the PMIP models (Bonfils et al. 1998)
PMIP Documentation for CCM1
State University of New York at
Albany: Model NCAR CCM1 (R15 L12) 1992
Dr. John E. Kutzbach, IES-Center for Climatic Research, University of Wisconsin,
1225 W. Dayton St. Madison, Wisconsin 53706-1695;Phone: 608 262 0392; Fax:
608 262 5964 ; e-mail: firstname.lastname@example.org
World Wide Web URL: http://www.ncar.ucar.edu/
NCAR CCM1 (R15 L12) 1992
Model Identification for PMIP
Number of days in each month: 31 28 31 30 31 30 31 31 30 31
The standard version 1 of the second generation NCAR Community Climate
Model (CCM1) was introduced in July of 1987, and included a number of significant
changes to the model formulation which were manifested in changes to the
It is identical to the SUNYA model, without for the addition of radiatively
active trace gases other than carbon dioxide.
Key documents for the standard CCM1 model are :
Williamson, D.L., J.T. Kiehl, V. Ramanathan, R.E. Dickinson, J.J. Hack,
1987: Description of NCAR Community Climate Model (CCM1). NCAR Technical
Note, NCAR/TN-285+STR, 112 pp.
Kiehl et al. (1987) , and Bath et al. (1987a , b ).
Covey, C. and S.L. Thompson (1989). Testing the effects of ocean heat
transport on climate. Paleogeography, Paleoclimatology, Paleoecology, 75,331-341.
Kutzbach,J.,R.Gallimore,S.Harrison,P.Behling,R.Selin, and F.Laarif (1998).
Climate and biome simulations for the past 21,000 years Quaternary Science
Reviews (in press).
Spectral (spherical harmonic basis functions) with transformation to a
Gaussian grid for calculation of nonlinear quantities and some physics.
Spectral rhomboidal 15 (R15), roughly equivalent to 4.5 x 7.5 degrees latitude-longitude.
Surface to 9 hPa; for a surface pressure of 1000 hPa, the lowest atmospheric
level is at 991 hPa.
Finite-difference sigma coordinates.
There are 12 unevenly spaced sigma levels with the following values: 0.009,
0.025, 0.060, 0.110, 0.165, 0.245, 0.355, 0.500. For a surface pressure
of 1000 hPa, 3 levels are below 800 hPa and 5 levels are above 200 hPa.
The PMIP simulation was run on a Cray-YMP computer using a single processor
in a UNICOS environment.
For the PMIP experiment, about 1.2 minutes Cray-YMP computer time per simulated
The control experiment was started from an AMIP simulation and for the
AMIP experiment, initial conditions for the atmospheric state, soil moisture,
and snow cover/depth were specified from the NCAR CCM1 model's standard
January initial dataset (cf. Bath et al. 1987a ). The model then was "spun
up" for 210 days in a perpetual January mode. The resulting climate state
was then taken as the 1 January 1979 starting point for the AMIP simulation.
Time Integration Scheme(s)
Time integration is by a semi-implicit Hoskins and Simmons (1975) scheme
with an Asselin (1972) frequency filter. The time step is 30 minutes for
dynamics and physics, except for full (at all Gaussian grid points and
vertical levels) radiation calculations which are done once every 12 hours
(see Solar Constant/Cycles).
Model spinup was from a previously tuned 0k control mixed-layer run
for Jan 15th. Model integration for 0k and 21k were then run for 25 years
with averages for last 5 years taken for PMIP results.
Orography is smoothed (see Orography). Negative values of atmospheric specific
humidity (which arise because of numerical truncation errors in the discretized
moisture equation) are filled by horizontal borrowing of moisture in a
globally conserving manner. See also Convection.
For the PMIP simulation, the model history is written once every 12 hours.
Primitive-equation dynamics are expressed in terms of vorticity, divergence,
potential temperature, specific humidity, and surface pressure. Energy-conserving
vertical finite-difference approximations are utilized (cf. Williamson
1983 , 1988 ).
Fourth-order (Ñ4) horizontal
diffusion of vorticity, divergence, temperature, and specific humidity
is computed locally on (approximately) constant pressure surfaces in grid-point
space, except at stratospheric levels, where second-order (Ñ2)
horizontal diffusion is applied (cf. Boville 1984 ).
Stability-dependent vertical diffusion is computed locally in grid-point
space at all levels. Cf. Williamson et al. (1987) for further details.
Gravity-wave drag is not modeled.
The solar constant is the AMIP-prescribed value of 1365 W/(m2).
The orbital parameters and seasonal insolation distribution are calculated
after PMIP recommendations. A seasonal, but not a diurnal cycle, in solar
forcing is simulated.
The carbon dioxide values for 0k and 21k are 330 ppm and 191 ppm respectively.
Atmospheric CO2 was set at the pre-industrial value of 267ppm (based on
a lowering from 330ppm Note the change in CO2 radiative forcing is consistent
with a decrease in CO2 from 345ppm to 280ppm).
The vertical distribution of zonal-mean mixing ratios of ozone is specified
from monthly data of Dütsch (1978) , updated by linear interpolation
every 12 hours. The radiative effects of water vapor and oxygen, as well
as methane, nitrous oxide, and chlorofluorocarbon compounds CFC-11 and
CFC-12 also are included, but not those associated with aerosols (see Radiation).
Shortwave radiation is treated in two spectral intervals--ultraviolet/visible
(0.0 to 0.9 micron) and near-infrared (0.9 to 4.0 microns). Shortwave absorption
by ozone, water vapor, carbon dioxide, and oxygen is modeled. Direct-beam
absorption by water vapor is after the method of Kratz and Cess (1985)
; the reflected-beam absorption (as well as Rayleigh scattering by gases)
follows Lacis and Hansen (1974) . Oxygen absorption is treated as in Kiehl
and Yamanouchi (1985) , and near-infrared absorption by carbon dioxide
is after Sasamori et al. (1972) . Gaseous absorption within clouds is included.
Cloud albedo depends on optical depth and solar zenith angle, with multiple
scattering effects included.
Longwave radiation is calculated in 5 spectral intervals (with wavenumber
boundaries at 0.0, 5.0 x 104, 8.0 x 104, 1.0x105,
1.2 x 105, and 2.2 x 105 m-1). Absorption/emission
by water vapor (cf. Ramanathan and Downey 1986 ), carbon dioxide (cf. Kiehl
and Briegleb 1991 ), and ozone (cf. Ramanathan and Dickinson 1979 ) is
treated; the standard CCM1 radiation code is modified to include absorption/emission
by methane, nitrous oxide, and chlorofluorocarbon compounds CFC-11 and
CFC-12 (cf. Wang et al. 1991a , b ). The emissivity of nonconvective cloud
is a function of diagnostic liquid water content. For purposes of the radiation
calculations, cloud is treated as randomly overlapped in the vertical.
Cf. Kiehl et al. (1987) and Wang et al. (1991a , b ) for further details.
See also Cloud Formation.
Moist convective adjustment after the method of Manabe et al. (1965) performs
several functions: removal of negative atmospheric moisture values (operating
with a scheme for horizontal borrowing of moisture--see Smoothing/Filling);
dry convective adjustment of unsaturated, unstable layers in the model
stratosphere, with vertical mixing of moisture; and moist static adjustment
of saturated unstable layers and of supersaturated stable layers.
Cloud forms in layers where the relative humidity exceeds 100 percent.
If the vertical lapse rate of the layer also exceeds the moist adiabatic
value, convective cloud forms (see Convection); otherwise, the cloud is
nonconvective, and the fractional cloud cover is set to 0.95 in the layer.
Convective cloud cover depends on the depth of the vertical instability,
with the cloud amount in each layer adjusted so that the total fractional
area is at most 0.30. If there is no associated precipitation (see Precipitation),
a minimum convective cloud fraction of 0.01 is specified in each layer.
Cloud is not allowed to form in the lowest model layer or in the top 3
layers, but clouds form together in the second and third layers above the
surface if either of these layers is supersaturated. Cf. Kiehl et al. 1987
for further details. See also Radiation for treatment of cloud-radiative
Precipitation results from application of convective adjustment (see Convection),
if the vertical column is supersaturated with a lapse rate exceeding moist
adiabatic. Precipitation also results if the column is supersaturated but
with a stable lapse rate. There is no subsequent evaporation of precipitation
before it falls to the surface.
The modern run contained a 'bug' that led to an inadvertent failure
to correctly update the saturation specific humidity. The bug runs produced
a shift of rainfall from convective to large scale and subsequent decrease
in total cloud cover and a warmer climate than no-bug runs. The sensitivity
of climate (21k minus modern) is, however, not significantly affected by
Planetary Boundary Layer
The height of the PBL top is assumed to be that of the first level above
the surface (sigma = 0.991), except for the calculation of a bulk Richardson
number (see Surface Fluxes). In that case, the PBL top is computed from
the temperature at the first sigma level but is constrained to be at least
For control, after interpolation of 1 x 1-degree Scripps Institution surface
height data (cf. Gates and Nelson 1975 ) to the model grid, the data are
smoothed using a Gaussian filter with 1.5-degree radius. The resulting
heights are transformed into spectral space and truncated at the R15 model
resolution. Cf. Pitcher et al. (1983) for further details.
21k Glacial data on coverage and height are taken from Peltier's 1x1
data sealevel was lowered by 106m to account for ice volume of glaciers;
Peltier's data was used to determine switches of ocean to land for 21k
that resulted from sea level lowering.
Both the 0k and 21k CCM1 PMIP runs were integrated using the Covey-Thompson
50m mixed-layer ocean model. The formulation included a fixed annual mean,
zonal mean ocean heat transport in the SST prediction equation to account
for poleward heat transport by ocean currents. The actual value used based
on tuning (see Covey-Thompson) is about 1/2 the observed transport. Poleward
of 60N and S the convergence of heat flux under sea ice is set at 2W/m2.
We followed a scheme suggested by Broccoli for PMIP to adjust the ocean
heat flux used in the 21k expt (which has reduced ocean area) so that the
total convergence of heat transport is unchanged from the control case.
Sea ice is predicted when the surface temperature falls below 271.2K.
The seaice thickness is predicted using a three-layer thermodynamic model
following Semtner (see Covey-Thompson for details).
Precipitation falls as snow if the temperatures of the surface and the
first two atmospheric levels above it are all < 0 degrees C. Snow cover
is determined from a combination of a monthly latitude-dependent climatology
(cf. Bath et al. 1987a ) and prognostic snow accumulation (on land only)
that is determined from a budget equation. A surface temperature > 0 degrees
C triggers snowmelt, which augments soil moisture (see Land Surface Processes).
Snow cover is also depleted by sublimation, which is calculated as part
of the surface evaporative flux (see Surface Fluxes).
The surface roughness length is specified as a uniform 0.25 m over land,
sea ice, and snow cover, and as 1.0 x 10-3 m over ocean.
Surface albedos for land surfaces are derived from the Matthews (1983)
1 x 1-degree soil/vegetation dataset, but with distinguished vegetation
types reduced to 10 and aggregated to the model resolution (see Horizontal
Resolution). Land albedo also depends on solar zenith angle and spectral
interval (ultraviolet/visible vs near-infrared--see Radiation). Snow cover
alters the land albedo; the composite value is determined from equally
weighted combinations of the local background albedo and that of the snow
(which depends on surface temperature for the diffuse beam and on solar
zenith angle for the direct beam). Over the ocean, surface albedos are
prescribed to be 0.0244 for the direct-beam (with sun overhead) and 0.06
for the diffuse-beam component of radiation; the direct-beam albedo varies
with solar zenith angle. The albedo of ice is a function of surface temperature.
Cf. Briegleb et al. (1986) for further details.
Longwave emissivities are set to unity (blackbody emission) for all
Surface solar absorption is determined from the albedos, and longwave emission
from the Planck equation with prescribed surface emissivity of 1.0 (see
Surface fluxes of momentum, sensible heat and moisture are determined
from bulk aerodynamic formulae, following the formulation of Deardorff
(1972) . Surface drag/exchange coefficients are a function of roughness
lengths (see Surface Characteristics) and bulk Richardson number (see Planetary
Boundary Layer). For computing these fluxes, the surface wind speed is
constrained to be at least 1 m/s.
The surface moisture flux also depends on the evapotranspiration efficiency
beta, which is unity over ocean, snow, and sea ice, but which over land
is a function of soil moisture (see Land Surface Processes).
Land Surface Processes
Land surface temperature is determined from the balance of surface energy
fluxes (see Surface Fluxes) by the diagnostic method of Holloway and Manabe
(1971) . (That is, there is no heat diffusion/ storage within the soil.)
Soil moisture is represented by the single-layer "bucket" model of Budyko
(1956) and Manabe (1969) , with field capacity a uniform 0.15 m of water.
Soil moisture is increased by both precipitation and snowmelt. It is decreased
by surface evaporation, which is determined from the product of the evapotranspiration
efficiency beta and the potential evaporation from a surface saturated
at the local surface temperature/pressure (see Surface Fluxes). Over land,
beta is given by the ratio of local soil moisture to the field capacity,
with runoff occurring implicitly if this ratio exceeds unity.
Last update November 9, 1998. For further information, contact: Céline