University of Illinois at Urbana-Champaign: Model UIUC MLAM-AMIP (4x5 L7) 1993

University of Illinois at Urbana-Champaign: Model UIUC MLAM-AMIP (4x5 L7) 1993

AMIP Representative(s)

Dr. Michael Schlesinger, Department of Atmospheric Sciences, University of Illinois at Urbana-Champaign, 105 South Gregory Avenue, Urbana, Illinois 61801; Phone: +1-217-333-2192; Fax: +1-217-244-4393; e-mail:; World Wide Web URL:

Model Designation

UIUC MLAM-AMIP (4x5 L7) 1993

Model Lineage

The UIUC multilevel atmospheric model (MLAM) traces its origins to the two-layer Oregon State University model described by Ghan et al. (1982) [1]. Subsequent modifications principally include an increase in vertical resolution from 2 to 7 layers, as well as substantial changes in the treatment of atmospheric radiation, convection, cloud/precipitation formation, and land surface processes.

Model Documentation

The dynamical structure and numerics of the UIUC model, as well as some of its surface schemes are as described by Ghan et al. (1982) [1] The parameterizations of radiation, cloud formation, and related physics are discussed by Oh (1989) [2] and by Oh and Schlesinger (1991a [3], b [4], c [5])

Numerical/Computational Properties

Horizontal Representation

Finite differences on a B-grid (cf. Arakawa and Lamb 1977) [6], conserving total atmospheric mass, energy, and potential enstrophy.

Horizontal Resolution

4 x 5-degree latitude-longitude grid.

Vertical Domain

Surface to 200 hPa (model top). For a surface pressure of 1000 hPa, the lowest prognostic level is at 990 hPa and the highest is at 280 hPa.

Vertical Representation

Finite-difference sigma coordinates.

Vertical Resolution

There are 7 unevenly spaced sigma layers between the surface and the model top at 200 hPa. (Proceeding from the surface, the thicknesses of the bottom three layers are about 20 hPa, 40 hPa, and 100 hPa, while the upper four layers are each 160 hPa thick).

Computer/Operating System

For the AMIP simulation, the model was run on a Cray C90 computer using one processor in a UNICOS environment.

Computational Performance

For the AMIP experiment, about 1.25 minutes of Cray C90 computer time per simulated day.


For the AMIP simulation, initial conditions for the atmosphere, soil moisture, and snow cover/depth for 1 January 1979 are specified from a previous model simulation of January.

Time Integration Scheme(s)

For integration of dynamics each hour, the first step by the Matsuno scheme is followed by a sequence of leapfrog steps, each of length 6 minutes. The diabatic terms (including full radiation calculations), dissipative terms, and the vertical flux convergence of the specific humidity are recalculated hourly.


Orography is area-averaged on the model grid (see Orography). A longitudinal smoothing of the zonal pressure gradient and the zonal and meridional mass flux is performed at latitudes polewards of 38 degrees (cf. Ghan et al. 1982 [1]). It is unnecessary to fill spurious negative values of atmospheric moisture, since these are not generated by the numerical schemes.

Sampling Frequency

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

Dynamical/Physical Properties

Atmospheric Dynamics

Primitive-equations dynamics are expressed in terms of u and v winds, temperature, surface pressure, and specific humidity. Cloud water is also a prognostic variable (see Cloud Formation).


  • Horizontal diffusion is not modeled.

  • Vertical diffusion of momentum, sensible heat, and moisture operates at all vertical levels. The diffusion depends on the vertical wind shear, but not on stability (cf. Oh 1989 [2] and Oh and Schlesinger 1991a [3]).

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. The daily horizontal distribution of column-integrated ozone is interpolated from prescribed monthly mean Total Ozone Mapping Spectrometer (TOMS) data (for example, cf. Stolarski et al. 1991 [7]). The radiative effects of water vapor, methane, nitrous oxide, and chlorofluorocarbon compounds CFC-11 and CFC-12 are also included, but not those of aerosols (see Radiation).


  • The spectral range of shortwave radiation is divided into three intervals: 0 to 0.44 micron, 0.44 to 0.69 micron, and 0.69 to 3.85 microns. The first two intervals are for the treatment of Rayleigh scattering (after Coakley et al. 1983 [8]) and ozone and carbon dioxide absorption (after Lacis and Hansen 1974 [9] and Fouquart 1988 [10], respectively); the last interval (further subdivided into six subintervals) is for water vapor absorption. Scattering and absorption by both gases and cloud droplets are modeled following a two-stream approach with delta-Eddington approximation. The optical depth and single-scattering albedo for cloud droplets are determined following Stephens (1978) [11] for non-ice clouds and Starr and Cox (1985) [12] for cirrus clouds.

  • The longwave flux calculations are based on a two-stream formulation with parameterized optical depths, but with scattering neglected. Longwave absorption is treated in four spectral bands, one each for carbon dioxide (5.4 x 10^4 to 8.0 x 10^4 m^-1) and ozone (9.8 x 10^4 to 1.1 x 10^5 m^-1), and the other two bands (with multiple subintervals between 0 and 3.0 x 10^5 m^-1) for the line centers, wings, and continuum of the water vapor absorption spectra. Pressure-broadening effects are included in all the absorption calculations, which follow Chou (1984) [13] for water vapor, Kneizys et al. (1983) [14] for ozone, Chou and Peng (1983) [15] for carbon dioxide, Donner and Ramanathan (1980) [16] for methane and nitrous oxide, and Ramanathan et al. (1985) [36] for chlorofluorocarbon compounds CFC-11 and CFC-12. The absorption by trace gases (methane, nitrous oxide, CFC-11, and CFC-12) is normalized in each subinterval. Absorption by cloud droplets is treated by an emissivity formulation based on data by Stephens (1978) [11] for non-ice clouds, and by Starr and Cox (1985) [12] and Griffith et al. (1980) [17] for extratropical and tropical cirrus clouds, respectively.

  • The radiation parameterization includes cloud-cover feedback by calculating separately the radiative fluxes for the cloudy and clear portions of each grid box; it includes cloud optical-depth feedback by linking the radiative properties to the prognostic cloud water content (see Cloud Formation). Clouds are vertically distributed by groups that make up an ensemble of contiguous cloud layers, and which are separated from each other by at least one layer of clear air. Following Geleyn (1977) [18], the contiguous cloud layers within each group overlap one another fully in the vertical, while the noncontiguous cloud groups overlap randomly. Cf. Oh (1989) [2] and Oh and Schlesinger (1991c) [5] for further details.


  • Penetrative convection is simulated by a modified Arakawa-Schubert (1974) [19] scheme. The scheme predicts mass fluxes from mutually interacting cumulus subensembles which have different entrainment rates and levels of neutral buoyancy (depending on the properties of the large-scale environment) that define the tops of the clouds and their associated convective updrafts. In turn, the predicted convective mass fluxes feed back on the large-scale fields of temperature (through latent heating and compensating subsidence), moisture (through precipitation and detrainment), and momentum (through cumulus friction). The effects on convective cloud buoyancy of phase changes from water to ice, and the drying and cooling effects of convective-scale downdrafts on the environment are not explicitly included.

  • The cloud-base mass flux for each cumulus sub-ensemble is determined following Arakawa (1969) [20] such that the convective instability for each subensemble is removed with an e-folding time of one hour (cf. Oh 1989[2]).

  • The model also simulates middle-level convection, defined by convective instability between any two adjacent layers, with the instability also being removed with an e-folding time of one hour. In addition, if the lapse rate becomes dry-convectively unstable anywhere within the model atmosphere, enthalpy is redistributed vertically in an energy-conserving manner.

Cloud Formation

  • The cloud parameterization is formulated separately for stratiform and cumuloform clouds, as described by Oh (1989) [2] and Oh and Schlesinger (1991b) [4]. For both cloud types, the liquid/ice water is computed prognostically, and the fractional cloud coverage of each grid box semiprognostically. The stratiform cloud fraction varies as the square root of the relative humidity. The cumuloform cloud fraction is determined as a function of the relative humidity and the convective mass flux (see Convection).

  • Cloud in the PBL (see Planetary Boundary Layer) is diagnostically computed on the basis of a cloud-topped mixed layer model (cf. Lilly 1968 [21] and Guinn and Schubert 1989 [22]).


  • Precipitation forms via the simulated microphysical processes (autoconversion from cloud liquid/ice water) in the prognostic cloud scheme (cf. Oh 1989 [2] and Oh and Schlesinger 1991b [4]). The large-scale precipitation rate is an exponential function of the liquid water mixing ratio and the cloud water content. The difference of these quantities multiplied by the cumulus mass flux yields the convective precipitation rate (see Convection).

  • The rate of evaporation of falling large-scale precipitation is proportional to the product of the rainfall rate, the relative humidity deficit from saturation, and the cloud-free fraction of the grid box. Evaporation of convective precipitation is proportional to the product of the relative humidity deficit and the cloud water content.

Planetary Boundary Layer

The top of the PBL is taken to be the height of the lowest three atmospheric layers (total thickness about 160 hPa for a surface pressure of 1000 hPa). PBL cloud is diagnostically computed on the basis of a cloud-topped mixed layer model. See also Cloud Formation, Diffusion, Surface Characteristics and Surface Fluxes


Orography, obtained from the 1 x 1-degree data of Gates and Nelson (1975) [23], is area-averaged over each 4 x 5-degree model grid square.


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

Sea Ice

AMIP monthly sea ice extents are prescribed. The surface temperature of the ice is determined prognostically from the surface energy balance (see Surface Fluxes) including heat conduction from the ocean below. The conduction flux is a function of the prescribed heat conductivity and ice thickness (a constant 3 m), and of the difference between the surface temperature and that of the ocean (a fixed 271.5 K). When snow accumulates on sea ice, this conduction flux can contribute to snowmelt. Cf. Ghan et al. (1982) [1] for further details.

Snow Cover

Precipitation falls as snow if the surface air temperature is < 0 degrees C. Snow mass is determined from a prognostic budget equation that includes the rates of accumulation, melting, and sublimation. Over land, the rate of snowmelt is computed from the difference between the downward heat fluxes at the surface and the upward heat fluxes that would occur for a ground temperature equal to the melting temperature of snow (0 degrees C); snowmelt contributes to soil moisture (see Land Surface Processes). Accumulation and melting of snow may also occur on sea ice (see Sea Ice). The surface sublimation rate is equated to the evaporative flux from snow (see Surface Fluxes) unless sublimation removes all the local snow mass in less than 1 hour; in that case the sublimation rate is set equal to the snow-mass removal rate. Snow cover also alters the surface albedo (see Surface Characteristics). Cf. Ghan et al. (1982) [1] for further details.

Surface Characteristics

  • Surface roughness is specified as in Hansen et al. (1983) [24]. Over land, the roughness length is a fit to the data of Fiedler and Panofsky (1972) [25] as a function of the standard deviation of the orography. The maximum of this value and that of the roughness of the local vegetation (including a "zero plane displacement" value for tall vegetation types--cf. Monteith 1973 [26]) determines the roughness length over land. Over sea ice, the roughness is a constant 4.3 x 10^-4 m after Doronin (1969) [27]. Over ocean, the roughness length is a function of the surface wind speed, following Garratt (1977) [28].

  • Snow-free surface albedo is updated monthly by interpolation using values for January, April, July, and October specified from data of Matthews (1983) [29]. The albedo of snow-covered surfaces is determined as a linear weighted (by snow depth) interpolation of snow-free and snow-covered values. The albedo of snow is a function of its temperature (cf. Manabe et al. 1991 [30]); it also depends on solar zenith angle (cf. Briegleb and Ramanathan 1982 [31]), but not on spectral interval.

  • Longwave emissivity is specified to be unity (blackbody emission) for all surfaces.

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).

  • The turbulent surface fluxes of momentum, sensible heat, and moisture are parameterized as bulk formulae that include surface atmospheric values of winds, as well as differences between skin values of temperatures and specific humidities and their surface atmospheric values. Following Oh and Schlesinger (1990) [32], the surface wind is taken as a fraction (0.7 over water and 0.8 over land and ice) of the winds extrapolated from the lowest two model layers. The surface atmospheric values of temperatures and humidities are taken to be the same as those at the lowest atmospheric level (sigma = 0.990).The aerodynamic drag and transfer coefficients depend on vertical stability (bulk Richardson number) and surface roughness length (see Surface Characteristics), with the same transfer coefficient used for the fluxes of sensible heat and moisture. In addition, the surface moisture flux depends on an evapotranspiration efficiency beta that is taken as unity over snow, ice and water; over land, beta is a function of the fractional soil moisture (see Land Surface Processes).

Land Surface Processes

  • Following Priestly (1959) [33] and Bhumralkar (1975) [34], the average ground temperature over the diurnal skin depth is computed from a prognostic budget equation whose source/sink terms include the net surface radiative flux and the sensible and latent heat fluxes (see Surface Fluxes); the thermal conductivity, volumetric heat capacity, and bulk heat capacity of snow, ice, and land are also taken into account. If the predicted ground temperature for land ice is > 0 degrees C, the ice is implicitly assumed to melt, since the model does not include a budget equation for land ice. See also Snow Cover.

  • Soil wetness is expressed as the ratio of soil moisture content to a field capacity that is specified as a function of soil texture and surface cover after data of Vinnikov and Yeserkepova (1991) [35]. Soil wetness is determined from a prognostic budget equation that includes the rates of precipitation, snowmelt, surface evaporation, and runoff. The evapotranspiration efficiency beta over land (see Surface Fluxes) is assigned a value that is the lesser of 1.33 times the soil wetness fraction or unity. The runoff rate is a nonlinear function of the soil wetness and the combined rates of precipitation and snowmelt. If the predicted soil wetness fraction exceeds unity, the excess moisture is taken as additional runoff.

Go to UIUC References

Return to UIUC Table of Contents

Return to Main Document Directory

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

LLNL Disclaimers