Institute of Atmospheric Physics (of the Chinese Academy of Sciences): Model IAP IAP-2L (4x5 L2) 1993

Institute of Atmospheric Physics (of the Chinese Academy of Sciences): Model IAP IAP-2L (4x5 L2) 1993

AMIP Representative(s)

Dr. Qing-cun Zeng, Laboratory of Numerical Modelling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Chinese Academy of Sciences, P.O. Box 2718, Beijing 100080, China; Phone: +86-1-2562347; Fax: +86-1-2562347

Model Designation

IAP IAP-2L (4x5 L2) 1993

Model Lineage

The IAP model consists of a special dynamical framework developed by Zeng and Zhang (1987) [1] and Zeng et al. (1987) [2] combined with physics similar to that of the Oregon State University model described by Ghan et al. (1982) [3].

Model Documentation

The principal documentation of the IAP model is provided by Zeng et al. (1989) [4].

Numerical/Computational Properties

Horizontal Representation

The available-energy conserving, finite-difference scheme of Zeng and Zhang (1982) [5] and Zeng et al. (1987) [2] is applied on a staggered C-grid (cf. Arakawa and Lamb 1977) [6].

Horizontal Resolution

4 x 5-degree latitude-longitude grid.

Vertical Domain

Surface to 200 hPa for dynamics (with the highest prognostic level at 400 hPa). For a surface pressure of 1000 hPa, the first atmospheric level is at 800 hPa. See also Vertical Representation and Vertical Resolution.

Vertical Representation

Finite differences in modified sigma coordinates: sigma = (P - PT)/(PS - PT), where P is atmospheric pressure, PT is 200 hPa (the dynamical top of the model), and PS is the surface pressure.

Vertical Resolution

There are two equally spaced, modified sigma levels (see Vertical Representation). For a surface pressure of 1000 hPa, these are at 800 hPa and 400 hPa (with the dynamical top at 200 hPa).

Computer/Operating System

The AMIP simulation was run on a Convex-C120 computer using a single processor.

Computational Performance

For the AMIP experiment, about 5 minutes of Convex-C120 time per simulated day.


For the AMIP experiment, the initial conditions for the atmosphere, soil moisture, and snow cover/depth are obtained from a model simulation of perpetual January using the AMIP-prescribed ocean temperatures and sea ice extents for 1 January 1979. See also Ocean and Sea Ice.

Time Integration Scheme(s)

The model uses a leapfrog scheme, followed by time filtering to damp the computational mode (cf. Robert 1966) [7]. The pressure gradient force terms are also smoothed (cf. Schuman 1971) [8] to permit use of a longer time step, which is 6 minutes for dynamics, 30 minutes for diffusion, and 1 hour for physics (including radiation). The vertical flux of atmospheric moisture is also computed hourly, and it is recomputed if conditional instability of a computational kind occurs (cf. Arakawa 1972) [9], as evidenced by relative humidities in excess of 100 percent.


  • Orography is smoothed (see Orography). Poleward of 70 degrees latitude, the wave-selected damping technique of Arakawa and Lamb (1977) [6] and the Fast Fourier Transform algorithm of Lu (1986) [10] are applied; from 38 to 70 degrees latitude, the recursive operator of Fjortoft (1953) [11] is also used. In addition, a Shapiro (1970) [12] smoothing operator is applied to perturbation values of surface pressure, temperature, and water vapor mixing ratio, and zonally on the wind field once per hour (cf. Liang 1986) [13]. A 9-point horizontal areal smoothing of the lapse rate and a three dimensional smoothing of the diabatic heating are also performed. See also Time Integration Scheme(s).

  • Filling of spurious negative values of moisture is accomplished by application of the numerical scheme of Liang (1986) [13] to the advection of atmospheric water vapor.

Sampling Frequency

For the AMIP simulation, the model history is written every 6 hours.

Dynamical/Physical Properties

Atmospheric Dynamics

Primitive-equations dynamics are expressed in terms of wind velocity, temperature, specific humidity, and a pressure parameter (PS-PT), where PS is the surface pressure and PT is 200 hPa, the pressure of the dynamical top of the model (see Vertical Representation). The dynamical framework utilizes perturbations from the temperature, geopotential, and surface pressure of the model's standard atmosphere (cf. Zeng 1979 [14], Zeng et al. 1987 [2], and Zeng et al. 1989 [4]).


  • Nonlinear horizontal diffusion of heat, momentum, and moisture following Smagorinsky (1963) [15] and Washington and Williamson (1977) [16] is applied on the modified sigma surfaces (see Vertical Representation).

  • There is no vertical diffusion as such, but momentum may be redistributed between the two atmospheric layers (see Vertical Resolution) by either eddy viscosity or convective friction. When the latter dominates, the friction coefficient depends on whether midlevel or penetrative convection occurs (see Convection). Cf. Zeng et al. (1989) [4] for further details.

Gravity-wave Drag

Gravity-wave drag is not modeled.

Solar Constant/Cycles

The solar constant is the AMIP-prescribed value of 1365 W/(m^2). Both seasonal and diurnal cycles in solar forcing are simulated.


The carbon dioxide concentration is the AMIP-prescribed value of 345 ppm. Above 200 hPa (the model dynamical top), a vertically integrated zonal ozone profile is specified from data of Dütsch (1971) [17], and is updated daily by linear interpolation between the 15th day of consecutive months. The radiative effects of water vapor, but not of aerosols, are also included (see Radiation).


  • Shortwave radiation is calculated in ultraviolet (wavelengths <0.5 micron) and visible (wavelengths 0.5-0.9 micron) spectral intervals, employing a delta-Eddington approximation (cf. Cess 1985 [18] and Cess et al. 1985 [19]). The shortwave calculations include treatment of Rayleigh scattering, absorption by water vapor using the exponential sum-fit method of Somerville et al. (1974) [20], absorption by ozone following Cess and Potter (1987) [21] and Lacis and Hansen (1974), [22] and scattering/absorption by cloud droplets. The optical properties (single-scattering albedo, asymmetry factor, and optical depth) of these droplets depend on cloud type (see Cloud Formation).

  • Longwave absorption by carbon dioxide and water vapor, with empirical transmission functions after Katayama (1972) [23], is calculated for five spectral intervals with wavelengths >0.9 micron. Cirrus clouds or cirrus anvils on convective clouds (see Cloud Formation) are treated as graybodies (emissivity of 0.5, expressed as a modified fractional cloudiness), and other clouds as blackbodies (emissivity of 1.0). For purposes of the radiation calculations, blackbody clouds overlap fully in the vertical, but graybody clouds only partially. Cf. Zeng et al. (1989) [4] for further details.


  • The parameterization of convection after Arakawa et al. (1969) [24] includes these elements: dry convective adjustment; shallow, midlevel, and penetrative convection; determination of the cumulus mass flux; and modification of the planetary boundary layer (PBL). If either atmospheric layer (see Vertical Resolution) is dry adiabatically unstable, the potential temperatures are adjusted to a common value, such that dry static energy is conserved. The layers that display evidence of moist convective instability determine whether (mutually exclusive) shallow, midlevel, or penetrative convection occur. Precipitation is associated with midlevel and penetrative convection (see Precipitation), and changes in PBL temperature and humidity are associated with shallow and penetrative convection (see Planetary Boundary Layer). Convective clouds are assumed to form a steady-state ensemble, with cloud air being saturated; the vertical profile of cloud temperature is computed assuming that moist static energy is conserved.

  • Cumulus mass flux is computed (differently) for midlevel and penetrative convection, assuming that the convective instability is removed with an e-folding time of 1 hour. (Cumulus mass flux is not associated with shallow convection in the model.) The mass flux from the PBL and that entrained from the free atmosphere are mixed in the lower part of the convective cloud, with detrainment of this mixture at the level of nonbuoyancy. The magnitude of the mass flux determines the amount of cumulus friction (with different coefficients for midlevel and penetrative convection), which brings about momentum transfer between the two atmospheric layers (see Diffusion). Cf. Zeng et al. (1989) [4] for further details.

Cloud Formation

  • Clouds may result either from large-scale condensation or from convection (see Convection). Large-scale clouds form when the relative humidity exceeds 90 percent in the lower vertical layer or 100 percent in the upper layer (see Vertical Resolution). Clouds are assumed to fill a grid box (cloud fraction = 1). (For purposes of longwave radiation calculations, however, graybody cirrus and cirrus anvil cloud fractions are 0.5.)

  • Four basic cloud types are modeled. Type 1 is formed by either midlevel or penetrative convection, type 2 when the relative humidity of the lower vertical layer exceeds 90 percent, type 3 as a result of shallow convection, and type 4 when large-scale precipitation occurs in the upper vertical layer (see Precipitation). (Precipitation is limited to cloud types 1, 2, and 4.) Types 2, 3, or 4 are defined as graybody cirrus clouds if they are nonprecipitating or if the average of their base and top temperatures is <-40 degrees C; convective cloud type 1 is also capped by a graybody cirrus anvil if it satisfies this temperature criterion.

  • In addition, a cloud type 5 is formed by the coexistence of types 2 and 4, and a cloud type 6 by the coexistence of types 3 and 4. (Types 1 and 3 cannot coexist; type 1 also overrides the formation of types 2 and 4, while type 2 overrides type 3.) Both types 5 and 6 are treated as low-level clouds for radiation purposes. Cf. Zeng et al. (1989) [4] for further details. See also Radiation.


  • Precipitation may result from midlevel or penetrative convection (see Convection). The amount of precipitation is equal to the net water vapor entrained from the environment, and falls at a rate that is a function of the cumulus mass flux. There is no subsequent evaporation of convective precipitation.

  • Precipitation also results from large-scale condensation. When the upper atmospheric layer becomes supersaturated, the resulting precipitation is assumed to evaporate completely in passing through the lower layer, thereby cooling and moistening the environment in proportion to the amount of evaporation (cf. Lowe and Ficke 1974) [25]. If the lower layer then becomes supersaturated, the resulting precipitation falls to the surface.

Planetary Boundary Layer

The PBL is parameterized as a constant flux surface layer of indefinite thickness (see Surface Fluxes). Its temperature and humidity are modified by shallow and penetrative convection (see Convection), with new values computed assuming conservation of moist static energy in the vertical.


The model's orography is determined by area-averaging the 1 x 1-degree topographic data of Gates and Nelson (1975) [26] within each 4 x 5-degree grid box. A 9-point smoothing of orography on neighboring model grid squares is also performed.


AMIP monthly sea surface temperatures are prescribed, with intermediate daily values determined by linear interpolation.

Sea Ice

AMIP monthly sea ice extents are prescribed. The surface temperature of the ice is determined from a budget equation that includes the surface heat fluxes (see Surface Fluxes) plus conduction heating from the ocean below the ice. This subsurface flux is a function of the heat conductivity and thickness (a constant 3 m) of the ice, and of the difference between the predicted ice temperature and that prescribed (271.5 K) for the ocean below. Snow is allowed to accumulate on sea ice (see Snow Cover), and melts if the ice surface temperature is >0 degrees C.

Snow Cover

Precipitation falls as snow if the surface air temperature is <0 degrees C. The snow mass is determined from a budget equation that includes the rates of snow accumulation, melting, and sublimation. Snowmelt (which contributes to soil moisture--see Land Surface Processes) is computed from the difference between the downward heat fluxes at the surface and the upward heat fluxes that would occur for a surface temperature equal to the melting temperature of ice (0 degrees C). (For snow on sea ice, the conduction heat flux from the ocean below also contributes to snowmelt--see Sea Ice.) The sublimation rate is set equal to the surface evaporative flux (see Surface Fluxes) unless all the snow mass is removed in less than one hour; in this case, sublimation is equated to the rate of snow mass removal. Cf. Zeng et al. (1989) for further details.

Surface Characteristics

  • Roughness lengths are not specified, since these are not required for the calculation of turbulent fluxes (see Surface Fluxes).

  • The surface albedo is prescribed for water, land ice, sea ice, and for six land surface types. Following Manabe and Holloway (1975) [27], the albedo of snow-covered surfaces varies as the square root of snow mass up to a maximum value that is assigned for a critical snow mass of 10 kg/(m^2) (see Snow Cover). Albedos for the land surfaces (with and without snow cover) are assigned from data of Posey and Clapp (1964) [28]. Over water, the albedo for diffuse solar flux is taken as 0.07, and that for the direct beam is a function of solar zenith angle (cf. Zeng et al. 1989) [4].

  • Longwave emissivity is prescribed as unity (blackbody emission) for all surface types.

Surface Fluxes

  • The absorbed surface solar flux is determined from the surface albedo, and surface longwave emission from the Planck function with constant surface emissivity of 1.0 (see Surface Characteristics).

  • Turbulent surface fluxes of momentum, heat, and moisture are calculated from bulk aerodynamic formulae. The momentum flux is proportional to the product of a drag coefficient and the effective surface wind, which is determined by extrapolating the wind at the two vertical levels and multiplying by a factor of 0.7 (but constrained to be at least 2 m/s in magnitude). The drag coefficient is not a function of vertical stability but it does depend on surface elevation. Over the oceans, the drag coefficient is a function of the surface wind speed, but it is constrained to be at most 2.5 x 10^-3.

  • The surface heat and moisture fluxes also depend on a product of the same drag coefficient and effective surface wind speed, as well as on the difference between the ground and surface atmospheric temperatures (for heat fluxes) or specific humidities (for moisture fluxes). These surface values of atmospheric temperature and humidity are determined by equating, respectively, the surface sensible heat and evaporative flux to corresponding fluxes at the top of the constant-flux surface layer. The latter are parameterized following K-theory (cf. Arakawa 1972) [9], where the eddy diffusivity depends on vertical stability but is constrained to be < 15 m^2/s over water surfaces, and to be <100 m^2/s elsewhere. The surface moisture flux also depends on an evapotranspiration efficiency factor beta, which is taken as unity over snow, ice, and water, and in areas of dew formation. Over land, beta is a function of soil moisture. Cf. Zeng et al. (1989) [4] for further details. See also Diffusion, Planetary Boundary Layer, and Land Surface Processes.

Land Surface Processes

  • Following Priestly (1959) [29] and Bhumralkar (1975) [30] the average ground temperature at the diurnal skin depth is computed from the net surface energy fluxes (see Surface Fluxes), taking account of the thermal conductivity, and the volumetric and bulk heat capacities of snow, ice, and soil.

  • Soil moisture is predicted as a fraction of a uniform field capacity of 0.15 m in a single layer (i.e., a "bucket" model). Fractional soil moisture is determined from a budget that includes the rates of precipitation, snowmelt, surface evaporation, and runoff. The evapotranspiration efficiency beta over land (see Surface Fluxes) is specified as the lesser of twice the fractional soil moisture or unity. Runoff is given by the product of the fractional soil moisture and the sum of precipitation and snowmelt rates. If the predicted fractional soil moisture is in excess of unity, the excess is taken as additional runoff. Cf. Zeng et al. (1989) [4] for further details.

Go to IAP References

Return to IAP Table of Contents

Return to Main Document Directory

Last update April 19, 1996. For further information, contact: Tom Phillips ( )

LLNL Disclaimers