (dotted). (c) Vertical profiles of the December high-latitude (65°–80°N) terrestrial air temperature response in CCSM3 (dashed), CAM3 (solid), and their difference (dotted).

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 15.
Fig. 15.

Accumulated cold season (October–March) precipitation and March snow depth responses (cm) in CCSM3. The thick black and gray contours on the snow panel denote the −10° and 0°C air temperature isotherms, respectively, during 2080–99.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 16.
Fig. 16.

Bimonthly SLP responses for CAM3 and CCSM3. The contour interval is 1 hPa, with positive (negative) values in red (blue) and the zero contours omitted. Shading indicates values that exceed the 5% confidence level based on a two-sided Student’s t test.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

Fig. 17.
Fig. 17.

As in Fig. 3 but for CCSM3. The net surface energy flux response curve is omitted.

Citation: Journal of Climate 23, 2; 10.1175/2009JCLI3053.1

* The National Center for Atmospheric Research is sponsored by the National Science Foundation.

Save