• Bonan, G., 1998: The land surface climatology of the NCAR Land Surface Model (LSM 1.0) coupled to the NCAR Community Climate Model (CCM3). J. Climate,11, 1307–1326.

  • Boville, B. A., and P. R. Gent, 1998: The NCAR Climate System Model, version one. J. Climate,11, 1115–1130.

  • ——, and J. W. Hurrell, 1998: A comparison of the atmospheric circulations by the CCM3 and CSM1. J. Climate,11, 1327–1341.

  • Briegleb, B., and D. H. Bromwich, 1998: Polar climate simulation of the NCAR CCM3. J. Climate,11, 1270–1286.

  • Bryan, F. O., B. G. Kauffman, W. G. Large, and P. R. Gent, 1996:The NCAR CSM Flux Coupler. National Center for Atmospheric Research NCAR/TN-424+STR, 46 pp.

  • Gates, W. L., 1992: AMIP: The Atmospheric Model Intercomparison Project. Bull. Amer. Meteor. Soc.,72, 1962–1970.

  • Gent, P. R., J. Willebrand, T. J. McDougall, and J. C. McWilliams, 1995: Parameterizing eddy-induced tracer transports in ocean circulation models. J. Phys. Oceanogr.,25, 463–474.

  • ——, F. O. Bryan, G. Danabasoglu, S. C. Doney, W. R. Holland, W. G. Large, and J. C. McWilliams, 1998: The NCAR Climate System Model global ocean component. J. Climate. 11, 1287–1306.

  • Hurrell, J. W., 1995: Comparison of the NCAR Community Climate Model (CCM) climates. Climate Dyn.,11, 25–50.

  • ——, H. van Loon, and D. J. Shea, 1998: The mean state of the troposphere. Meteorology of the Southern Hemisphere, D. Karoly and D. Vincent, Eds., Amer. Meteor. Soc., in press.

  • Kiehl, J. T., J. J. Hack, G. Bonan, B. A. Boville, D. Williamson, and P. Rasch, 1998: The National Center for Atmospheric Research Community Climate Model: CCM3. J. Climate,11, 1131–1149.

  • Large, W. G., J. C. McWilliams, and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parameterization. Rev. Geophys.,32, 363–403.

  • Mitchell, J. F. B., and T. S. Hills, 1986: Sea-ice and the antarctic winter circulation, a numerical experiment. Quart. J. Roy. Meteor. Soc.,112, 953–969.

  • ——, and C. A. Senior, 1989: The antarctic winter; simulations with climatological and reduced sea-ice extents. Quart. J. Roy. Meteor. Soc.,115, 225–246.

  • Quintanar, A. I., and C. R. Mechoso, 1995a: Quasi-stationary waves in the Southern Hemisphere. Part II: Generation Mechanisms. J. Climate,8, 2673–2690.

  • ——, and ——, 1995b: Quasi-stationary waves in the Southern Hemisphere. Part I: Observational data. J. Climate,8, 2659–2672.

  • Schoeberl, M. R., and M. A. Geller, 1976: The structure of stationary planetary waves in winter in relation to the polar night jet intensity. Geophys. Res. Lett.,3, 177–180.

  • Simmonds, I., 1981: The effect of sea-ice on a general circulation model of the Southern Hemisphere. Sea Level, Ice and Climatic Change, I. Allison, Ed., IAHS, 253 pp.

  • ——, and Y. Lin, 1983: Topographical and thermal forcing in a circulation model of the Southern Hemisphere. Publication No. 24, University of Melbourne, Meteorology Department, Melbourne, Victoria, Australia, 78 pp.

  • Trenberth, K. E., 1979: Interannual variability of the 500 mbar zonal mean flow in the Southern Hemisphere. Mon. Wea. Rev.,107, 1515–1524.

  • ——, 1980: Planetary waves at 500 mb in the Southern Hemisphere. Mon. Wea. Rev.,108, 1378–1389.

  • van Loon, H. V., and R. Jenne, 1972: The zonal harmonic standing waves in the Southern Hemisphere. J. Geophys. Res.,77, 992–1003.

  • ——, ——, and K. Labitzke, 1973: Zonal harmonic standing waves. J. Geophys. Res.,78, 4463–4471.

  • ——, J. W. Kidson, and A. B. Mullen, 1993: Decadal variation of the annual cycle in the Australian dataset. J. Climate,6, 1227–1231.

  • Watterson, I. G., and I. N. James, 1992: Baroclinic waves propagating from a high latitude source. Quart. J. Roy. Meteor. Soc.,118, 23–50.

  • Weatherly, J., B. Briegleb, W. G. Large, and T. Bettge, 1998: Sea ice and polar climate in the NCAR CSM. J. Climate,11, 1472–1486.

  • Xu, J.-S., H. von Storch, and H. van Loon, 1990: The performance of four spectral GCMs in the Southern Hemisphere: The January and July climatology and the semi-annual wave. J. Climate,3, 53–70.

  • View in gallery

    Geopotential height distribution at 500 mbar for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 100 m.

  • View in gallery

    Distribution of temperature differences at 500 mbar for (a) NCEP–CCM3, (b) NCEP–CSM1, and (c) CCM3–CSM1. Contour interval is 0.5 K.

  • View in gallery

    Standard deviation of the geopotential heights at 500 mbar for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 m.

  • View in gallery

    Geopotential height differences at 500 mbar for (a) NCEP–CCM3, (b) NCEP–CSM1, and (c) CCM3–CSM1. Contour interval is 20 m.

  • View in gallery

    Distribution of surface air temperature differences for CCM3–CSM1. Contour interval is 3 K.

  • View in gallery

    Geopotential height deviations from the zonal mean at 500 mbar for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 20 m.

  • View in gallery

    Temperature deviations from the zonal mean at 500 mbar for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 0.5 K.

  • View in gallery

    Zonal-mean amplitude of quasi-stationary wave 1 for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 gpm.

  • View in gallery

    Latitude–height section of the mean zonal wind speed for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 5 m s−1.

  • View in gallery

    Latitude–height section of the difference in mean zonal wind speed for (a) NCEP–CCM3 and (b) NCEP–CSM1. Contour interval is 1 m s−1.

  • View in gallery

    Time–longitude variation of quasi-stationary wave 1 at 60°S at 500 mbar: (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 20 gpm.

  • View in gallery

    Amplitude of quasi-stationary wave 1 at 500 mbar: (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 25 gpm.

  • View in gallery

    Zonal-mean amplitude of quasi-stationary wave 2 for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 gpm.

  • View in gallery

    Amplitude of quasi-stationary wave 2 at 500 mbar: (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 gpm.

  • View in gallery

    Zonal-mean amplitude of quasi-stationary wave 3 for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 gpm.

  • View in gallery

    Amplitude of quasi-stationary wave 3 at 500 mbar: (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 5 gpm.

  • View in gallery

    Amplitude of quasi-stationary wave 3 for CCM3 with satellite-observed concentrations. Contour interval is 10 gpm.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 145 138 2
PDF Downloads 51 47 0

Quasi-Stationary Waves in the Southern Hemisphere: An Examination of Their Simulation by the NCAR Climate System Model, with and without an Interactive Ocean

View More View Less
  • 1 Department of Geography, University of California, Los Angeles, Los Angeles, California
© Get Permissions
Full access

Abstract

The three primary quasi-stationary waves in the geopotential height field of the Southern Hemisphere, as simulated by the National Center for Atmospheric Research (NCAR) Climate System Model (CSM1) and the Community Climate Model, version 3 (CCM3), are examined and compared with the NCAR–National Centers for Environmental Prediction reanalyses. Fourier analysis is used to decompose the geopotential heights into its zonal harmonic components. Both models are able to simulate the mean and zonal asymmetry of the geopotential heights; however, the CSM1 simulates the interannual variability considerably better than the CCM3. The amplitude and phase of wave 1 are well simulated by the models, particularly in the subantarctic region. The models are also able to reproduce the interannual variation in phase and amplitude of wave 1. The success of the simulation is attributed to the models’ ability to simulate well the important features of the geopotential height and temperature distributions. The models vary in their ability to simulate waves 2 and 3. Reasons for these variations are discussed.

Corresponding author address: Dr. Marilyn Raphael, Department of Geography, UCLA, Los Angeles, CA 90095-1524.

Email: raphael@geog.ucla.edu

Abstract

The three primary quasi-stationary waves in the geopotential height field of the Southern Hemisphere, as simulated by the National Center for Atmospheric Research (NCAR) Climate System Model (CSM1) and the Community Climate Model, version 3 (CCM3), are examined and compared with the NCAR–National Centers for Environmental Prediction reanalyses. Fourier analysis is used to decompose the geopotential heights into its zonal harmonic components. Both models are able to simulate the mean and zonal asymmetry of the geopotential heights; however, the CSM1 simulates the interannual variability considerably better than the CCM3. The amplitude and phase of wave 1 are well simulated by the models, particularly in the subantarctic region. The models are also able to reproduce the interannual variation in phase and amplitude of wave 1. The success of the simulation is attributed to the models’ ability to simulate well the important features of the geopotential height and temperature distributions. The models vary in their ability to simulate waves 2 and 3. Reasons for these variations are discussed.

Corresponding author address: Dr. Marilyn Raphael, Department of Geography, UCLA, Los Angeles, CA 90095-1524.

Email: raphael@geog.ucla.edu

1. Introduction

General circulation models (GCMs) are commonly used to study the earth’s climate system. With relatively few exceptions, however (e.g., Quintanar and Mechoso 1995a; Simmonds and Lin 1983), these studies have focused upon the Northern Hemisphere (NH) atmospheric circulation and climate. This is partly because GCMs have failed to reproduce well some of the key features of the large-scale Southern Hemisphere (SH) circulation. For example, Xu et al. (1990) examined the ability of five GCMs to simulate the SH circulation climate. They found that key features such as the double jet structure, and the zonal asymmetry in the geopotential height field dominated by wave 1, are poorly simulated. Recently a new GCM was released by the National Center for Atmospheric Research (NCAR). It provides a fresh opportunity to see how much simulation of the SH climate has been improved from previous GCMs. Hurrell (1995) compared the climates simulated by several earlier versions of the NCAR Climate Community Models (CCMs) with the aim of documenting common strengths and deficiencies as well as providing a benchmark against which future CCM versions may be judged. The present study does not attempt to repeat that work. Rather, it focuses upon GCM simulation of the large-scale SH circulation climate, comparing the results of a model that uses specified sea surface temperatures (SSTs) as a lower boundary forcing with the results that are obtained when that same model is coupled to a dynamic, interactive ocean.

This study is an examination of the SH atmospheric circulation as simulated by the NCAR Climate System Model, version 1 (CSM1), and the Climate Community Model, version 3 (CCM3). The motivation for this research is twofold. First, to evaluate how well these models simulate the large-scale atmospheric circulation in the SH and, second, if they prove reliable, as appears to be the case, to use them as mechanistic tools to examine the mechanisms that force and maintain the quasi-stationary waves in the SH. In this paper the first step, the abilities of the CCM3 and the CSM1 to simulate the three primary quasi-stationary waves in the SH circulation, is reported. Model results are compared with the NCAR–National Centers for Environmental Prediction (NCAR–NCEP) reanalyses.

The observed features of quasi-stationary waves 1, 2, and 3 are briefly described in section 2. The models and the data are briefly described in section 3. In section 4 the results are presented and discussed. Section 5 summarizes the significance of the model results and presents some conclusions.

2. The mean quasi-stationary waves

In the climatological mean, the large-scale circulation of the SH is almost completely described (99%) by the first three quasi-stationary waves in the geopotential height field (van Loon and Jenne 1972; Trenberth 1980). In this field, wave 1 is by far the dominant of these three, accounting for some 80%–90% of the variance. Its peak amplitude occurs between 50°S and 60°S. This peak is relatively constant at these latitudes although it experiences strong growth in the austral spring (Quintanar and Mechoso 1995b) and is largest in the southern winter when it propagates into the stratosphere (see Fig. 1.23 in Hurrell et al. 1998). A secondary peak in amplitude of wave 1 occurs in the subtropics, and it too reaches a maximum in the southern winter. Wave 1 is quasi-barotropic in both the subtropical and subantarctic regions. In the latter region it occurs in close association with the zonal anomalies of temperature, prompting the suggestion that thermal forcing is a key component in forcing the pressure wave (van Loon and Jenne 1972). Modeling studies have been few, and they suggest both the influence of Antarctica’s asymmetry (e.g., Watterson and James 1992) as well as lower latitude forcing from the Indian Ocean region (Quintanar and Mechoso 1995a). Observational studies by Trenberth (1980) suggest that subtropical wave 1 is coupled to the Southern Oscillation and to variations in the zonal westerlies.

Compared to wave 1, waves 2 and 3 have considerably smaller amplitudes and account for much less of the variance in the data. Quintanar and Mechoso (1995b) suggest that they contribute so little to the quasi-stationary wave field because they are primarily eastward traveling. Wave 2 reaches its maximum amplitudes in the same latitudes as wave 1, the subantarctic and polar regions. Wave 2 is thought to be forced by Antarctica’s orography since its ridges are aligned with the highest (and coldest) regions over that continent (van Loon and Jenne 1972). Wave 2 has strong interannual variation (Trenberth 1979; van Loon et al. 1993) not always present in the mean and accounts for as much as 10% of the variance. Wave 3 reaches its maximum amplitudes at 50°–55°S, and in the southern winter. These amplitudes also vary considerably from year to year. Van Loon et al. (1993), using a dataset from the Australian Bureau of Meteorology, show that the mean amplitude of wave 3 increased remarkably after 1977. Wave 3 is thought to be associated with the three lower latitude continents to which its ridges lie close (van Loon and Jenne 1972).

3. Description of models and data

The following description of the models assessed in this study is brief. General details and other aspects of the models’ performance are addressed by relevant articles in this issue. The CSM1 is NCAR’s new, comprehensive climate system model whose components are the CCM3, a dynamic, interactive ocean model, a sea-ice model, and a land surface process model. A full description of the CSM1, including initialization and spinup, is given by Boville and Gent (1998) in this issue.

The NCAR CSM ocean model (NCOM) is described in Gent et al. (1998). It has 45 levels in the vertical, 4 of which are in the top 50 m and 25 in the top 1 km. The horizontal resolution of NCOM varies latitudinally; 1.2° at the equator and poles, and 2.3° in the midlatitudes. Its longitudinal resolution is 2.4°. NCOM uses the Gent–McWilliams eddy mixing parameterization (Gent et al. 1995) and the nonlocal K- profile boundary layer parameterization from Large et al. (1994). The sea-ice model (Weatherly et al. 1998) accounts for snow on top of ice, accumulating and ablating snow as the surface energy balance requires. It also computes the temperature of the snow–ice interface. Weatherly et al. (1998) show that the CSM1 simulates ice concentrations and thickness similar to that observed in the SH. The land surface model (Bonan 1998) is a one-dimensional model of energy, momentum, and water and carbon dioxide exchanges between the atmosphere and the land. It accounts for ecological differences among vegetation types, hydraulic and thermal differences among soil types, and allows for multiple surfaces within one grid cell. Interaction among the component models of the CSM1 is modulated by a flux coupler. The flux coupler controls the exchange of information among the models and ensures the conservation of heat, momentum, and freshwater within the climate model system during the exchange processes (Bryan et al. 1996).

The CCM3 is a three-dimensional global atmospheric general circulation model (GCM). It descends directly from the CCM2 and has improved physical representation of specific climate processes. These include substantial improvements to the top of the atmosphere and surface energy budgets, a reduction in the magnitude of the hydrological cycle, and a more realistic distribution of tropical precipitation. In this study, the CCM3 and the CSM1 are similar in most aspects, the chief difference being the part in thermal forcing played by the interactive ocean in the CSM1. Both models have 18 levels in the vertical and realistic orography. Their resolution is a triangular truncation at wavenumber 42 (T42) on a Gaussian grid. The surface air temperature used by the CSM1 atmosphere is obtained from the interaction among the model atmosphere, land surface model, sea-ice model, and ocean model. The CCM3 is run without an ocean. Its SST is specified using monthly averaged, annually varying data from the Atmospheric Model Intercomparison Project (AMIP) dataset for the period 1979–93 (Gates 1992). In this dataset, when significant sea ice is present, the SST is assigned a value of −1.8°C. In the CCM3, grid points with this value are assumed to be completely covered with ice 2 m deep, with 5 mm (water equivalent) of overlying snow. The ocean beneath the ice has a temperature fixed at −2°C and the surface temperature is determined using a four-layer diffusion model. This treatment of ice in the CCM3 results in ice that is too deep and too extensive compared to observations in the SH. The CSM1 is more conservative in its treatment of ice, allowing partial grid ice coverage; thus its surface temperatures are warmer and closer to observations around Antarctica in winter (Boville and Hurrell 1998). Weatherly et al. (1998) show that in the SH the simulated ice-covered area and the interannual variability of the ice-covered area in CSM1 are similar to observations. Also the ice thickness pattern in the Antarctic region is realistic but too thin away from the continent. In this issue, Kiehl et al. (1998) give a comprehensive description of the CCM3, and Boville and Hurrell (1998) examine the climate simulated by the CCM3.

The NCAR–NCEP reanalyses are an improvement on the old National Meteorological Center (NMC, now known as NCEP) data. They represent an international effort in which historic data (1957–96) were reanalyzed, incorporating data and methods that were previously unavailable. NCAR–NCEP reanalyses lack the inhomogeneities that have been introduced into the operational analyses due to changes in the NCEP model’s data assimilation system over many years. The new analyses are computed at a much finer resolution (T62) and do not contain model-induced climatic shifts. To make them directly comparable to the model data, the NCAR–NCEP data are spectrally truncated to T42 before processing. In this paper they are referred to as observed data.

CCM3 results from an integration performed with specified thermal forcing using observed, monthly averages of SST for the period 1979–93, NCAR–NCEP reanalyses for the concurrent period, and CSM1 simulations for 15 yr are compared. The CSM1 data are extracted from the years 11–24 of the 300-yr simulation, the period after the model reached equilibrium (no surface trends). The primary variable examined here is the geopotential height for the southern winter, July–September (JAS). Evaluation of the geopotential height variations simulated by the models should provide an assessment of their ability to simulate the large-scale atmospheric circulation, since they are strongly correlated to variations in the balanced, mainly rotational component of the large-scale flow that dominates the climate of the extratropics. Since geopotential heights are closely associated with temperature and zonal wind variations, the latter are used to help understand some of the variations seen in the geopotential heights and the quasi-stationary waves. However, a full analysis of the reasons underlying the performance of the models is beyond the scope of this paper.

Modeled and observed data at 1000, 850, 700, 500, 300, 200, and 100 mb are evaluated. Discussion of the geographical aspects of the circulation focus on the 500-mbar level. Fourier analysis is used to decompose the geopotential height field into its harmonic components and the first three waves are examined. The amplitudes and phases of the waves are determined from the mean Fourier coefficients, as done by van Loon et al. (1973). For each simulation, the amplitude, phase, and percent variation explained by the first three waves are compared to that of the observed data (NCAR–NCEP reanalyses).

4. Results and discussion

a. Geopotential height

The geopotential height distributions simulated by the models agree well with the observed data (Fig. 1). The isolines are concentric with lowest heights occurring over Antarctica and highest in the subtropics. Both models exhibit lower heights than observed over Antarctica, the result of cooler simulated tropospheric temperatures (Fig. 2), but are practically identical to the observed in the subtropics. In addition, the steeper gradients over the southern Indian Ocean and the gentler gradients over the southern Pacific Ocean are well simulated by both models.

The interannual variability in the geopotential height field at 500 mbar is shown in Fig. 3. In the observed (Fig. 3a), the variability is small compared to the mean height, and the greatest variability occurs in proximity to Antarctica. The most variable regions lie over the Bellingshausen and Amundsen Seas and eastern Antarctica. Least variable regions lie over the open oceans and at lower latitudes. The CSM1 simulation (Fig. 3c) very closely matches the observed magnitudes and spatial patterns of interannual variability. The CCM3 is less successful. While it simulates the general pattern of variability, it is weaker by almost 50%. This is particularly obvious over the oceans and over eastern Antarctica. Reasons for the limited success of the CCM3 in this regard are discussed later.

The differences between modeled and observed data are shown in Fig. 4. Over most of the SH the 500-mbar surface is higher in the observed than in the models. These differences are greatest in association with Antarctica, neither model appears markedly better than the other and the general pattern of differences (between simulation and observed) is similar. The differences between the CCM3 and the CSM1 are more distinctive, exhibiting a clear wave 2 pattern of alternating negative and positive values around Antarctica (Fig. 4c). The CCM3 500-mbar heights are higher than the CSM1 over the Antarctic Peninsula and between Australia and Antarctica. This pattern of geopotential height differences is echoed in the 500-mbar temperature differences (Fig. 2c). The differences in surface air temperatures between the models (Fig. 5) suggest that the 500-mbar patterns arise from the surface where the CCM3 is colder than the CSM1 except over the Antarctic peninsula and between Australia and Antarctica. This arises largely from the fact that in the the CCM3 sea-ice coverage is too large and too extensive. See section 3 above and Briegleb and Bromwich (1998).

The concentric alignment of the isolines in Fig. 1 belies the fact that there is strong zonal asymmetry in the geopotential height field. This is illustrated in Fig. 6. In the middle and high latitudes of the observed (Fig. 6a), there is a clear wave 1 in the field such that heights are higher than average over the southern Pacific Ocean and lower than average over the southern Atlantic and Indian Oceans. This asymmetry is a characteristic of the SH circulation and is closely associated with the temperature distribution (van Loon et al. 1973). Comparison with Fig. 7a shows that the lower than average heights are associated with lower than average temperatures and vice versa. Hurrell et al. (1998) suggest that these asymmetries are linked to the asymmetry of Antarctica (around the south pole) and the stress of the easterly surface winds on the water and ice along the periphery of the continent.

The simulated asymmetry in temperature (Figs. 7b,c) and heights (Figs. 6b,c) display good agreement with the observed. Differences occur largely in the size of the zonal anomaly, the simulated values being larger than the observed. The CSM1 captures the spatial structure better than the CCM3. The simulation of the zonal asymmetry in the height field is a necessary condition for representation of quasi-stationary wave 1 in the subpolar region. While the forcing mechanisms of this wave are not yet verified, it is clear that it is associated with the zonal asymmetries in the temperature and geopotential height fields (Hurrell et al. 1998).

b. Quasi-stationary wave analysis

1) Wave 1

In the SH, wave 1 is the most significant fluctuation in the observed geopotential height field (Fig. 8a). Vertically, it is marked by two well-defined peaks in amplitude in the upper troposphere; one between 30° and 40°S (subtropical) and one that propagates into the stratosphere between 55° and 60°S (subantarctic). A third smaller maximum occurs over Antarctica. Wave 1 is responsible for 80%–90% of the variance (not shown) in the subtropical and subantarctic troposphere. Over Antarctica, it is responsible for 20%–90%, with the lowest variance being explained immediately over the continental surface. As Figs. 8b,c illustrate, the mean subantarctic wave 1 simulated by the models is in excellent agreement with the observed, and it accounts for as much of the variance.

Despite the good agreement, some important differences in amplitude occur. The CCM3 has a stronger amplitude than the observed, by 20 gpm, whereas the CSM1 is weaker by 10 gpm. These differences in amplitude occur largely above 500 mb. Stationary wave structure is strongly dependent on the zonal wind intensity (e.g., Schoeberl and Geller 1976) and in these models, the differences between the observed and the simulated wave 1 can be associated with the zonal wind distribution. Figure 9 shows the zonally averaged zonal wind speeds. It is clear that in the subtropical maximum, the simulated wind speeds are greater than the observed. Also, the secondary maximum in observed wind speeds (from approximately 40°–60°S) is not well developed in the simulations. However, poleward of 60°S, the differences in magnitude appear slight. Figure 10 illustrates more clearly the differences between the modeled and observed zonal winds. The westerlies poleward of 60°S are weaker than observed, whereas equatorward of 60°S they are stronger than observed. The magnitudes of the zonal winds in the models suggest a slightly weaker polar vortex in the latitudes of the maximum in wave 1 and stronger midlatitude westerlies in the upper troposphere and lower stratosphere. Given the actual distribution of the winds in Fig. 9, it is possible that the weaker winds in Fig. 10 arise from a slight latitudinal shift in circulation rather than from a generally weakened circulation. However, these differences in the zonal winds can affect the transfer of energy and momentum, and therefore the quasi-stationary wave amplitude. These discrepancies occur at latitudes and at heights where differences in wave 1 are apparent.

The average state of the subantarctic wave 1 appears well simulated by the model in most respects. It is also of interest to note whether or not the interannual variation of wave 1 is captured by the models. Figure 11 shows the amplitude of wave 1 at 500 mbar as a function of time and longitude. In the observed (Fig. 11a), the phase of the wave does not vary significantly; however, although it is small there is a clear variation in the amplitude with a timescale that approaches 9 yr. Apart from the obvious differences in amplitude, this variation is approximated by the CCM3 (Fig. 11b). This could be expected for the CCM3 because it is forced by SSTs for the same time period as the observed data. Unlike the CCM3, the simulations of the CSM1 cannot be expected to match closely the period of observed data; however, Fig. 11c indicates that the CSM1 is successful at reproducing decadal-scale variation in wave 1.

The subtropical wave 1 simulated by CCM3 and the CSM1 propagates into the stratosphere (Figs. 8b,c). This is contrary to the observed. Also, over Antarctica the secondary maximum in wave 1 is only weakly simulated in the CSM and absent from the CCM3. While the discrepancies in the subtropics might also be explained by the differences in zonal wind intensity (suggested above), in the present analysis, it is not clear why the polar wave 1 is not simulated by the models. This wave 1 is confined to the continent, between the surface and 500 mbar poleward of 70°S, so its absence might be due to the specification of the topography or the spatial resolution of the land in the models. Also, both models exhibit lower temperatures and geopotential heights over the continent (not shown) than the observed suggesting a difference in the thermal forcing.

The phase of wave 1 is unchanging with height (not shown) in the subantarctic region in both the models and the observed. This is a characteristic feature of the SH quasi-stationary waves and is associated with the almost unbroken expanse of ocean, which allows the temperature and the geopotential height waves to be in phase with each other (van Loon and Jenne 1972). At 500 mbar the observed ridge of wave 1 is located over the eastern Atlantic in the subtropics and over the central Pacific at subantarctic latitudes (Fig. 12a). As in Fig. 8a, it has an amplitude minimum at 40°S where the wave nearly vanishes and there is very little wave activity over the continent itself. Both models place the ridge of wave 1 in approximately the same locations as that observed (Figs. 12b,c). The forcing mechanisms for wave 1 are not fully established [Quintanar and Mechoso (1995a) suggest lower latitude transient eddy forcing]; however, this wave is clearly related to the temperature, following closely the the pattern of zonal asymmetry in the temperature field. Compare with Fig. 7 (note the difference in map projection). Both models simulate the general distribution of these zonal temperature anomalies well and perhaps this underlies their ability to simulate this wave 1. At 500 mbar the simulated subtropical wave 1 also resembles the observed in amplitude and in phase.

2) Wave 2

In the observed data (Fig. 13), wave 2 is characterized by a peak in amplitude at 55°S where it accounts for 10% of the variance and a peak of similar magnitude over Antarctica between 70° and 80°S where it accounts for 20%–60% of the variance. There is little activity in the Tropics. The CCM3 simulates a stronger than observed wave 2 in the subtropics and over Antarctica. but is responsible for similar amounts of variance in the heights as the observed. The CSM1 does not simulate wave 2 over Antarctica but produces a very strong wave 2 at approximately 55°S and 30°S. At 55°S wave 2 strongly resembles wave 1. Some 20%–30% of the variance in the heights simulated by CSM1 is explained by this feature, while in the subtropics it is 15%.

The observed geographic distribution of wave 2 at 500 mbar is shown in Fig. 14a. The subantarctic and subtropical waves illustrated in Fig. 13 are clearly seen. Over the continent the ridges lie near 90°E and 90°W. Over Antarctica the CCM3 resembles the observed but is not as successful with the subantarctic wave 2. On the other hand, the CSM1 does not simulate wave 2 over Antarctica but captures the subantarctic wave, in both the amplitude and phase. Wave 2 is thought to be associated with the presence and orography of Antarctica since its ridges occur over the highest parts of the continent (van Loon and Jenne 1972). Therefore, it is not immediately clear why the models’ simulations over the continents differ since both models have identical orography. In the subantarctic, the difference between the models might be related to the surface temperature distributions that they simulate. The differences in temperature and the accompanying differences in geopotential heights (Figs. 2c and 3c) have a strong wave 2 component with a ridge at 90°W.

3) Wave 3

Wave 3 has been associated with the southern continents in the subtropical region, contributes very little to the variance in the geopotential height field, and its amplitude varies considerably from one period to the next (van Loon et al. 1993). In the observed data wave 3 has a peak in amplitude at 50°S where it propagates into the stratosphere, and a secondary peak over Antarctica (Fig. 15a). At 50°S wave 3 closely resembles waves 1 and 2 in structure. The CCM3 exhibits no wave 3 except for some small activity over Antarctica. The CSM1 produces a wave 3 that resembles the observed very closely but there is limited propagation into the stratosphere. Like the observed, in the CSM1, wave 3 explains no more than 10% of the variance. Figure 16 shows the geographic distribution of wave 3 at 500 mbar. In the observed, wave 3 lies between 25°S and 60°S with ridges at 50°S, in the vicinity of the three low latitude continents. The CSM1 simulates a strong wave 3 whose phase is displaced east of the observed. The phase of wave 3 simulated by the CCM3 is distinctly similar to the observed but the amplitude of the simulated wave is much weaker.

It is interesting that the CCM3 does not resolve a strong wave 3 in the period 1979–93 since van Loon et al. (1993) show that the amplitude of wave 3 quadrupled after 1977, and a relatively strong wave 3 is resolved from the observed. The wave differences between the CCM3 and the CSM1 are in latitudes whose surfaces are oceanic or composed of sea ice. Given that the major difference between the models in those latitudes is sea-ice extent and depth, then it is possible that wave 3 in the CCM3 is weak because of the thermal influence of too much ice.

There have been a number of numerical modeling studies on the effects of varying Antarctic sea-ice extents (e.g., Simmonds 1981; Mitchell and Hills 1986;Mitchell and Senior 1989). The results of these studies suggest that sea ice has a strong effect upon the transfer of energy between the ocean and atmosphere by reducing the fluxes of latent and sensible heat from ocean to atmosphere as well as by reducing the amount of solar radiation absorbed. The studies also show that changes in sea-ice extents affect the surface pressure as well as the upper-level tropospheric flow. To examine the possibility that wave 3 is affected by the presence of too much ice in the CCM3, recent results from a CCM3 simulation with sea-ice concentration derived from satellite observations for the same period (1979–93) as the CCM3 run discussed here were analyzed and are now presented. The boundary conditions of this recent run differ from the other only in its use of satellite-observed sea-ice concentrations and a slightly modified surface drag coefficient. The satellite-derived sea-ice concentrations resemble other observations closely (B. Boville 1997, personal communication).

Figure 17 depicts the zonal wave 3 resolved from the sea-ice concentration simulations. It is clear that compared to that in Fig. 15b, zonal wave 3 has a much higher amplitude and one that approximates that resolved from the NCAR–NCEP reanalyses (Figs. 15a and 16a). Wave 3 at 500 mb is well formed although its phase is shifted east like that of the CSM1 (Fig. 16c). Also, the interannual variation of the geopotential height and temperature fields (not shown) produced by this simulation resemble the observed much more closely than the initial CCM3 data. Given the foregoing discussion it is probable that in the CCM3, zonal wave 3 is strongly influenced by the effect of too much ice on the thermal influence of the surface. Perhaps the presence of the ice changes the surface flux of heat sufficiently to affect the surface temperature and the surface flow with the resulting reduction in the amplitude of zonal wave 3 in the initial CCM3 simulations.

5. Summary and conclusions

The major consequence of this study is the determination that both the CSM1 and CCM3 are able to simulate important aspects of the large-scale SH circulation. This has been confirmed by verification with the NCAR–NCEP reanalyses and by validating the models against each other. This is a significant achievement for the NCAR models since previous GCMs including the CCM2 have been limited in their ability to simulate the large-scale SH circulation (see, e.g., Xu et al. 1990). The success of the simulations varies between the models, however.

Despite the fact that one model is forced by specified SSTs and the other is coupled to an interactive ocean, wave 1, the dominant fluctuation in the large-scale SH circulation, is well represented by both models in the subantarctic latitudes, the region of strongest and most consistent fluctuation. This emphasizes that wave 1 is a stable and necessary fluctuation in the general circulation of the southern atmosphere. The success of the simulation of wave 1 lies in the models’ ability to reproduce the key features of the geopotential height and temperature fields, namely, their mean distribution, interannual variability, and the zonal asymmetry poleward of 40°S. These are essential for good simulation of wave 1. That the CCM3 can produce wave 1 as well as it does shows that an atmospheric GCM alone can be used to study the mean circulation in the SH. The discrepancies that appear in the model simulations of wave 1 appear to be due to problems largely within the atmospheric component of the model rather than the ocean.

The roles of the interactive ocean and in particular sea ice become perhaps more significant when waves 2 and 3 are examined. Wave 2 is better simulated by CSM1 over the southern oceans than by the CCM3. Wave 3 is only weakly simulated by the CCM3, whereas the CSM1 produces a strong, albeit phase-shifted wave 3. The chief differences between the two models are the interactive ocean and the specified sea-ice distribution. Not surprisingly therefore, the chief differences between their simulations occur in latitudes with oceanic and sea-ice surfaces. The differences in wave 2 simulations appear linked with surface temperature differences between the models. Thus, the sea-ice distributions may play an important role in forcing waves 2 and 3.

The role of sea ice becomes clearer when the results of a CCM3 run whose boundary conditions differ from the one under discussion by the use of satellite-observed sea-ice concentrations and a slightly modified surface drag coefficient are examined. Wave 3 produced by this simulation closely resembles the observed and it is suggested here that too much sea ice in the earlier run affected the flux of energy from the ocean to the atmosphere, and that in turn served to reduce the amplitude of zonal wave 3. Therefore, the discrepancies between the simulations, and between the modeled data and the observed, do not only indicate avenues for improving model importance. They also suggest some of the factors that might be important in forcing or maintaining the waves.

The distribution of SSTs (and now sea ice), the asymmetry (around the South Pole) of Antarctica, and the orography of Antarctica have all been suggested as forcing mechanisms for the fluctuations in the circulation. Some have suggested forcing from lower latitude transient activity (Quintanar and Mechoso 1995a). However, it is not yet firmly proven what the relative roles of these elements are in forcing the general circulation in the SH. In the NH, the roles of thermal forcing and topography in the generation of the stationary waves in winter are much clearer. However, in the SH the same is not true. One reason is that the physical processes underlying the climate are better understood and specified in work that focused on the NH. The CCM3 and CSM1 both simulate wave 1 sufficiently well enough to allow a more detailed examination of the mechanisms that force the primary quasi-stationary wave in the model atmosphere. To this end, we have conducted a number of experiments designed to alter the symmetry and topography of Antarctica and the symmetry of the SST distribution in the CCM3, in order to isolate their role in forcing and thereby generate a working hypothesis for the real world. This will be reported upon in later articles.

Acknowledgments

I thank the CSM group at NCAR for the funding and the data that made this work possible. I also thank Harry van Loon, James Hurrell, Byron Boville, Roberto Mechoso, and two anonymous reviewers for their helpful discussions during the analysis and write up of this article.

REFERENCES

  • Bonan, G., 1998: The land surface climatology of the NCAR Land Surface Model (LSM 1.0) coupled to the NCAR Community Climate Model (CCM3). J. Climate,11, 1307–1326.

  • Boville, B. A., and P. R. Gent, 1998: The NCAR Climate System Model, version one. J. Climate,11, 1115–1130.

  • ——, and J. W. Hurrell, 1998: A comparison of the atmospheric circulations by the CCM3 and CSM1. J. Climate,11, 1327–1341.

  • Briegleb, B., and D. H. Bromwich, 1998: Polar climate simulation of the NCAR CCM3. J. Climate,11, 1270–1286.

  • Bryan, F. O., B. G. Kauffman, W. G. Large, and P. R. Gent, 1996:The NCAR CSM Flux Coupler. National Center for Atmospheric Research NCAR/TN-424+STR, 46 pp.

  • Gates, W. L., 1992: AMIP: The Atmospheric Model Intercomparison Project. Bull. Amer. Meteor. Soc.,72, 1962–1970.

  • Gent, P. R., J. Willebrand, T. J. McDougall, and J. C. McWilliams, 1995: Parameterizing eddy-induced tracer transports in ocean circulation models. J. Phys. Oceanogr.,25, 463–474.

  • ——, F. O. Bryan, G. Danabasoglu, S. C. Doney, W. R. Holland, W. G. Large, and J. C. McWilliams, 1998: The NCAR Climate System Model global ocean component. J. Climate. 11, 1287–1306.

  • Hurrell, J. W., 1995: Comparison of the NCAR Community Climate Model (CCM) climates. Climate Dyn.,11, 25–50.

  • ——, H. van Loon, and D. J. Shea, 1998: The mean state of the troposphere. Meteorology of the Southern Hemisphere, D. Karoly and D. Vincent, Eds., Amer. Meteor. Soc., in press.

  • Kiehl, J. T., J. J. Hack, G. Bonan, B. A. Boville, D. Williamson, and P. Rasch, 1998: The National Center for Atmospheric Research Community Climate Model: CCM3. J. Climate,11, 1131–1149.

  • Large, W. G., J. C. McWilliams, and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parameterization. Rev. Geophys.,32, 363–403.

  • Mitchell, J. F. B., and T. S. Hills, 1986: Sea-ice and the antarctic winter circulation, a numerical experiment. Quart. J. Roy. Meteor. Soc.,112, 953–969.

  • ——, and C. A. Senior, 1989: The antarctic winter; simulations with climatological and reduced sea-ice extents. Quart. J. Roy. Meteor. Soc.,115, 225–246.

  • Quintanar, A. I., and C. R. Mechoso, 1995a: Quasi-stationary waves in the Southern Hemisphere. Part II: Generation Mechanisms. J. Climate,8, 2673–2690.

  • ——, and ——, 1995b: Quasi-stationary waves in the Southern Hemisphere. Part I: Observational data. J. Climate,8, 2659–2672.

  • Schoeberl, M. R., and M. A. Geller, 1976: The structure of stationary planetary waves in winter in relation to the polar night jet intensity. Geophys. Res. Lett.,3, 177–180.

  • Simmonds, I., 1981: The effect of sea-ice on a general circulation model of the Southern Hemisphere. Sea Level, Ice and Climatic Change, I. Allison, Ed., IAHS, 253 pp.

  • ——, and Y. Lin, 1983: Topographical and thermal forcing in a circulation model of the Southern Hemisphere. Publication No. 24, University of Melbourne, Meteorology Department, Melbourne, Victoria, Australia, 78 pp.

  • Trenberth, K. E., 1979: Interannual variability of the 500 mbar zonal mean flow in the Southern Hemisphere. Mon. Wea. Rev.,107, 1515–1524.

  • ——, 1980: Planetary waves at 500 mb in the Southern Hemisphere. Mon. Wea. Rev.,108, 1378–1389.

  • van Loon, H. V., and R. Jenne, 1972: The zonal harmonic standing waves in the Southern Hemisphere. J. Geophys. Res.,77, 992–1003.

  • ——, ——, and K. Labitzke, 1973: Zonal harmonic standing waves. J. Geophys. Res.,78, 4463–4471.

  • ——, J. W. Kidson, and A. B. Mullen, 1993: Decadal variation of the annual cycle in the Australian dataset. J. Climate,6, 1227–1231.

  • Watterson, I. G., and I. N. James, 1992: Baroclinic waves propagating from a high latitude source. Quart. J. Roy. Meteor. Soc.,118, 23–50.

  • Weatherly, J., B. Briegleb, W. G. Large, and T. Bettge, 1998: Sea ice and polar climate in the NCAR CSM. J. Climate,11, 1472–1486.

  • Xu, J.-S., H. von Storch, and H. van Loon, 1990: The performance of four spectral GCMs in the Southern Hemisphere: The January and July climatology and the semi-annual wave. J. Climate,3, 53–70.

 Fig. 1.
Fig. 1.

Geopotential height distribution at 500 mbar for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 100 m.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

 Fig. 2.
Fig. 2.

Distribution of temperature differences at 500 mbar for (a) NCEP–CCM3, (b) NCEP–CSM1, and (c) CCM3–CSM1. Contour interval is 0.5 K.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

 Fig. 3.
Fig. 3.

Standard deviation of the geopotential heights at 500 mbar for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 m.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

 Fig. 4.
Fig. 4.

Geopotential height differences at 500 mbar for (a) NCEP–CCM3, (b) NCEP–CSM1, and (c) CCM3–CSM1. Contour interval is 20 m.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 5.
Fig. 5.

Distribution of surface air temperature differences for CCM3–CSM1. Contour interval is 3 K.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

 Fig. 6.
Fig. 6.

Geopotential height deviations from the zonal mean at 500 mbar for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 20 m.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 7.
Fig. 7.

Temperature deviations from the zonal mean at 500 mbar for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 0.5 K.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 8.
Fig. 8.

Zonal-mean amplitude of quasi-stationary wave 1 for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 gpm.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 9.
Fig. 9.

Latitude–height section of the mean zonal wind speed for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 5 m s−1.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 10.
Fig. 10.

Latitude–height section of the difference in mean zonal wind speed for (a) NCEP–CCM3 and (b) NCEP–CSM1. Contour interval is 1 m s−1.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 11.
Fig. 11.

Time–longitude variation of quasi-stationary wave 1 at 60°S at 500 mbar: (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 20 gpm.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 12.
Fig. 12.

Amplitude of quasi-stationary wave 1 at 500 mbar: (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 25 gpm.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 13.
Fig. 13.

Zonal-mean amplitude of quasi-stationary wave 2 for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 gpm.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 14.
Fig. 14.

Amplitude of quasi-stationary wave 2 at 500 mbar: (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 gpm.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 15.
Fig. 15.

Zonal-mean amplitude of quasi-stationary wave 3 for (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 10 gpm.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 16.
Fig. 16.

Amplitude of quasi-stationary wave 3 at 500 mbar: (a) NCEP, (b) CCM3, and (c) CSM1. Contour interval is 5 gpm.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

Fig. 17.
Fig. 17.

Amplitude of quasi-stationary wave 3 for CCM3 with satellite-observed concentrations. Contour interval is 10 gpm.

Citation: Journal of Climate 11, 6; 10.1175/1520-0442(1998)011<1405:QSWITS>2.0.CO;2

* An electronic supplement to this article may be found on the CD-ROM accompanying this issue or at http://www.ametsoc.org/AMS.

Save