Vertical Discretization in the NCAR Community Atmospheric Model CAM 2'0

1
(No Transcript)
2
Method of Vertical Discretization in the NCAR
Community Atmospheric Model (CAM 2.0)

• AT 745
• September 25, 2002
• Jonathan Vigh

3
What is the CAM 2.0?

• The atmospheric component of the Community
  Climate System Model (CCSM), a comprehensive
  climate modeling system developed and maintained
  by NCAR and the university community
• Previously called the Community Climate Model
  (CCM)

4
Brief Overview CCM0A (c. 1982)

• CCM0A was based on the Australian spectral model
  developed by Bourke, McAvaney, Puri, and Thurling
• Was modified to adopt more efficient Fourier
  transform routines and the improved
  radiation/cloudiness parameterizations of
  Ramanathan and Dickinson
• Focus was climate simulations

5
Brief Overview CCM0B (c. 1983)

• In 1981, decision was made that one model should
  serve both climate research and forecasting
  research
• CCM0A was scrapped in favor of a new version
  based on the adiabatic, inviscid version of the
  spectral model developed at ECMWF
• Vertical and temporal finite differences matched
  the Australian spectral model
• New model included physical parameterizations
  for
• Radiation and cloud routines
• Convective adjustment
• Stable condensation
• Vertical diffusion
• Surface fluxes
• Surface-energy-balance prescription

6
Brief Overview CCM1 (Jul 1987)

• Substantial changes included
• Radiation scheme modified
• Vertical finite-difference approximations were
  modified to conserve energy without adversely
  affecting the model simulations
• Frictional heating included
• Horizontal diffusion modified to hyperdiffusion
  form in troposphere
• Local moisture adjustment was generalized for a
  global horizontal borrowing in a conserving
  manner
• Vertical diffusion was converted to a nonlinear
  form for which eddy-mixing coefficient depended
  on local shear and stability (applied throughout
  whole atmosphere eliminated need for dry
  convective adjustment in troposphere)
• Surface drag coefficient made a function of
  stability (Deardorff, 1972)
• Equation of state modified to formally account
  for moisture in the atmosphere (using Tv when
  appropriate)
• Climate mode had optional seasonal mode with
  interactive hydrology
• Method of initialization for global forecast mode

7
Brief Overview CCM2 (Oct 1992)

• Changes in model formulation were so extensive
  that it should not be thought of as being evolved
  from CCM1
• T42L18 (approximately 2.8 x 2.8 deg resolution,
  top at 2.917 mb)
• The following were carried over from CCM1
• Semi-implicit, leap-frog time integration scheme
• Spectral transform method for treating the dry
  dynamics
• Bi-harmonic horizontal diffusion operator

8
CCM2 Changes, contd.

• Shape preserving Semi-Lagrangian Transport (SLT)
  methods used for the advection of water vapor and
  other scalar fields
• Incorporation of a terrain-following hybrid
  vertical coordinate (Simmons and Strüfing, 1981),
  but allows an upper boundary at finite height as
  in Kasahara (1974)
• A d-Eddington approximation used to calculate
  solar absorption
• Cloud radiative parameterization of Slingo (1989)
• Diurnal cycle included
• Land and sea ice surfaces are included with and
  without snow cover
• Subsurface temperatures obey a thermal diffusion
  equation
• Net energy flux at surface/atmosphere interface
  is calculated using bulk exchange formulae in
  which the transfer coefficients depend on
  stability
• Cloud fraction parameterization a generalization
  of Slingo (1987)
• Cloud fraction depends on RH, vertical motion,
  static stability, and convective precipitation
  rate
• Turbulence in PBL mixes heat, moisture, momentum
  and passive scalars
• Simple mass flux scheme by Hack (1993) used to
  represent all types of moist convection
• Biosphere-Atmosphere Transfer Scheme of Dickinson
  et al. (1986) available

9
Brief Overview CCM3 (1996)

• Major changes in parameterized physics, modest
  changes in dynamical formulation
• Modification in physics made to
• Address serious systematic errors and biases in
  CCM2 simulations (in TOA and surface energy
  budgets)
• Allow for more suitable coupling of the model
  atmosphere with land, ocean, and sea-ice
  component models
• Changes fall under five categories
• Modifications to the representation of radiative
  transfer through both clear and cloudy
  atmospheric columns
• Modifications to hydrologic processes (PBL, moist
  convection, and surface energy exchange)
• Incorporation of sophisticated land surface model
• Incorporation of an optional slab mixed-layer
  ocean/thermodynamic sea-ice component
• Other changes in model formalism which dont
  significantly change the model climate

10
Brief overview of CAM 2.0 (2002)

• Incorporated prognostic cloud-water
  parameterization (Rasch and Kristjansson,1998)
• Incorporated scheme to allow a greater variety of
  cloud overlap assumptions (Collins, 2001)
• Updated water vapor absorptivity and emissivity
  with work of Collins, Hackney, and Edward (2002)
• Standard model grid has 26 vertical grid levels,
  with the additional levels added near the
  tropopause
• New ozone dataset (1990 NOAA) and new topography
  with diffusive filter to smooth oceanic ridges
• Sea-ice model replaced with simplified
  thermodynamic version of the CCSM Community
  Sea-Ice Model (CSIM4)
• Community Land Model (CLM2.0) replaces previous
  land model (LSM1)
• Model operates on true land, ocean, and ice
  fractional areas rather than binary
  representation
• Atmospheric model now determines partitioning of
  rain and snow rather than the land-surface model
• Added precipitation evaporation to model
• Orographic drag was removed from ocean
• Code made more modular for testing of various
  physics packages

11
Computational Environment

• Code has been reworked for a distributed
  computing environment (massively parallel
  processing)
• Support for IBM SP3 clusters and Linux PC
• Many runs are done at NCAR on the IBM SP3 running
  at 1.27 TF/s (18th fastest computer in the world)
• BlueSky gets delivered Oct. 2002
• (9 TF/s, 2nd ranked if list stood still)
• Compare to
• NEC Earth Simulator (35.86 TF/s, 1st ranked)
• ECMWF/IBM (1.84 TF/s, 12th ranked)
• NAVOCEANO (1.41 TF/s, 16th ranked)
• NCEP/IBM (1.18 TF/s, 20th and 21st ranked) -gt 5
  TF/s by 2003, 50 TF/s by 2009

12
(No Transcript)