Meeting Talks


Clouds in the operational Met Office mesoscale model

(Damian Wilson 18/3/03)
This web page provides a summary of the way cloud and condensate is diagnosed or predicted in the Met Office mesoscale model. I have assembled this for the Cloudnet project although it may be a useful reference source for others.

I have assembled the information in the form of a summary table, with links to papers and more complete summaries, which should supply required information at three levels of depth.


To access these documents, Email for the password

Cloud Fraction:

Condensed water:



Level RHcrit
1 0.91
2 0.90
3 0.88
4 0.86
5 0.84
6 0.82
7-38 0.80

Anvil parametrization parameters:

Anvil factor 3.0
Tower factor 0.25

Summary of q+l-qsat PDF method:

Here we assume that, within a gridbox, there exists a known distribution, G, of a variable 's', where

s = aL (qT' - qsat').

qT is the sum of the vapour plus the liquid water content and qsat is the saturation specific humidity with respect to liquid water at temperature TL.

aL = 1 / (1 + alpha L/cp)

where alpha is dqsat/dT at constant pressure, L is the latent heat of vaporization of water and cp is the heat capacity of air.

TL = T - l L/cp.

The ' symbol represents deviations from grid box means. The distribution of `s' is a symmetric triangle (about s = 0) where the width (where the function goes to zero) is parametrized by plus or minus aL (1-RHcrit) qsat. The water content of a point in the distribution is given by

l = aL (qT - qsat) = Qc + s


Qc = aL (qT_mean - qsat_mean)

and qT_mean and qsat_mean are gridbox means. Hence the liquid cloud fraction can be given simply by the integral of G(s) above the limit where s = -Qc. The liquid water content is given by the first moment of this integral.
Further details are in the Smith 1990 paper and UM documentation paper 29

Summary of empirically adjusted cloud fraction method:

This is an adjustment to the Smith scheme in order to increase the cloud fractions which it produces, in order to better match observations. This code will replace the Qc used in the Smith scheme calculations of cloud fraction with an increased value of Qc, to produce increased liquid cloud fractions. It does this with the linear mapping:

Qc'/bs = (Qc/bs + K2) / K1


K1 = 1 - K2 and K2 = 0.184 for levels 1-13 and 0.0955 for levels 14 and above.

The liquid water content is not adjusted. A similar modification is done to the ice cloud fraction diagnostic: an equivalent Qc is calculated, consistent with the Smith scheme, this is modified as above, then the cloud fraction is recalculated. Again, this will increase the cloud fraction.
Further details are in the Cusack 2002 note.

Summary of area cloud fraction method:

This parametrization aims to better produce thin stratocumulus cloud where the vertical resolution is not sufficient to resolve it. Each model layer is split into three sub layers, and values of qT and TL interpolated from the the full model layer (and those surrounding it) into the sub layer. The interpolation method depends on whether there are strong jumps in temperature across the full model layers, thus suggesting the presence of a sharp inversion, the sharpness of which needs to be maintained. The cloud fraction in each of the sub-layers is then calculated as above and averaged to produce the bulk cloud fraction (hence this may be greater than the bulk cloud fraction calculated using the full model layer). The radiation will see an area cloud fraction, ade up of the maximum of the fraction in each of the sub-layers.
Further details are in UM documentation paper 29.

Summary of ice cloud fraction method:

The ice cloud fraction cannot be calculated using a PDF formulation since the assumption of instantaneous condensation is not applied to ice. Instead we invert the implicit relationship that exists in the Smith 1990 scheme between cloud fraction and condensed water content (using saturation specific humidity with respect to liquid water). This provides a diagnostic relationship for ice cloud fraction as a function of ice content, qsat and RHcrit. It is NOT a function of vapour content. The functional form is given in Wilson and Ballard, 1999. This ice fraction is then modified by the empirically adjusted cloud fraction formulation.

Summary of convective cloud fraction method:

This is a straightforward diagnostic.

convective cloud = 0.7873 + 0.06 ln(TCW)

where TCW is the total cloud water content before convective precipitation is applied in kg m-2.
Further details are in UM documentation paper 27.

Summary of anvil cloud:

This scheme modifies the shape of the convective cloud fraction so that it is not a single, uniform value from cloud base to cloud top, but has a more anvil shape associated with it.
Details will follow when I have located them!

Summary of microphysics scheme:

This scheme considers transfer processes which act between vapour, liquid water, ice and rain. The processes which are parametrized are: fall of ice; nucleation of ice; deposition and sublimation of ice; riming; capture of rain by snow; melting of ice; evaporation of melting snow; evaporation of rain; accretion; autoconversion. Rain is parametrized as a diagnostic and is assumed to fall out in a single timestep.
A transfer diagram is available here.
Further details are in UM documentation paper 26 and in Wilson and Ballard 1999.

Damian Wilson 18/3/03.

These pages are maintained by .
Return to Radar Group | Department of Meteorology | University of Reading