• Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor, 4 , 11471167.

    • Search Google Scholar
    • Export Citation
  • Boer, G. J., and M. Lazare, 1988: Some results concerning the effect of horizontal resolution and gravity wave drag on simulated climate. J. Climate, 1 , 789806.

    • Search Google Scholar
    • Export Citation
  • Boville, B. A., 1991: Sensitivity of simulated climate to model resolution. J. Climate, 4 , 469485.

  • Bryan, F. O., G. Danabasoglu, N. Nakashiki, Y. Yoshida, D-H. Kim, J. Tsutsui, and S. C. Doney, 2006: Response of North Atlantic thermohaline circulation and ventilation to increasing carbon dioxide in CCSM3. J. Climate, 19 , 23822397.

    • Search Google Scholar
    • Export Citation
  • Chen, T-C., and J. J. Tribbia, 1993: An effect of the model's horizontal resolution on stationary eddies simulated by the NCAR CCM1. J. Climate, 6 , 16571664.

    • Search Google Scholar
    • Export Citation
  • Chin, T. M., R. F. Milliff, and W. G. Large, 1998: Basin-scale, high-wavenumber sea surface wind fields from a multiresolution analysis of scatterometer data. J. Atmos. Oceanic Technol, 15 , 741763.

    • 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
  • Danabasoglu, G., W. G. Large, J. J. Tribbia, P. R. Gent, B. P. Briegleb, and J. C. McWilliams, 2006: Diurnal coupling in the tropical oceans of CCSM3. J. Climate, 19 , 23472365.

    • Search Google Scholar
    • Export Citation
  • Davey, M. K., and Coauthors, 2002: STOIC: A study of coupled model climatology and variability in tropical ocean regions. Climate Dyn, 18 , 403420.

    • Search Google Scholar
    • Export Citation
  • Deser, C., A. Capotondi, R. Saravanan, and A. Phillips, 2006: Tropical Pacific and Atlantic variability in CCSM3. J. Climate, 19 , 24512481.

    • Search Google Scholar
    • Export Citation
  • DeWeaver, E., and C. M. Bitz, 2006: Atmospheric circulation and its effect on Arctic sea ice in CCSM3 simulations at medium and high resolution. J. Climate, 19 , 24152432.

    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., K. W. Oleson, G. Bonan, F. Hoffman, P. Thornton, M. Vertenstein, Z-L. Yang, and X. Zeng, 2006: The Community Land Model and its climate statistics as a component of the Community Climate System Model. J. Climate, 19 , 23022324.

    • Search Google Scholar
    • Export Citation
  • Fekete, B., C. Vorosmarty, and W. Grabs, 2002: High-resolution fields of global runoff combining observed river discharge and simulated water balance. Global Biogeochem. Cycles, 16 .1042, doi:10.1029/1999GB001254.

    • Search Google Scholar
    • Export Citation
  • Hack, J. J., J. M. Caron, S. G. Yeager, K. W. Oleson, M. M. Holland, J. E. Truesdale, and P. J. Rasch, 2006: Simulation of the global hydrological cycle in the CCSM Community Atmosphere Model version 3 (CAM3): Mean features. J. Climate, 19 , 21992221.

    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., J. J. Hack, A. Phillips, J. Caron, and J. Yin, 2006: The dynamical simulation of the Community Atmospheric Model version 3 (CAM3). J. Climate, 19 , 21622183.

    • Search Google Scholar
    • Export Citation
  • Johnson, G. C., B. M. Sloyan, W. S. Kessler, and K. E. McTaggart, 2002: Direct measurements of upper ocean currents and water properties across the tropical Pacific Ocean during the 1990s. Progress in Oceanography, Vol. 52, Pergamon, 31–36.

    • Search Google Scholar
    • Export Citation
  • Kållberg, P., A. Simmons, S. Uppala, and M. Fuentes, 2004: The ERA-40 archive. Tech. Rep. ERA-40 Project Rep. 17, European Centre for Medium-Range Weather Forecasts, Reading, United Kingdom, 35 pp.

  • Kiehl, J. T., 1994: On the observed near cancellation between longwave and shortwave cloud forcing in tropical regions. J. Climate, 7 , 559565.

    • Search Google Scholar
    • Export Citation
  • Kiehl, J. T., and V. Ramanathan, 1990: Comparison of cloud forcing derived from the Earth Radiation Budget Experiment with that simulated by the NCAR Community Climate Model. J. Geophys. Res, 95 , 1167911698.

    • Search Google Scholar
    • Export Citation
  • Kiehl, J. T., and D. L. Williamson, 1991: Dependence of cloud amount on horizontal resolution in the National Center for Atmospheric Research Community Climate Model. J. Geophys. Res, 96 , 1095510980.

    • Search Google Scholar
    • Export Citation
  • Kiehl, J. T., and K. E. Trenberth, 1997: Earth's annual global mean energy budget. Bull. Amer. Meteor. Soc, 78 , 197208.

  • Kiehl, J. T., C. A. Shields, J. J. Hack, and W. D. Collins, 2006: The climate sensitivity of the Community Climate System Model version 3 (CCSM3). J. Climate, 19 , 25842596.

    • Search Google Scholar
    • Export Citation
  • King, M. D., and Coauthors, 2003: Cloud and aerosol properties, precipitable water, and profiles of temperature and water vapor from MODIS. IEEE Trans. Geosci. Remote Sens, 41 , 442458.

    • Search Google Scholar
    • Export Citation
  • Kistler, R., and Coauthors, 2001: The NCEP–NCAR 50-year reanalysis: Monthly means CD-ROM and documentation. Bull. Amer. Meteor. Soc, 82 , 247267.

    • Search Google Scholar
    • Export Citation
  • Large, W. G., and G. Danabasoglu, 2006: Attribution and impacts of upper-ocean biases in CCSM3. J. Climate, 19 , 23252346.

  • Levitus, S., and Coauthors, 1998: Introduction. Vol. 1, World Ocean Database 1998, NOAA Atlas NESDIS 18, 346 pp.

  • Manabe, S. J., J. L. Smagorinsky, J. Holloway, and H. M. Stone, 1970: Simulated climatology of a general circulation model with a hydrologic cycle: III. Effects of increased horizontal computational resolution. Mon. Wea. Rev, 98 , 175212.

    • Search Google Scholar
    • Export Citation
  • Manabe, S. J., D. G. Hahm, and J. L. Holloway Jr., 1979: Climate simulations with GFDL spectral models of the atmosphere: Effect of spectral truncation. JOC Study Conf. on Climate Models: Performance, Intercomparison and Sensitivity Studies, Geneva, Switzerland, WMO, Global Atmospheric Research Programme Publication Series 22, Vol. 1, 41–94.

  • Moncrieff, M. W., S. K. Krueger, D. Gregory, J-L. Redelsperger, and W-K. Tao, 1997: GEWEX cloud system study (GCSS) working group 4: Precipitating convective cloud systems. Bull. Amer. Meteor. Soc, 78 , 831845.

    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and Coauthors, 2004: Effects of land use change on North American cilmate: Impact of surface datasets and model biogeophysics. Climate Dyn, 23 , 117132.

    • Search Google Scholar
    • Export Citation
  • Pope, V. D., and R. A. Stratton, 2002: The processes governing horizontal resolution sensitivity in a climate model. Climate Dyn, 19 , 211236.

    • Search Google Scholar
    • Export Citation
  • Ramanathan, V., and W. Collins, 1991: Thermodynamic regulation of ocean warming by cirrus clouds deduced from observations of the 1987 El Niño. Nature, 351 , 2732.

    • Search Google Scholar
    • Export Citation
  • Randel, D. L., T. H. Vonder Haar, M. A. Ringerud, G. Stephens, T. J. Greenwald, and C. L. Combs, 1996: A new global water vapor dataset. Bull. Amer. Meteor. Soc, 77 , 12331246.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc, 80 , 22612287.

  • Steele, M., R. Morley, and W. Ermold, 2001: PHC: A global ocean hydrography with a high-quality Arctic Ocean. J. Climate, 14 , 20792087.

    • Search Google Scholar
    • Export Citation
  • Warren, S. G., C. J. Hahn, J. London, R. M. Chervin, and M. L. Jenne, 1988: Global distribution of total cloud cover and cloud type amounts over the ocean. Tech. Rep. NCAR Tech. Note TN-317+STR, National Center for Atmospheric Research, 42 pp. + 170 maps.

  • Whitworth III, T., 1983: Monitoring the transport of the Antarctic Circumpolar Current at Drake Passage. J. Phys. Oceanogr, 13 , 20452057.

    • Search Google Scholar
    • Export Citation
  • Whitworth III, T., and R. G. Peterson, 1985: Volume transport of the Antarctic Circumpolar Current from bottom pressure measurements. J. Phys. Oceanogr, 15 , 810816.

    • Search Google Scholar
    • Export Citation
  • Williamson, D. L., 1988: The effect of vertical finite difference approximations on simulations with the NCAR Community Climate Model. J. Climate, 1 , 4058.

    • Search Google Scholar
    • Export Citation
  • Williamson, D. L., J. T. Kiehl, and J. J. Hack, 1995: Climate sensitivity of the NCAR Community Climate Model (CCM2) to horizontal resolution. Climate Dyn, 11 , 377397.

    • Search Google Scholar
    • Export Citation
  • Willmott, C. J., and K. Matsuura, 2000: Terrestrial air temperature and precipitation: Monthly and annual climatologies. [Available online at http://climate.geog.udel.edu/~climate.].

  • Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc, 78 , 25392558.

    • Search Google Scholar
    • Export Citation
  • Yeager, S. G., C. A. Shields, W. G. Large, and J. J. Hack, 2006: The low-resolution CCSM3. J. Climate, 19 , 25452566.

  • Zhang, Y. C., W. B. Rossow, A. A. Lacis, V. Oinas, and M. I. Mishchenko, 2004: Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets: Refinements of the radiative transfer model and the input data. J. Geophys. Res, 109 .D19105, doi:10.1029/2003JD004457.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Large-scale advective temperature tendency in K day−1 for (top) the T42 CAM3, (middle) the T85 CAM3, and (bottom) Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE) dataset (Moncrieff et al. 1997).

  • View in gallery

    Zonally averaged annual temperature difference (left) between the T85 CAM3 and the T42 CAM3 and (right) between the T42 CAM3 and ERA-40. Contour intervals are 0.8 K.

  • View in gallery

    Scatterplots showing shortwave vs longwave cloud forcing in the tropical west Pacific warm pool region for (left) the T42 CAM3, (middle) the T85 CAM3, and (right) ERBE.

  • View in gallery

    (top) The net longwave flux difference and (bottom) net shortwave flux difference between the T85 CAM3 and the T42 CAM3 in W m −2.

  • View in gallery

    (left) The absorbed solar radiation anomaly for the T42 CAM3, the T85 CAM3, and ERBE for ENSO warm minus cold conditions. (right) The outgoing longwave radiation anomaly for the T42 CAM3, the T85 CAM3, and ERBE for ENSO warm minus cold conditions. Contour intervals are 5 W m −2.

  • View in gallery

    Surface energy residual for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) their difference in W m −2 for the Tropics. The contour interval for the top and middle panels is 25. The contour levels for the bottom panel are (−80.0, −60.0, −40.0, −30.0, −20.0, −10.0, −5.0, −1.0, 0.0, 1.0, 5.0, 10.0, 20.0, 30.0, 40.0, 60.0, 80.0).

  • View in gallery

    Shortwave cloud forcing difference between the T85 CAM3 and the T42 CAM3 for (left) North America, (middle) South America, and (right) Africa. Contour intervals are 5 W m−2.

  • View in gallery

    (top) The surface wind stress vectors for (left) the T85 CAM3, (middle) the T42 CAM3, and (right) their difference. (bottom) The surface wind stress vectors for (left) ERS, (middle) the difference between T85 CAM3 and ERS, and (right) the difference between T42 CAM3 and ERS. Color contours show corresponding surface geopotential height fields (PHIS) in units of 10−3 m2 s−2.

  • View in gallery

    (left) The surface wind stress magnitude (color contours) and vectors over the Tropics for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) their difference. (right) The surface wind stress for (top) ERS, (middle) the difference between T85 CAM3 and ERS, and (bottom) the difference between T42 CAM3 and ERS.

  • View in gallery

    The surface wind stress magnitude (color contours) and vectors over the Southern Hemisphere for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) the T85–T42 difference. Contour intervals are 0.02 m s−2 for wind magnitudes and 0.01 m s−2 for differences.

  • View in gallery

    Near-surface wind difference (top) between the T85 CAM3 and the T42 CAM3 and (bottom) between the ERA-40 and T42 CAM3 in m s−1.

  • View in gallery

    Annually averaged precipitation over the tropical Indian and Pacific Oceans for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) the T85–T42 difference. Contour intervals are 1 mm day−1 for total precipitation rate and 0.25 mm day−1 for differences.

  • View in gallery

    June–August (JJA) mean precipitation over the tropical oceans for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) their difference. Contour intervals are 1 mm day−1 for total precipitation rate and 0.5 mm day−1 for differences.

  • View in gallery

    Annually averaged precipitation over the tropical Indian and Pacific Oceans for (top) the T85 CCSM3, (middle) the T42 CCSM3, and (bottom) the T85–T42 difference. Contour intervals are 1 mm day−1 for total precipitation rate and 0.25 mm day−1 for differences.

  • View in gallery

    Time- and horizontal-mean global differences between models and observations for potential (left) temperature and (right) salinity.

  • View in gallery

    Time-mean, vertically integrated mass transport (barotropic) streamfunction from T42x1 and its difference from the T85x1 solution. (top) Contour intervals are 10 and 20 Sv for transports smaller and greater than 60 Sv, respectively. Also, the thick and thin (shaded regions) lines denote clockwise and counterclockwise circulations, respectively. (bottom) The contour interval is 5 Sv, and the negative differences (thin lines) are shaded.

  • View in gallery

    Seasonally averaged 2-m temperature difference between (left) the T85x1 and observations and (right) the T42x1 and observations for (top) December–February (DJF) and (bottom) JJA. Observations are from Willmott and Matsuura (2000).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 187 105 13
PDF Downloads 75 43 4

CCSM–CAM3 Climate Simulation Sensitivity to Changes in Horizontal Resolution

View More View Less
  • 1 National Center for Atmospheric Research, Boulder, Colorado
  • | 2 University of Washington, Seattle, Washington
  • | 3 National Center for Atmospheric Research, Boulder, Colorado
Full access

Abstract

The latest version of the Community Climate System Model (CCSM) Community Atmosphere Model version 3 (CAM3) has been released to allow for numerical integration at a variety of horizontal resolutions. One goal of the CAM3 design was to provide comparable large-scale simulation fidelity over a range of horizontal resolutions through modifications to adjustable coefficients in the parameterized treatment of clouds and precipitation. Coefficients are modified to provide similar cloud radiative forcing characteristics for each resolution. Simulations with the CAM3 show robust systematic improvements with higher horizontal resolution for a variety of features, most notably associated with the large-scale dynamical circulation. This paper will focus on simulation differences between the two principal configurations of the CAM3, which differ by a factor of 2 in their horizontal resolution.

Corresponding author address: Dr. James J. Hack, NCAR, P.O. Box 3000, Boulder, CO 80307. Email: jhack@ncar.ucar.edu

Abstract

The latest version of the Community Climate System Model (CCSM) Community Atmosphere Model version 3 (CAM3) has been released to allow for numerical integration at a variety of horizontal resolutions. One goal of the CAM3 design was to provide comparable large-scale simulation fidelity over a range of horizontal resolutions through modifications to adjustable coefficients in the parameterized treatment of clouds and precipitation. Coefficients are modified to provide similar cloud radiative forcing characteristics for each resolution. Simulations with the CAM3 show robust systematic improvements with higher horizontal resolution for a variety of features, most notably associated with the large-scale dynamical circulation. This paper will focus on simulation differences between the two principal configurations of the CAM3, which differ by a factor of 2 in their horizontal resolution.

Corresponding author address: Dr. James J. Hack, NCAR, P.O. Box 3000, Boulder, CO 80307. Email: jhack@ncar.ucar.edu

1. Introduction

Solution of the continuous nonlinear differential equations governing atmospheric motions requires the use of a discrete approximation, most frequently utilizing finite-difference or spectral methods. The horizontal and vertical resolution at which global climate applications are numerically integrated is normally chosen on the basis of computational expense, generally weighed against gross measures of solution accuracy. To place any confidence in numerical simulations of climate, the horizontal and vertical resolution must be fine enough to accurately represent the phenomenological motion scales of most importance to the climate system. A common spectral truncation used in global climate models is a 42-wave triangular truncation(T42), which provides an isotropic representation of scalar information and very accurately treats features and their horizontal derivatives down to approximately 950 km. Motion scales below this truncation limit must be treated in some other way, and generally enter the solution in the form of a forcing term. In a spectral model these subgrid-scale terms are evaluated on a transform grid that is also used to evaluate nonlinear terms in the equations, and whose grid intervals are directly related to the spectral truncation. In a T42 model the transform grid interval is approximately 300 km at the equator. These terms are almost always evaluated using parameterization techniques, which can be highly nonlinear and are generally functions of the explicitly resolved atmospheric state variables. Ideally, one would select a horizontal resolution for which the solutions are in a convergent regime; that is, one in which additional increases in resolution would not greatly alter the solutions. Under such circumstances it might also be expected that the behavior of parameterized forcing terms would not significantly change with additional increases in resolution.

Exploration of global atmospheric simulation sensitivity to horizontal resolution goes back more than 30 yr (Manabe et al. 1970), and has continued sporadically in the intervening years (e.g., Manabe et al. 1979; Boer and Lazare 1988; Boville 1991; Kiehl and Williamson 1991; Chen and Tribbia 1993; Pope and Stratton 2002). Most investigations have identified some systematic improvements related to increases in horizontal resolution. In an earlier version of the CAM, Williamson et al. (1995) showed that many statistics used to characterize climate properties began to converge in the range of a T63 spectral truncation for midlatitudes. They also showed that scales of motion included at T63 and higher resolutions were needed to capture the nonlinear processes that appear to drive some larger scale circulations. The more discouraging outcome of that investigation was that they were unable to demonstrate convergence for many other quantities, even at a T106 truncation.

In this paper we will explore simulation sensitivity to horizontal resolution in the most recent version of the Community Climate System Model (CCSM) Community Atmosphere Model version 3 (CAM3). This model is the latest in a succession of atmospheric general circulation models that have been made widely available to the scientific community, originating with the National Center for Atmospheric Research (NCAR) Community Climate Model (CCM). The CAM3 incorporates a significant number of changes to the dynamical formulation, the treatment of cloud and precipitation processes, radiation processes, and atmospheric aerosols, and is described in Collins et al. (2006b). One of the unique design goals for the CAM3 was to provide simulations with comparable large-scale fidelity over a range of horizontal resolutions. This is accomplished through modifications to adjustable coefficients in the parameterized physics package associated with clouds and precipitation. The standard configuration of the CAM3 is based on an Eulerian spectral dynamical core, where the vertical discretization makes use of 26 levels (L26) treated using second-order finite differences (Williamson 1988). Three standard configurations of the CAM3 are distributed, including horizontal spectral truncations of T31 (∼3.75° transform grid), T42 (∼2.8° transform grid), and T85 (∼1.4° transform grid). The discussion that follows will focus on simulation differences between the T85L26 and T42L26 CAM3 configurations, which we anticipate will be the most commonly used versions. Twenty-two-year uncoupled simulations, using observed sea surface temperatures (SSTs) and observed sea ice, are used to characterize features of the simulated climate at the two resolutions. These simulation characteristics are then contrasted with simulation properties obtained from the fully coupled CCSM3 (Collins et al. 2006a). Characteristics of the T31L26 simulation are discussed separately in Yeager et al. (2006) for uncoupled applications, and when coupled to a nominal 3° resolution ocean model.

2. Tuning of the physical parameterization package

Changes to horizontal resolution in a global atmospheric model directly affect the scales of motion available for the explicit solution of the governing equations. For example, doubling the horizontal resolution from T42 to T85 allows motion scales that did not exist in the T42 model to be explicitly resolved in the solution of the large-scale equations of motion. A familiar adjustment to compensate for such changes is the need to maintain reasonable energy characteristics for the smallest resolved scales. In the case of the CAM3 spectral model, this requires that the coefficient on the biharmonic diffusion operator vary with resolution. This variation is determined experimentally so that in the mid- to upper troposphere the two-dimensional kinetic energy spectra have a reasonably well-behaved distribution as a function of the high-order wavenumbers (i.e., near the truncation limit), an approach discussed in detail by Boville (1991).

Similarly, the behavior of the parameterized treatment of physical processes also changes with resolution, for a variety of reasons. One example is shown in Fig. 1, which illustrates the time series of the resolved three-dimensional advective tendency of temperature over the tropical West Pacific warm pool (∼1°N, 155°E) for the T42 and T85 CAM3 configurations, along with an observationally derived version of the same field. We note that the T85 time series has been spatially averaged to the equivalent T42 area average for a fair comparison. This quantity is one of the primary destabilizing resolved-scale terms seen by the parameterized treatment of moist convection. It is clear from simple inspection that the fundamental character of this term changes with resolution in terms of vertical structure, temporal behavior, and the amplitude of the deviations from the long-term mean. The high-resolution time series is much more consistent with observational estimates of this quantity (bottom panel). This is admittedly a convolved product of both the additional scales of explicitly resolved motions and their interactions with the parameterized physics. But as shown in Hack and Caron (2005, unpublished manuscript) the more realistic behavior of the resolved motion field at higher resolution is very robust and largely determined by the internal behavior of the dynamical motion field. As might be expected, the temporal characteristics of parameterized processes like precipitation or cloud water [e.g., their probability density functions (PDFs)] differ significantly, even for cases where the time-mean properties are very similar.

A second example of how changes in resolution can affect the behavior of parameterized physics is shown in Fig. 2. This figure shows the difference in zonally and annually averaged temperature between the T85 and T42 CAM3 simulations, as well as the T42 CAM3 simulation and the European Centre for Medium-Range Weather Forecasts (ECMWF) 40-yr Re-Analysis (ERA-40) reanalysis. The figure shows a systematic warming of the troposphere, with the largest signals at high latitudes. The enhanced warming associated with higher resolution is a desirable signal when compared with observational estimates, particularly at high latitudes. It also is a very robust signal that has a weak dependence on the formulation of the parameterized physics package and is presumed to be attributable to improved accuracy in the treatment of the large-scale motion field. Systematic changes like this warming can have a significant impact on the treatment of parameterized processes. For example, the CAM3 exploits relative humidity thresholds in the treatment of cloud formation. Systematic changes in the temperature field accompanying changes in horizontal resolution require that the selection of these thresholds be revisited in order to maintain a similar cloud field and similar cloud radiative properties.

Changing free parameters in parameterized physics formulations, such as relative humidity thresholds or cloud water autoconversion thresholds in pursuit of a specific simulation goal is most frequently, and often disparagingly, referred to as tuning. Tuning generally involves the exploration of simulation sensitivity to a limited number of loosely constrained coefficients in the parameterized physics. The most common goal is to identify a parameterization configuration that yields simulation results that best agree with observations on some arbitrary combination of time and space scales. Generally, these time and space scales involve zonally averaged seasonal means of quantities like cloud radiative forcing (CRF). In the case of CAM3, adjustments are made to spatially and temporally invariant coefficients incorporated in the physical parameterization package. For the CAM3, these parameters include large-scale relative humidity thresholds on cloud formation, rainfall evaporation efficiencies in stratiform and convective precipitation processes, adjustment time scales associated with moist convection, and autoconversion thresholds for transforming cloud and ice water to rainwater and snow, respectively.

The goal of modifying the physics package as a function of horizontal resolution in the CAM3 is to ensure that the top-of-model (TOM) energy budget is closed and that the top-of-atmosphere (TOA) energy budget components are as close to observational estimates as possible, at all resolutions. We note that there is a subtle difference between TOM and TOA since there is a small amount of atmospheric mass above the top of the model, where TOM radiative fluxes define the net source or sink of energy to the simulated atmosphere. The nonlinear tuning exercise is done experimentally, exploiting expert knowledge of the way in which the various physical processes are formulated and are likely to interact. The CAM3 was initially developed at the T42 resolution, and the T31 and T85 configurations were developed as derivatives. A selected set of global annual climate metrics for the T42 and T85 CAM3 configurations are shown in Table 1. The TOM annual energy balance remains well within 0.5 W m−2 for both resolutions, where the component fluxes in the energy budget are both within 1 W m−2 between the two resolutions, and well within observational uncertainty. It is reasonable to ask how these quantities would compare without changes to the physics parameterizations. Using the T42 physics parameters, a T85 configuration would have a TOM and surface energy imbalance of −3.6 W m−2, principally due to differences in the longwave portion of the energy budget. This would render the configuration unsuitable for coupled simulation applications. Most other measures in Table 1 are virtually unchanged with the exception of the vertical distribution of cloud, the clear-sky TOM longwave flux, the all-sky net surface longwave flux, and the surface sensible heat flux. The change in the cloud distribution is required to keep the all-sky top-of-model radiative fluxes relatively unchanged. The changes in the clear-sky longwave flux are attributable to subtle changes in the vertical distribution of water vapor and changes to the tropospheric static stability. Generally speaking, the T42 and T85 configurations are very similar to each other in terms of the large-scale global annual energy and water cycle budget. However, as we will show, the details of how this balance is maintained can be quite different depending on the spatial and temporal averaging procedures.

An important check on the changes made to the cloud and precipitation processes is to examine the response of the cloud field to anomalies in SST. One approach exploits the observed linear correlation between longwave cloud forcing (LWCF) and shortwave cloud forcing (SWCF) as discussed in Kiehl and Ramanathan (1990), Ramanathan and Collins (1991), and Kiehl (1994). Figure 3 shows this relationship over the tropical western Pacific (10°S–20°N, 110°–160°E) for both Earth Radiation Budget Experiment (ERBE) and the T42 and T85 configurations of CAM3. The ERBE data show a strong linear correlation between the SWCF and LWCF, most closely approximated by the T42 CAM3. The T85 configuration exhibits a nonlinear correlation between the CRF anomalies, with a much greater range in the magnitude of the CRF anomalies. This unusual behavior of the cloud scheme in the T85 configuration will require additional research at the process level to better understand and improve the relationship between SWCF and LWCF. More importantly, it does indicate that the radiative behavior of clouds is likely to be different at different resolutions due to the changes made to achieve global energy balance. This may explain, in part, why the CAM3 exhibits slightly different climate sensitivity as a function of horizontal resolution (Kiehl et al. 2006).

3. Simulation differences: Uncoupled atmosphere

In many respects, the large-scale simulation properties of the T42 and T85 CAM3 configurations are very similar, exhibiting analogous biases with respect to observational data. Fields like surface temperature generally show changes that would be expected from differences in elevation associated with changes in horizontal resolution. One of the larger and more obvious systematic changes to the simulation is a general warming of the troposphere (see Fig. 2), with a relatively widespread warming of the tropopause at virtually all latitudes. The T85 simulation shows a modest drying of the atmosphere outside of the deep Tropics, most notably over Northern Hemisphere land areas. There are also a collection of other significant simulation differences that are of importance to coupling the atmosphere to other component models. These differences fall into three categories worthy of discussion: differences in radiative forcing, differences in the low-level dynamical circulation, and differences in surface water exchange processes, principally attributable to the precipitation component. Other, more modest differences in quantities like the dynamical circulation of the free atmosphere, including aspects of internal variability, are contained in companion manuscripts (e.g., Hurrell et al. 2006).

As shown in Table 1, the T42 and T85 CAM3 configurations exhibit similar global annual average properties. We note that one area where both configurations differ markedly from observations is cloud amount, which to some extent may be associated with uncertainty in observational estimates. For example, although there is a large difference in low-level cloud amount compared to the International Satellite Cloud Climatology Project (ISCCP), the simulated results are more consistent with the surface-based observations of Warren et al. (1988). Both configurations exhibit very similar energy budget properties on global annual time scales. There is, however, a redistribution of energy in the system at T85, which is easiest to discuss in terms of zonal means. The T85 model shows a reduction in outgoing longwave radiation (OLR) of approximately 3 W m−2 in the deep Tropics, and an increase in OLR on the order of 4 W m−2 poleward of 30°. Similarly, the absorbed solar radiation shows a reduction of ∼8 W m−2 in the deep Tropics and an increase in the extratropics, maximizing in the storm tracks around 8 W m−2. The CRF is enhanced at low latitudes, and decreased at high latitudes, most notably in the storm tracks. The largest differences in the tropical radiation budget are generally confined to the western Pacific and Indian Oceans, where both the longwave and shortwave budgets show large spatially coherent increases in CRF (see Fig. 4). These regions are convectively active, and show significant increases in liquid and ice water loading at higher resolution, consistent with the increases in CRF. Generally speaking, CRF and the associated condensed water loading appear to be biased high when compared to available observational estimates. Despite the systematic biases in the longwave and shortwave radiation budgets over the Pacific and Indian Oceans, the radiative response to the El Niño–Southern Oscillation (ENSO) cycle is generally improved for the higher-resolution configuration. This is especially true for the shortwave response. Figure 5 shows the spatial pattern of the anomaly response of monthly averaged differences in absorbed solar and outgoing longwave radiation between November 1984 (ENSO warm phase) and October 1989 (ENSO cold phase) as seen by ERBE and as simulated by CAM3. Generally, the T85 response is much stronger than the T42 response, and more consistent with the observed response. The most extreme difference is seen in the shortwave response, for which the T42 configuration exhibits an extremely weak response in the western Pacific. The longwave response is also weaker than observed, and the strongest response is positioned well to the east of the observed maximum.

Changes in the simulated cloud field and the CRF introduce relatively large local differences to the net surface energy budget (see Fig. 6). Some of these differences can be characterized as local reflections of changes to the CRF, a consequence of tuning cloud and precipitation processes to achieve global energy balance. Others are nonlocal changes, associated with differences in the low-level dynamical circulation that affect properties like surface latent heat fluxes in the subtropics. In either case, these changes represent significant differences in the forcing of the ocean circulation, something to keep in mind when we discuss the coupled model simulation results.

Several of the differences that can be seen in the radiative forcing of the climate system motivated a strong interest in higher resolution. A persistent simulation deficiency for most global atmospheric climate models is the representation of stratocumulus clouds along the eastern coasts of the Atlantic and Pacific Ocean Basins. Figure 7 shows the change in annually averaged absorbed solar radiation off the coasts of Baja, Peru, and Namibia. Note the significant reductions in absorbed solar radiation immediately along the coast, attributable to a much improved representation of stratocumulus cloud cover. Unlike the changes to the radiation budget in the convectively active deep Tropics, these signals are entirely associated with higher horizontal resolution since they have a very weak dependence on the cloud radiative budget tuning. The east–west dipole structure in the Southern Hemisphere (SH) plots of absorbed shortwave radiation reflects the displacement of stratus clouds toward the coast where they should be located. Previous coupled models exhibited large warm SST biases in these regions, attributed in part to excessive solar insolation at the surface. Local reductions of more than 40 W m−2 represent a significant improvement to the simulation. We will discuss the impact of these radiative changes on the coupled simulation in later sections.

Another area where the T85 configuration exhibits systematic improvements when compared to the T42 configuration is with respect to the low-level dynamical circulation. The differences in the low-level circulation represent noteworthy simulation improvements, although some biases continue to exist. We begin by examining features of the surface wind stress in the eastern ocean stratocumulus regimes. The upper panels in Fig. 8 show the annually averaged surface wind stress off the west coast of Peru as simulated by the T42 and T85 CAM3 configurations, along with the difference between the two. The surface height field, in geopotential meters, is also illustrated using color contours in the T85 and T42 panels. The surface geopotential illustrates the difficulties in representing structures like the Andes Mountains at low resolution. One consequence of the horizontal averaging process is the extension of elevated topography over the ocean surface, as seen in Fig. 8. The bottom panels show surface wind stress as estimated from Earth Remote Sensing (ERS) scatterometer retrievals and the differences with the T42 and T85 CAM3 configurations. As can be seen, there is virtually no equatorward stress on the ocean surface in the vicinity of the coast in the T42 model. This component of the wind stress is responsible for the upwelling of cold ocean water in these regions, and its absence was another simulation deficiency attributed to the generation of warmer than observed sea surface temperatures over this region in coupled applications of the CCSM. The lower-right panel shows the magnitude and structure of the wind stress bias as compared to ERS estimates. A clear improvement in the low-level wind field can be seen in the T85 configuration of the model. The upwelling wind component immediately along the coast continues to be weak, but the overall structure of the T85 low-level wind field represents a major improvement in circulation. This change in the low-level circulation is also a very robust simulation feature that is entirely attributable to changes in horizontal resolution.

Two other low-latitude areas benefit from circulation changes associated with higher horizontal resolution. Figure 9 shows surface wind stress over the Pacific Basin for the T85 and T42 configurations along with ERS scatterometer estimates. Two features that have plagued T42 versions of the CAM include excessively strong trades in the subtropical Pacific, and a very weak westerly wind stress on the equator. The T85 configuration pushes both of these biases toward observational estimates, with significantly enhanced westerly equatorial wind stress in the central Pacific, and a marked decrease in the subtropical Pacific trades. This latter difference is associated with comparably large reductions in surface latent heat flux. One area where the low-level circulation degrades is in the Indian Ocean, which exhibits an enhanced and anomalous cross-equatorial flow.

One of the most reliable simulation signals associated with higher horizontal resolution in the atmosphere is the location of the SH storm track. Figure 10 shows the surface wind stress over the SH for the T42 and T85 configurations of the CAM3, along with their difference. This figure clearly shows the poleward migration of the SH storm track, which is in much better agreement with observational analyses. This response is one of the signals that monotonically improved with higher resolution as shown in Williamson et al. (1995).

The Arctic surface wind simulation changes considerably when the CAM3 resolution is increased to T85. Although there continue to be notable differences with analyses, there are large reductions in circulation biases, where the most prominent difference is a reduction in an anomalous polar summer anticyclone as seen in Fig. 11. This anomalous flow pattern at T42 has been thought to be a major contributor to bases in the simulated Arctic sea ice distribution, which exhibits ice that is too thick off the Siberian coast and too thin along the Canadian coast. We will briefly touch on this when we discuss coupled simulation results.

The final area in which there are conspicuous large-scale simulation differences between the T42 and T85 models involves freshwater exchange with the surface. Although changes to the low-level dynamical circulation introduce desirable improvements in the evaporation of water (e.g., central Pacific subtropics), differences in the net exchange of water are more often dominated by changes in precipitation. Both configurations of the model continue to reproduce the major features of the hydrological cycle, but the T85 configuration exhibits marked redistributions of precipitation at low latitudes, as seen in Fig. 12.

The Pacific intertropical convergence zone (ITCZ) sharpens in the meridional direction at higher resolution, particularly during boreal summer (Fig. 13). There are similar improvements to the representation of the Atlantic ITCZ, historically a very difficult feature to reproduce. The improved definition of the Pacific ITCZ is manifested primarily in the form of reductions in subtropical precipitation, with very modest increases in ITCZ precipitation along the equator. Some of the more important differences are seen in the tropical western Pacific, which exhibits a substantial increase in precipitation rate, maximizing northeast of New Guinea and north of the Solomon Islands. Overall, there is a tendency to move precipitation closer to the equator, and in some cases into the equatorial Northern Hemisphere, as seen in the Indian Ocean. If we use the Climate Prediction Center (CPC) merged analysis of precipitation (CMAP) product as our standard observed precipitation climatology (Xie and Arkin 1997), these changes can be viewed as improvements. Reductions in precipitation rates over Indonesia and the Indian peninsula represent desirable simulation improvements, as are the reductions in precipitation rate over the western Arabian Sea and Gulf of Oman (not shown). The boreal winter simulation shows a significant and realistic enhancement of the South Pacific convergence zone (SPCZ), as well as a reduction of precipitation in the Northern Indian Ocean (not shown). The improvements to the SPCZ are only weakly seen in the annual mean plots shown in Fig. 12, and arise from a weakening of the double-ITCZ-like structure extending across the Pacific just south of the equator, and an enhancement of precipitation in the southern extratropical extension of the SPCZ.

Changes to the precipitation distribution represent significant alterations to the net water exchange between the atmosphere and ocean. The increase in precipitation rate in the tropical western Pacific alters the T42 freshwater budget by approximately 20% and is another area where we might expect to see large simulation differences when these atmospheres are coupled to a fully interactive ocean component model.

4. Simulation differences: Coupled configuration

In this section we provide an overview of the simulation differences as a function of resolution in CCSM3 coupled configurations with an emphasis on the simulation results obtained from the land, ocean, and sea ice components. All simulations employ the CAM3 as the atmospheric component, either at T42 or T85 resolution. The land surface model is discretized on the same horizontal transform grid as the atmosphere: ∼2.8° at T42, and ∼1.4° at T85. The ocean and sea ice models make use of a nominal 1° horizontal finite-difference discretization for all the coupled simulations to be discussed. We will make use of the nomenclature T85x1 to refer to the T85 atmosphere coupled to the 1° ocean model, and T42x1 to refer to the T42 atmosphere coupled to the 1° ocean. The T85x1 configuration of the coupled model has been used to document the CCSM3 simulations for international climate-change assessment purposes (see Collins et al. 2006a).

a. Atmosphere

In the previous section we discussed a broad class of atmospheric simulation differences, including a warming and drying of the simulated atmosphere at high resolution, along with three specific classes of simulation differences that would affect coupled component models: localized radiation budget differences, large-scale differences in the dynamical circulation, and localized changes in the freshwater budget.

The differences in global annual measures of the coupled atmospheric simulation for both resolutions are remarkably similar to the differences documented in Table 1 for the uncoupled configuration. Other large-scale simulation differences also carry over to the coupled framework, including the tendency for a slightly warmer and drier simulation at high resolution. The resolution differences in the vertical distribution of water, in both condensed and vapor phases, are also in qualitative agreement with the uncoupled solutions (see also Hack et al. 2006). Regional radiation biases are generally confined to convectively active regions in the western Pacific and Indian Oceans as in the uncoupled model. Interestingly, the shortwave absorption anomalies in the eastern ocean stratocumulus regimes persist in the coupled framework, but are not as apparent in the coupled net surface energy budget differences. This is due to adjustments in the other components of the surface energy budget in response to differences in SST (e.g., a warmer ocean) and local changes in the low-level dynamical circulation (affecting sensible and latent heat transfers). Generally speaking, low latitude differences in the net surface energy budget as a function of resolution are considerably more complex than shown in Fig. 6. The complexity of the resolution response is largely attributable to differences in meridional shifts in deep convection, and the associated changes to the cloud field and dynamical circulation. These differences will become more apparent when we discuss the coupled freshwater budget differences at the two resolutions.

Improvements in the dynamical circulation due to higher resolution are generally similar to what is seen in the uncoupled framework. Flow along the eastern ocean coastlines is improved, excessively strong trades are reduced, the SH storm track moves poleward in agreement with observations, and seasonal anomalies in the low-level Arctic circulations are reduced. The major exception includes the equatorial Indian Ocean where the low-level surface circulation improves along with the distribution of diabatic heating. A portion of this improvement appears to be associated with the coupled atmosphere–ocean configuration (e.g., see Hack et al. 2006), and further improves at higher resolution.

Although many global annual atmospheric measures of the energy and water cycles are virtually identical to the uncoupled simulations, there are some notable local differences in the coupled and uncoupled configuration. One of the most egregious biases in the coupled framework is associated with the behavior of the simulated hydrological cycle, which exhibits large anomalies in the exchange and storage of water in the atmosphere when compared to the uncoupled CAM3 configuration (Hack et al. 2006). One feature is the shift in the surface exchange of water from the Northern to Southern Hemisphere Tropics, producing a significant and unrealistic change to the freshwater budget over the tropical oceans, most notably during the boreal winter. This shift is largely attributable to differences in the precipitation distribution. Although precipitation anomalies appear in both the Atlantic and Pacific Basins, the zonal mean anomaly is dominated by changes over the Pacific. This takes the form of an unrealistic enhancement of a southern and more vigorous branch of ITCZ convection extending across the Pacific Basin from the warm pool to the Ecuador coast. There are hints of this tendency in the uncoupled model (see Fig. 12), a tendency which is amplified in the coupled configuration. The change to the precipitation distribution is symptomatic of the so-called double-ITCZ problem that plagues many coupled models (e.g., see Davey et al. 2002). Figure 14 shows the coupled model precipitation distribution for the T42x1 and T85x1 configurations along with their difference. The figure shows locally significant meridional redistributions of precipitation as a function of horizontal resolution. This redistribution of diabatic heating is also associated with changes to the low-latitude dynamical circulation, which alters the surface exchange of latent and sensible energy, and is responsible for the complex differences in the net surface energy budget discussed earlier. This figure also serves to illustrate an important and more general observation about the contribution of horizontal resolution to many systematic large-scale biases in the coupled CCSM3. As can be seen, the double-ITCZ problem is qualitatively present at both resolutions where the localized precipitation differences are consistent with the resolution biases seen in the uncoupled model (e.g., poleward extension of the SPCZ at high resolution). This result is typical of many other systematic large-scale coupled simulation biases, such as measures of internal variability (e.g., the Madden–Julian oscillation; MJO), which also show little, if any, sensitivity to changes in horizontal resolution (see Hurrell et al. 2006). This result suggests that the parameterized treatment of processes like moist convection may play the most important role in the introduction of such biases.

b. Ocean

In this section we present a brief summary of the ocean model solutions from the T42x1 and T85x1 configurations, and refer to other papers in this volume for additional detail. The ocean component in both configurations is identical (for model details see Danabasoglu et al. 2005), and is initialized with January-mean climatological potential temperature (θ) and salinity (S) (Levitus et al. 1998; Steele et al. 2001) at a state of rest. For consistency, the following analysis of the mean states is based on the same 30-yr time-mean period (years 571–600) as in the T85x1 analysis of Large and Danabasoglu (2006, hereafter LD06).

Both the T42x1 and T85x1 ocean simulations show modest, linear cooling trends after the first 50 years of integration. We compute these average trends as −0.19 and −0.24 W m−2 heat losses at the surface between years 400 and 600 in T42x1 and T85x1, respectively. In contrast, the global-mean S is well preserved in both the T85x1 and T42x1 integrations.

The time- and horizontal-mean global difference profiles from observations (Levitus et al. 1998; Steele et al. 2001) for θ and S are plotted in Fig. 15. The T42x1 profile is warmer between 250 and 2250 m, and colder elsewhere when compared to the T85x1 profile. The depth ranges over which this warming and cooling occurs are very similar in all major ocean basins except the deep Pacific where T42x1 is uniformly warmer by about 0.1°C below 2000 m, in better agreement with observations. Although the global-mean S is essentially the same at both resolutions, the S profiles (Fig. 15) exhibit different vertical distributions. The T85x1 profile is fresher than the T42x1 profile in the upper 1500 m and saltier below this depth, where all major ocean basin profiles show the same systematic behavior. The T42x1 profile generally agrees better with observations. This is particularly true below 2250 m, where the T85x1 simulation bias is reduced by one-half in T42x1. We note that the different vertical distributions of temperature and salinity in the two configurations are density compensating.

Figure 16 shows the time-mean, vertically integrated mass transport (barotropic) streamfunction distribution from T42x1 and its difference from the T85x1 solution. A detailed discussion of the T85x1 circulation, including comparisons with observations, is presented in LD06. In general, all gyre circulation patterns and magnitudes are similar in the two configurations suggesting that the wind stress curls are similar at both atmospheric resolutions. Localized differences are at most order 5–10 Sv (1 Sv ≡ 106 m3 s−1) in the northern Gulf Stream, southern Agulhas, and equatorial Pacific gyre. The exception to this is the Southern Ocean where the Antarctic Circumpolar Current (ACC) is driven primarily by the zonal wind stress. In T85x1 the latitude of the maximum zonal-mean westerlies is in good agreement with satellite scatterometer data (Chin et al. 1998), and represents an improvement over T42x1. The simulated winds, however, are too strong in the latitude band of the ACC (Yeager et al. 2006, hereafter Y06). The erroneous equatorward displacement of the maximum zonal wind at T42x1, coupled with a modest reduction in magnitude, produces an equatorward shift in the ACC with a mean transport at the Drake Passage of 177 Sv. This represents a reduction of 15 Sv with respect to mean transport in T85x1, but still exceeds the observational estimate of 134 ± 13 Sv [Whitworth (1983) as corrected by Whitworth and Peterson (1985)]. Generally, the T85x1 configuration exhibits systematically larger transport along the ACC, which locally exceeds 35 Sv compared to T42x1. Although the magnitude of the ACC transport is too strong at high resolution, the interannual variability of the ACC transport (∼15 Sv) is comparable in both configurations of the model.

The SST error patterns and magnitudes are very similar in the T42x1 and T85x1 configurations (see LD06). This is primarily due to the same circulation errors in each of the coupled configurations, leaving the associated SST biases virtually unchanged. This is also true in the Southern Ocean where the equatorward shift of the ACC in T42x1 does not significantly change the SST errors. Important exceptions are the large positive SST biases off the west coasts of South America, South Africa, and Baja California. These biases are modestly reduced at high resolution, where the spatially averaged differences for a 15°-wide strip immediately off these coasts (see Y06) show reductions of approximately 0.8°C off of South America and South Africa in the T85x1 simulation, but only 0.1°C off Baja. As discussed earlier, regional changes in the surface winds and absorbed solar radiation between the two atmospheric resolutions are the likely contributors to the reductions in the SST biases. However, the principal source of the remaining biases remains an unresolved coupled problem, because subsurface ocean temperatures and upwelling patterns also exhibit differences in the T42x1 and T85x1 ocean solutions.

In general, simulation differences in the sea surface salinity (SSS) in the tropical regions primarily reflect the changes in the precipitation fields. For example, the equatorial and southern tropical Atlantic exhibit excessive precipitation rates in T42x1, resulting in reduced SSS when compared to observations. Although precipitation rates in T85x1 are improved in these regions, the reduction in precipitation leads to only a small improvement in the fresh bias by about 0.5 psu when compared to observations (see Fig. 1 of LD06).

The average Eulerian-mean meridional overturning circulation (MOC) is very similar in the two configurations, as are the northward transports of heat and freshwater. The T85x1 MOC shows some modest improvements compared to the T42x1 MOC, exhibiting a deeper penetration of the North Atlantic Deep Water cell. It also shows better agreement with some observational estimates of southward flows at high latitude in certain density classes as detailed in Bryan et al. (2006). In contrast with the time-mean MOC, the amplitudes of decadal time-scale variability differ substantially. Decadal time scale variability for the T85x1 configuration is about a factor of 2 larger than in the lower resolution configuration. Further details of the MOC comparisons, particularly for the Atlantic Ocean, are given in Bryan et al. (2006).

Finally, in the eastern equatorial Pacific, both solutions have a strong semiannual signal in the seasonal cycle of SST anomalies. The characteristics of ENSO variability are also very similar, as documented by Y06. The T85x1 Equatorial Undercurrent maximum is stronger than in T42x1 and closer to the observations of Johnson et al. (2002). On the other hand, the asymmetric structure of the South Equatorial Current (with a stronger southern branch) is degraded in the T85x1 when compared to the T42x1 simulation (see Figs. 10 and 11 of LD06).

Overall, despite large differences in the boundary forcing of the ocean by the high-resolution atmosphere, the T85x1 simulation is quite similar to the T42x1 simulation. Each configuration enjoys specific strengths and weaknesses, although the simulation differences suggest that neither configuration is systematically better when compared to observational estimates of the global ocean circulation.

c. Land

From the perspective of the Community Land Model (CLM), there are three aspects of the simulation that change with increased resolution. First, because the land model currently runs on the same grid as that of the host atmospheric model, a finer-scale representation of the land surface is required. Consequently, the underlying land surface can change in terms of the distribution of land cover and soil types. Second, the land surface is forced by and responds to changes in the near-surface atmosphere (e.g., precipitation, temperature, specific humidity) attributable to changes in atmospheric circulation. Third, there may be feedback mechanisms between the land surface and the atmosphere that may amplify or dampen the response of the land surface to the changes in near-surface forcing. There are no scaling or tuning modifications made within the land model itself to accommodate higher resolution.

Attributing changes in land surface climatology to any of these mechanisms separately is difficult and outside the scope of this paper. However, the contribution of the finer-scale land surface representation is likely to be small compared to other factors. Globally, the land cover types change by at most 0.2%. Regionally, the largest differences occur where there is extensive coastline or where land cover is fragmented. Generally, it takes more substantive changes in land cover to affect CLM regional surface climate (e.g., Oleson et al. 2004). Similarly, the largest change in soil texture (%sand or %clay) is 7%. For the larger regions we examine below, these relative differences are even smaller; a maximum of 3% for land cover change and 2% for soil texture. Therefore, in this section we simply address the issue of whether increased horizontal resolution significantly affects land surface climatology in response to changes in atmospheric radiative and hydrological forcing and subsequent feedbacks. We divide the land surface into continental-sized regions, using subdivisions with known biases in the T42x1 simulation, to assess whether higher spatial resolution has improved or degraded the simulation (Table 2).

Global land seasonal averages of precipitation, evapotranspiration, runoff, air temperature, net radiation, and the Bowen ratio are fairly similar at T85x1 and T42x1 resolutions (Table 2). In general, changes in the air temperature correspond to changes in radiative forcing and/or changes in the partitioning of net radiation into sensible and latent heat as reflected by the Bowen ratio. There is about a 0.3°C decrease (increase) in T85x1 global surface temperature in boreal winter (summer). These changes compare favorably to observations, and correspond to decreases (increases) in net radiation.

In boreal winter, higher resolution results in cooler temperatures at low and middle latitudes of North America (0.9° and 0.2°C cooling) and Europe (0.8°C; see Fig. 17). A decrease in downward longwave radiation (not shown) and hence net radiation is likely to be partially responsible for this cooling. At low and middle latitudes over North America longwave radiation decreases by 7 and 10 W m−2, and in Europe by 9 W m−2. The 4 W m−2 decrease in global net radiation reflects a 6 W m−2 decrease in downward longwave radiation. Global absorbed solar radiation is virtually unchanged. However, there are changes in downward solar radiation and albedo in some of these regions that compensate somewhat for the decrease in longwave radiation and modulate the cooling. For example, at middle latitudes over North America, an increase in downward solar radiation along with a decrease in albedo increases absorbed solar radiation, which compensates somewhat for the decrease in longwave radiation. Similarly, changes in the partitioning of net radiation into sensible and latent heat also interact with changes in radiative forcing. At low latitudes over North America, a wet bias in precipitation increases with higher resolution resulting in a shift in the Bowen ratio toward larger latent heat fluxes that contribute to cooling. The cooling in Europe and at middle latitudes over North America reduces the T42x1 warm bias in these regions. However, a cold bias at low latitudes in Asia is enhanced (not shown) and a cold bias is introduced at low latitudes over North America in T85x1.

Warming in the high latitudes of North America and Asia offsets some of the boreal winter cooling described above. The 3°C warm bias in the T42x1 simulation in these regions increases by about 1.5°C at higher resolution. This does not appear to have much to do with the atmospheric radiative forcing since absorbed solar and downward longwave radiation change by less than 2 W m−2. Changes in atmospheric circulation appear to be responsible for the warming in these regions (Dickinson et al. 2006).

In boreal summer, the increase in global temperature appears to be primarily driven by the response of the land surface to increases in net radiation caused by increases in absorbed solar radiation (not shown). In particular, there are increases in absorbed solar radiation of 22 W m−2 at midlatitudes in North America (1.4°C warming), 12 W m−2 in Europe (1.1°C), and 10 and 16 W m−2 at mid- and high latitudes in Asia (1.5° and 1.6°C). Changes in albedo are minimal at these scales (<0.01), with the exception of high latitudes in Asia, where higher incoming solar radiation is primarily responsible for the increase in absorbed solar radiation. Snow cover in high-latitude Asia is lower in the T85x1 model, which results in lower albedo and contributes to increased absorbed solar radiation. The warming in boreal summer reduces cold biases in these regions except over Europe where a warm bias is introduced at higher resolution.

Over North Africa, land air temperature is somewhat cooler in the T85x1 simulation in both seasons (Fig. 17). In summer, this appears to be due in part to an increase in precipitation and evapotranspiration and a decrease in sensible heat that lowers the Bowen ratio. Radiative forcing of the surface is also lower due to less incoming solar radiation. These changes contribute to a year-round cold bias in this region.

The global average hydrologic cycle is not strongly affected by changes in horizontal resolution. Precipitation and runoff are biased high and low in boreal winter and summer, respectively, in both simulations. However, there are noteworthy changes with resolution at smaller scales. There is an overactive hydrological cycle at northern high latitudes year-round in the T42x1 simulation. This appears to be slightly enhanced at higher resolution. In particular, winter precipitation and snow depth at high latitudes in North America are overestimated. Consequently, the snowmelt season is delayed and snow persists into early summer, which likely contributes to the cold summer bias. This problem is less severe in high-latitude Asia because of smaller biases in precipitation. However, as noted previously, the cold bias in summer in this region appears to be improved at higher resolution because there are fairly large increases in incoming solar radiation in spring and summer that compensate for the increase in winter snow depth. Snow appears to melt back at about the same rate as the T42x1 simulation.

Summer runoff at high latitudes in North America increases by 50% at higher resolution and is about double what the observations suggest. Much of this is due to a much stronger runoff peak in June caused by melting of the deeper snowpack. In high-latitude Asia, despite a small increase in winter precipitation and snow depth, the runoff in summer is actually lower in the T85x1 simulation. This is because warmer temperatures and increased solar radiation combine to melt snow sooner and create a runoff peak that occurs in May in the T85x1 simulation as compared to June in the T42x1 simulation. However, the runoff peak in May is still substantially higher than in the observations.

Other regions that have significant biases in hydrology in the T42x1 simulation show improvement at higher resolution. Summer precipitation in Europe and North Africa has improved slightly. In the Amazonia region, higher resolution increases dry season precipitation by 67%, resulting in favorable increases in evapotranspiration and runoff and cooler temperatures. However, improvements in precipitation do not necessarily translate to improvements in the simulation of runoff in other regions. For example, summer precipitation over North Africa agrees quite well with observations but runoff is biased high at higher resolution, suggesting that improvements to the treatment of runoff processes are likely required.

d. Sea ice

A detailed analysis of the simulation of Arctic sea ice as a function of horizontal resolution is addressed in DeWeaver and Bitz (2006). Therefore, in this section we will only summarize the major findings. Historically, simulated Arctic sea ice in the CCSM model has been too thick off the Siberian coast and too thin along the Canadian coast using a T42 atmosphere. Both of these biases are significantly reduced by moving to the higher-resolution T85 atmosphere. The improvement in ice distribution is associated with the improvement in the surface wind distribution discussed in the context of the uncoupled simulation, most notably an erroneous north polar summer anticyclone, which is a feature of the T42 configuration, but absent at T85. DeWeaver and Bitz (2006) employ an offline sea ice model to explore the reasons for the simulation improvement in great detail. Although improvements in the surface wind forcing of the ice plays a major role in the ice distribution improvement, their results also suggest that differences in thermodynamic forcing of the ice are an important contributor. We refer the interested reader to DeWeaver and Bitz (2006) for more detail.

5. Summary

Like many other major climate modeling activities, the CCSM global atmospheric model has employed the same moderately low-resolution truncation for more than a decade. The development and release of the CAM3 provides the opportunity to comprehensively explore the benefits of exploiting higher horizontal resolutions in a framework that provides comparable large-scale simulation fidelity when compared with simulations using more traditional horizontal truncations.

We have presented selected features of the CAM3 simulation as a function of two horizontal spectral truncations, T42 and T85, for both coupled and uncoupled configurations. The model formulations differ only in the rate at which energy is dissipated by horizontal diffusion processes, and in the choices of a limited number of coefficients associated with parameterized convection and cloud processes. The uncoupled high-resolution model exhibits a number of systematic simulation improvements, including a desirable systematic warming of the troposphere, large-scale improvements in the low-level dynamical circulation, and local improvements to the representation of clouds and precipitation regimes, such as the improved structure of the Atlantic ITCZ. Despite anomalies in the near cancellation of longwave and shortwave cloud forcing in the Tropics, the T85 model demonstrates a superior radiative response to the ENSO cycle, particularly in the shortwave component of the radiation budget. In the context of atmospheric teleconnections, the T85 CAM3 provides an extremely realistic Pacific–North American (PNA) response, where the amplitude of the T42 response is only about half as strong (Deser et al. 2006). Using warm − cold composites, Deser et al. (2006) also show that although there is no clear benefit of high resolution to temperature teleconnections, the T85 model is demonstrably better than the T42 model with regard to precipitation teleconnections.

Most of the resolution-dependent simulation improvements seen in the uncoupled CAM3 carry over to the coupled model framework. From a sea ice perspective, the higher-resolution atmosphere has a significant positive impact on the simulation of Arctic ice, as shown by DeWeaver and Bitz (2006). From a land perspective, the higher-resolution atmosphere produces an improved simulation of air temperature and hydrology in some regions. Most notably, there is a 0.8°C cooling in boreal winter over Europe, which reduces a warm bias in the T42 simulation. The T85 atmosphere also results in warming during boreal summer at mid- and high latitudes over North America and Asia, reducing a cold T42 bias. There are, however, some aspects of the simulation that degrade with the high-resolution atmosphere. The boreal winter warm bias at high latitudes in the T42 simulation is worse at higher resolution. Similarly, the overactive hydrological cycle at northern high latitudes year-round appears to be enhanced, particularly over North America.

Despite a number of significant localized changes in the radiative, dynamical, and freshwater forcing of the ocean with the higher-resolution atmosphere, there is virtually no resolution response seen in the ocean component simulation. This is perhaps most remarkable in the eastern ocean stratus regions where the radiative forcing of the ocean surface is reduced by more than 40 W m−2 and the coastal surface wind stress has been significantly improved with regard to upwelling. These large surface forcing improvements produce relatively minor responses to the warm sea surface temperature anomalies simulated by the ocean component. The structure and amplitude of other large-scale simulation anomalies like the Pacific double ITCZ are also largely independent of horizontal resolution. The overall ocean response to the high-resolution atmosphere leads to the conclusion that the ocean model solutions from the high- and low-resolution configurations are not significantly different. Similarly, Deser et al. (2006) find that the structure, period, and amplitude of ENSO are relatively insensitive to the resolution of the atmospheric component model.

Overall, the high-resolution version of the CAM3 exhibits a mixed message, especially with regard to the coupled modeling framework. Although the higher-resolution atmosphere has a limited impact on improving the quality of some component model solutions, there is little question that the overall quality of the atmospheric simulation is improved. Perhaps one of the most important observations about the T85 configuration is contained in Fig. 1. It is clear that the properties of the resolved-scale motions seen by the parameterized physics are much more realistic at T85 than at T42. As such, the T85 truncation would appear to be a more appropriate resolution at which to test parameterization techniques developed using observational data. It remains to be seen why the pointwise-resolved scale motions appear to be so much more energetic at high resolution, but that modes of internal variability like the MJO exhibit no improvement with increased horizontal resolution (see Hurrell et al. 2006). This is but one of many questions that will be answered as the differences in simulation quality associated with changes in horizontal resolution are explored in greater detail.

Acknowledgments

We would like to acknowledge the substantial contributions to the CCSM project from the National Science Foundation (NSF), Department of Energy (DOE), the National Oceanic and Atmospheric Administration, and the National Aeronautics and Space Administration. In particular, Hack, Truesdale, and Caron wish to acknowledge support from the DOE Office of Science, NCAR Water Cycle Across Scales Initiative, and the Central Research Institute for the Electric Power Industry (CRIEPI). This study is based on model integrations performed by NCAR and CRIEPI with support and facilities provided by NSF, DOE, MEXT, and ESC/JAMSTEC.

REFERENCES

  • Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor, 4 , 11471167.

    • Search Google Scholar
    • Export Citation
  • Boer, G. J., and M. Lazare, 1988: Some results concerning the effect of horizontal resolution and gravity wave drag on simulated climate. J. Climate, 1 , 789806.

    • Search Google Scholar
    • Export Citation
  • Boville, B. A., 1991: Sensitivity of simulated climate to model resolution. J. Climate, 4 , 469485.

  • Bryan, F. O., G. Danabasoglu, N. Nakashiki, Y. Yoshida, D-H. Kim, J. Tsutsui, and S. C. Doney, 2006: Response of North Atlantic thermohaline circulation and ventilation to increasing carbon dioxide in CCSM3. J. Climate, 19 , 23822397.

    • Search Google Scholar
    • Export Citation
  • Chen, T-C., and J. J. Tribbia, 1993: An effect of the model's horizontal resolution on stationary eddies simulated by the NCAR CCM1. J. Climate, 6 , 16571664.

    • Search Google Scholar
    • Export Citation
  • Chin, T. M., R. F. Milliff, and W. G. Large, 1998: Basin-scale, high-wavenumber sea surface wind fields from a multiresolution analysis of scatterometer data. J. Atmos. Oceanic Technol, 15 , 741763.

    • 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
  • Danabasoglu, G., W. G. Large, J. J. Tribbia, P. R. Gent, B. P. Briegleb, and J. C. McWilliams, 2006: Diurnal coupling in the tropical oceans of CCSM3. J. Climate, 19 , 23472365.

    • Search Google Scholar
    • Export Citation
  • Davey, M. K., and Coauthors, 2002: STOIC: A study of coupled model climatology and variability in tropical ocean regions. Climate Dyn, 18 , 403420.

    • Search Google Scholar
    • Export Citation
  • Deser, C., A. Capotondi, R. Saravanan, and A. Phillips, 2006: Tropical Pacific and Atlantic variability in CCSM3. J. Climate, 19 , 24512481.

    • Search Google Scholar
    • Export Citation
  • DeWeaver, E., and C. M. Bitz, 2006: Atmospheric circulation and its effect on Arctic sea ice in CCSM3 simulations at medium and high resolution. J. Climate, 19 , 24152432.

    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., K. W. Oleson, G. Bonan, F. Hoffman, P. Thornton, M. Vertenstein, Z-L. Yang, and X. Zeng, 2006: The Community Land Model and its climate statistics as a component of the Community Climate System Model. J. Climate, 19 , 23022324.

    • Search Google Scholar
    • Export Citation
  • Fekete, B., C. Vorosmarty, and W. Grabs, 2002: High-resolution fields of global runoff combining observed river discharge and simulated water balance. Global Biogeochem. Cycles, 16 .1042, doi:10.1029/1999GB001254.

    • Search Google Scholar
    • Export Citation
  • Hack, J. J., J. M. Caron, S. G. Yeager, K. W. Oleson, M. M. Holland, J. E. Truesdale, and P. J. Rasch, 2006: Simulation of the global hydrological cycle in the CCSM Community Atmosphere Model version 3 (CAM3): Mean features. J. Climate, 19 , 21992221.

    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., J. J. Hack, A. Phillips, J. Caron, and J. Yin, 2006: The dynamical simulation of the Community Atmospheric Model version 3 (CAM3). J. Climate, 19 , 21622183.

    • Search Google Scholar
    • Export Citation
  • Johnson, G. C., B. M. Sloyan, W. S. Kessler, and K. E. McTaggart, 2002: Direct measurements of upper ocean currents and water properties across the tropical Pacific Ocean during the 1990s. Progress in Oceanography, Vol. 52, Pergamon, 31–36.

    • Search Google Scholar
    • Export Citation
  • Kållberg, P., A. Simmons, S. Uppala, and M. Fuentes, 2004: The ERA-40 archive. Tech. Rep. ERA-40 Project Rep. 17, European Centre for Medium-Range Weather Forecasts, Reading, United Kingdom, 35 pp.

  • Kiehl, J. T., 1994: On the observed near cancellation between longwave and shortwave cloud forcing in tropical regions. J. Climate, 7 , 559565.

    • Search Google Scholar
    • Export Citation
  • Kiehl, J. T., and V. Ramanathan, 1990: Comparison of cloud forcing derived from the Earth Radiation Budget Experiment with that simulated by the NCAR Community Climate Model. J. Geophys. Res, 95 , 1167911698.

    • Search Google Scholar
    • Export Citation
  • Kiehl, J. T., and D. L. Williamson, 1991: Dependence of cloud amount on horizontal resolution in the National Center for Atmospheric Research Community Climate Model. J. Geophys. Res, 96 , 1095510980.

    • Search Google Scholar
    • Export Citation
  • Kiehl, J. T., and K. E. Trenberth, 1997: Earth's annual global mean energy budget. Bull. Amer. Meteor. Soc, 78 , 197208.

  • Kiehl, J. T., C. A. Shields, J. J. Hack, and W. D. Collins, 2006: The climate sensitivity of the Community Climate System Model version 3 (CCSM3). J. Climate, 19 , 25842596.

    • Search Google Scholar
    • Export Citation
  • King, M. D., and Coauthors, 2003: Cloud and aerosol properties, precipitable water, and profiles of temperature and water vapor from MODIS. IEEE Trans. Geosci. Remote Sens, 41 , 442458.

    • Search Google Scholar
    • Export Citation
  • Kistler, R., and Coauthors, 2001: The NCEP–NCAR 50-year reanalysis: Monthly means CD-ROM and documentation. Bull. Amer. Meteor. Soc, 82 , 247267.

    • Search Google Scholar
    • Export Citation
  • Large, W. G., and G. Danabasoglu, 2006: Attribution and impacts of upper-ocean biases in CCSM3. J. Climate, 19 , 23252346.

  • Levitus, S., and Coauthors, 1998: Introduction. Vol. 1, World Ocean Database 1998, NOAA Atlas NESDIS 18, 346 pp.

  • Manabe, S. J., J. L. Smagorinsky, J. Holloway, and H. M. Stone, 1970: Simulated climatology of a general circulation model with a hydrologic cycle: III. Effects of increased horizontal computational resolution. Mon. Wea. Rev, 98 , 175212.

    • Search Google Scholar
    • Export Citation
  • Manabe, S. J., D. G. Hahm, and J. L. Holloway Jr., 1979: Climate simulations with GFDL spectral models of the atmosphere: Effect of spectral truncation. JOC Study Conf. on Climate Models: Performance, Intercomparison and Sensitivity Studies, Geneva, Switzerland, WMO, Global Atmospheric Research Programme Publication Series 22, Vol. 1, 41–94.

  • Moncrieff, M. W., S. K. Krueger, D. Gregory, J-L. Redelsperger, and W-K. Tao, 1997: GEWEX cloud system study (GCSS) working group 4: Precipitating convective cloud systems. Bull. Amer. Meteor. Soc, 78 , 831845.

    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and Coauthors, 2004: Effects of land use change on North American cilmate: Impact of surface datasets and model biogeophysics. Climate Dyn, 23 , 117132.

    • Search Google Scholar
    • Export Citation
  • Pope, V. D., and R. A. Stratton, 2002: The processes governing horizontal resolution sensitivity in a climate model. Climate Dyn, 19 , 211236.

    • Search Google Scholar
    • Export Citation
  • Ramanathan, V., and W. Collins, 1991: Thermodynamic regulation of ocean warming by cirrus clouds deduced from observations of the 1987 El Niño. Nature, 351 , 2732.

    • Search Google Scholar
    • Export Citation
  • Randel, D. L., T. H. Vonder Haar, M. A. Ringerud, G. Stephens, T. J. Greenwald, and C. L. Combs, 1996: A new global water vapor dataset. Bull. Amer. Meteor. Soc, 77 , 12331246.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc, 80 , 22612287.

  • Steele, M., R. Morley, and W. Ermold, 2001: PHC: A global ocean hydrography with a high-quality Arctic Ocean. J. Climate, 14 , 20792087.

    • Search Google Scholar
    • Export Citation
  • Warren, S. G., C. J. Hahn, J. London, R. M. Chervin, and M. L. Jenne, 1988: Global distribution of total cloud cover and cloud type amounts over the ocean. Tech. Rep. NCAR Tech. Note TN-317+STR, National Center for Atmospheric Research, 42 pp. + 170 maps.

  • Whitworth III, T., 1983: Monitoring the transport of the Antarctic Circumpolar Current at Drake Passage. J. Phys. Oceanogr, 13 , 20452057.

    • Search Google Scholar
    • Export Citation
  • Whitworth III, T., and R. G. Peterson, 1985: Volume transport of the Antarctic Circumpolar Current from bottom pressure measurements. J. Phys. Oceanogr, 15 , 810816.

    • Search Google Scholar
    • Export Citation
  • Williamson, D. L., 1988: The effect of vertical finite difference approximations on simulations with the NCAR Community Climate Model. J. Climate, 1 , 4058.

    • Search Google Scholar
    • Export Citation
  • Williamson, D. L., J. T. Kiehl, and J. J. Hack, 1995: Climate sensitivity of the NCAR Community Climate Model (CCM2) to horizontal resolution. Climate Dyn, 11 , 377397.

    • Search Google Scholar
    • Export Citation
  • Willmott, C. J., and K. Matsuura, 2000: Terrestrial air temperature and precipitation: Monthly and annual climatologies. [Available online at http://climate.geog.udel.edu/~climate.].

  • Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc, 78 , 25392558.

    • Search Google Scholar
    • Export Citation
  • Yeager, S. G., C. A. Shields, W. G. Large, and J. J. Hack, 2006: The low-resolution CCSM3. J. Climate, 19 , 25452566.

  • Zhang, Y. C., W. B. Rossow, A. A. Lacis, V. Oinas, and M. I. Mishchenko, 2004: Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets: Refinements of the radiative transfer model and the input data. J. Geophys. Res, 109 .D19105, doi:10.1029/2003JD004457.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Large-scale advective temperature tendency in K day−1 for (top) the T42 CAM3, (middle) the T85 CAM3, and (bottom) Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE) dataset (Moncrieff et al. 1997).

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 2.
Fig. 2.

Zonally averaged annual temperature difference (left) between the T85 CAM3 and the T42 CAM3 and (right) between the T42 CAM3 and ERA-40. Contour intervals are 0.8 K.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 3.
Fig. 3.

Scatterplots showing shortwave vs longwave cloud forcing in the tropical west Pacific warm pool region for (left) the T42 CAM3, (middle) the T85 CAM3, and (right) ERBE.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 4.
Fig. 4.

(top) The net longwave flux difference and (bottom) net shortwave flux difference between the T85 CAM3 and the T42 CAM3 in W m −2.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 5.
Fig. 5.

(left) The absorbed solar radiation anomaly for the T42 CAM3, the T85 CAM3, and ERBE for ENSO warm minus cold conditions. (right) The outgoing longwave radiation anomaly for the T42 CAM3, the T85 CAM3, and ERBE for ENSO warm minus cold conditions. Contour intervals are 5 W m −2.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 6.
Fig. 6.

Surface energy residual for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) their difference in W m −2 for the Tropics. The contour interval for the top and middle panels is 25. The contour levels for the bottom panel are (−80.0, −60.0, −40.0, −30.0, −20.0, −10.0, −5.0, −1.0, 0.0, 1.0, 5.0, 10.0, 20.0, 30.0, 40.0, 60.0, 80.0).

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 7.
Fig. 7.

Shortwave cloud forcing difference between the T85 CAM3 and the T42 CAM3 for (left) North America, (middle) South America, and (right) Africa. Contour intervals are 5 W m−2.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 8.
Fig. 8.

(top) The surface wind stress vectors for (left) the T85 CAM3, (middle) the T42 CAM3, and (right) their difference. (bottom) The surface wind stress vectors for (left) ERS, (middle) the difference between T85 CAM3 and ERS, and (right) the difference between T42 CAM3 and ERS. Color contours show corresponding surface geopotential height fields (PHIS) in units of 10−3 m2 s−2.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 9.
Fig. 9.

(left) The surface wind stress magnitude (color contours) and vectors over the Tropics for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) their difference. (right) The surface wind stress for (top) ERS, (middle) the difference between T85 CAM3 and ERS, and (bottom) the difference between T42 CAM3 and ERS.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 10.
Fig. 10.

The surface wind stress magnitude (color contours) and vectors over the Southern Hemisphere for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) the T85–T42 difference. Contour intervals are 0.02 m s−2 for wind magnitudes and 0.01 m s−2 for differences.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 11.
Fig. 11.

Near-surface wind difference (top) between the T85 CAM3 and the T42 CAM3 and (bottom) between the ERA-40 and T42 CAM3 in m s−1.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 12.
Fig. 12.

Annually averaged precipitation over the tropical Indian and Pacific Oceans for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) the T85–T42 difference. Contour intervals are 1 mm day−1 for total precipitation rate and 0.25 mm day−1 for differences.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 13.
Fig. 13.

June–August (JJA) mean precipitation over the tropical oceans for (top) the T85 CAM3, (middle) the T42 CAM3, and (bottom) their difference. Contour intervals are 1 mm day−1 for total precipitation rate and 0.5 mm day−1 for differences.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 14.
Fig. 14.

Annually averaged precipitation over the tropical Indian and Pacific Oceans for (top) the T85 CCSM3, (middle) the T42 CCSM3, and (bottom) the T85–T42 difference. Contour intervals are 1 mm day−1 for total precipitation rate and 0.25 mm day−1 for differences.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 15.
Fig. 15.

Time- and horizontal-mean global differences between models and observations for potential (left) temperature and (right) salinity.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 16.
Fig. 16.

Time-mean, vertically integrated mass transport (barotropic) streamfunction from T42x1 and its difference from the T85x1 solution. (top) Contour intervals are 10 and 20 Sv for transports smaller and greater than 60 Sv, respectively. Also, the thick and thin (shaded regions) lines denote clockwise and counterclockwise circulations, respectively. (bottom) The contour interval is 5 Sv, and the negative differences (thin lines) are shaded.

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Fig. 17.
Fig. 17.

Seasonally averaged 2-m temperature difference between (left) the T85x1 and observations and (right) the T42x1 and observations for (top) December–February (DJF) and (bottom) JJA. Observations are from Willmott and Matsuura (2000).

Citation: Journal of Climate 19, 11; 10.1175/JCLI3764.1

Table 1.

Global annual-mean climatological properties of CAM3 at T85 and T42.

Table 1.
Table 2.

CCSM DJF and JJA land surface climatology. P: precipitation (mm day−1), E: evapotranspiration (mm day−1), R: runoff (mm day−1), T: 2-m air temperature (K), Q: net radiation (W m−2), and B: Bowen ratio. Observations are from Willmott and Matsuura (2000) (air temperature and precipitation) and Fekete et al. (2002) (runoff). Italics represents improvements in the T85 simulation as compared to observations while bold represents deterioration. The asterisk denotes that the T85 simulation is not significantly different from the T42 simulation at the 95% confidence level.

Table 2.
Save