Centre National de Recherches Météorologiques: Model CNRM ARPEGE Cy11 (T42 L30) 1995

Centre National de Recherches Météorologiques: Model CNRM ARPEGE Cy11 (T42 L30) 1995

Model Designation

CNRM ARPEGE Cy11 (T42 L30) 1995

Model Lineage

The model is a version (Cycle 11) of the ARPEGE climate model, which is designed for use by the French climate community. (ARPEGE is the acronym for Action de Recherche Petite Echelle Grande Echelle: Research Project on Small and Large Scales.) ARPEGE is the successor to the EMERAUDE baseline model, and differs substantially in its numerics (especially the horizontal representation), as well as in the formulation of horizontal diffusion, gravity-wave drag, cloud formation, and land surface schemes.

Model Documentation

The ARPEGE Cycle 11 model is described by Deque and Piedelievre (1995)[35], with reference to documentation of an earlier version of the model by Deque et al. (1994)[33]. The model's horizontal representation is discussed by Hortal and Simmons (1991)[36] and by Courtier and Geleyn (1988)[34]. Land surface processes are represented by the scheme of Noihlan and Planton (1989)[37] which was implemented in the ARPEGE model by Mahfouf et al. (1995)[38].

Numerical/Computational Properties

Horizontal Representation

As in the baseline model, the horizontal representation is spectral (spherical harmonic basis functions) with transformation to a Gaussian grid for calculation of nonlinear quantities and some physics. However, in the repeated AMIP integration with the ARPEGE Cy11 model, the Gaussian grid is reduced longitudinally near the poles so that its horizontal resolution is everywhere approximately the same (cf. Hortal and Simmons (1991)[36]). (The ARPEGE Cy11 model is coded so that the spectral basis functions may be mapped conformally from the geographical sphere to a transformed sphere with a different pole, while preserving local horizontal derivatives to within a latitude-dependent mapping factor m. That is, the effective resolution of the model may be varied, and the center of highest resolution may be located at any geographical point--cf. Courtier and Geleyn 1988[34]. However, the option of using these stretched and rotated coordinates was not exercised in the repeated AMIP integration.)

Horizontal Resolution

Spectral triangular 42 (T42), but with transformation to a reduced Gaussian grid such that its resolution is everywhere approximately 300 km (cf. Hortal and Simmons (1991)[36]). The reduced grid resulted in a savings of about 20 percent in computation time.

Computer/Operating System

The repeated AMIP simulation was run on a Cray 2 computer using 4 processors in a UNICOS environment.

Computational Performance

For the repeated AMIP experiment, approximately 12 minutes Cray 2 computation time per simulated day.


For the repeated AMIP simulation, the model was initialized for 1 January 1979 from the climate state obtained after a one-year integration with climatological sea surface temperatures. The initial conditions for this precursor integration were obtained from a December 1988 ECMWF analysis that was interpolated to the model grid.


The horizontal representation of the ARPEGE model necessitates use of a vertical moisture borrowing scheme to correct spurious negative humidities, rather than the horizontal borrowing scheme employed in the baseline model.

Sampling Frequency

For the repeated AMIP simulation, the history of selected variables is written every 6 hours, with a full model history saved at 5-day intervals.

Dynamical/Physical Properties

Atmospheric Dynamics

As in the baseline model, primitive-equation dynamics are expressed in terms of vorticity and divergence, temperature, and specific humidity, and the natural logarithm of surface pressure (or, on option, the surface pressure itself). Also as in the baseline model, ozone is treated as a prognostic variable. In addition, however, the nonconservation of atmospheric mass due to moisture sources/sinks (i.e., evaporation/precipitation) is accounted for in the ARPEGE model's continuity equation; inclusion of this effect results in increased tropical rainfall (cf. Deque and Piedelievre (1995)[35]).


A linear del^6 formulation of horizontal diffusion of vorticity, divergence, temperature, specific humidity, and ozone mixing ratio on constant hybrid sigma-pressure vertical surfaces replaces the del^4 scheme of the baseline model. Below 100 hPa, the diffusivity K increases as (n/N)^6, where n and N are the meridional and truncation wavenumbers, respectively (i.e., n<=N=42 at T42 resolution). Above 100 hPa, K increases as the inverse of the pressure, yielding very strong diffusion in the model stratosphere. In addition, a mesospheric drag is applied to the model winds and temperatures above 1 hPa: the winds are relaxed toward zero and temperatures toward those of a standard atmosphere at these levels. Cf. Deque et al. (1994)[33] for further details.

Gravity-wave Drag

The parameterization of gravity-wave drag is different from that of the baseline model.
  • The gravity-wave stress is assumed to be maximum at the surface, and in a direction that depends on the (2 x 2) covariance matrix of the unresolved oragraphy (a measure of the anisotropy of the mountains) as well as on the mean wind in the planetary boundary layer. The magnitude of the surface stress is proportional to the product of the air density, the Brunt-Vaisalla frequency at the surface, the wind speed at the lowest vertical level, and the root-mean square of the unresolved orography in the direction of this wind (calculated from the subgrid-scale orographic variance).
  • At levels above the surface, the gravity-wave stress is assumed to be in the same direction as the surface stress. The vertically propagating gravity waves do not interact with the mean flow below a critical level of resonance. Above this level, their resonant amplification follows the experimental results of Clark and Peltier (1984)[39], while dissipation proportional to the square of the Froude number also operates. The gravity waves are trapped and reflected with dissipation at the vertical level where the Brunt-Vaisalla frequency becomes zero. Cf. Deque et al. (1994))[33] for further details.

Cloud Formation

Cloud formation is by the same diagnostic method as is used in the baseline model, but a different critical humidity profile (humidities above which cloud forms at different pressure levels) is specified. The 2 empirical coefficients of the critical profile are tuned to accomplish several objectives:

  • to produce generally larger cloud fractions than in the baseline model;
  • to yield a planetary albedo of about 0.30;
  • to result in approximate balance of global annual-mean radiation at the top of the atmosphere.
Cf. Deque and Piedelievre (1995)[35] and Deque et al. (1994)[33] for further details.

Surface Characteristics

Surface characteristics of land are different from those of the baseline model (cf. Mahfouf et al. (1995)[38]).
  • Primary and secondary vegetation cover and type specified in each grid box from 13 classes determined by the Manzi and Planton (1994)[40] simplification of the Wilson and Henderson-Sellers (1985)[41] dataset. Roughness length, leaf area index (LAI), and minimum stomatal resistance required by the land surface scheme are specified according to the vegetation class and soil characteristics of the grid box by blending values associated with the primary and secondary vegetation weighted in a 3 to 1 ratio, respectively. The coverage and roughness length of deciduous and cultivated vegetation also undergo a seasonal cycle. The roughness length includes a contribution from local subgrid-scale orography as well.
  • Soil color, column depth (derived from drainage data of Wilson and Henderson-Sellers (1985)[41]), and texture (fraction of sand and clay required by the land land surface scheme obtained from Webb et al. (1991)[42] data) also are specified for each grid box. The depth of the active soil layer required by the land surface scheme is assigned according to the larger of the vegetation root depth vs the bare-soil depth.
  • As in the baseline model, surface albedos are a function of solar zenith angle, but their values are assigned differently. Albedos over land are prescribed according to vegetation cover (blended as described above for primary and secondary vegetation types) and soil color. The chosen albedo values are validated against Earth Radiation Budget Experiment (ERBE) clear-sky data and METEOSAT data provided by Arino et al. (1991)[43]. Surface longwave emissivities are prescribed in the same manner as in the baseline model.

Surface Fluxes

  • The model follows the Louis et al. (1981)[17] formulation of turbulent surface fluxes, as in the baseline model, but with roughness lengths for calculation of the momentum flux over land set 10 times larger than those for the heat flux.
  • Over land, the surface moisture flux is made up of evaporation from bare ground and from moisture intercepted by the vegetation canopy, as well as from transpiration by the foliage according to formulations of the land surface scheme.

Land Surface Processes

Land surface processes are simulated by the Interactions between Soil-Biosphere-Atmosphere (ISBA) scheme of Noilhan and Planton (1989)[37] as implemented in the ARPEGE model by Mahfouf et al. (1995)[38]. (Use of ISBA results in less extreme ground temperatures over the summer continents than in the baseline model.)
  • The ISBA scheme includes 5 prognostic variables: surface temperature, mean surface temperature, surface volumetric water content, mean volumetric water content, and the water amount intercepted by the vegetation canopy. The time dependence of the prognostic variables are formulated as force-restore equations after Deardorff (1978)[13].
  • ISBA also requires 7 parameters that are prescribed or derived from other surface characteristics: the vegetation cover, leaf area index (LAI), minimum stomatal resistance, surface shortwave albedo, longwave emissivity, active soil depth, and surface roughness length (see Surface Characteristics). In addition, climatoligical/equilibrium temperatures and volumetric water contents, the maximum moisture capacity of the vegetation canopy, as well as transfer coefficients and restoring time constants are specified in the prognostic equations.
  • The turbulent flux of moisture from the surface to the atmosphere includes direct evaporation from the vegetation canopy and from bare soil, as well as transpiration by the foliage (see Surface Fluxes).
  • Transpiration ceases when the soil moisture reaches a specified wilting point corresponding to a water potential of -15 bar; evaporation occurs at the potential rate when soil moisture is intermediate between its field capacity (corresponding to a hydraulic conductivity of 1x10^-4 m/day) and a saturation value that depends on the soil texture (see Surface Characteristics). Surface runoff occurs when the saturation value of soil moisture is exceeded.

Go to CNRM References

Return to Model Table of Contents

Return to Main Document Directory

Last update July 2, 1996. For further information, contact: Tom Phillips ( phillips@tworks.llnl.gov )

 LLNL Disclaimers