Cloud evaluation

Cloud and large-scale precipitation processes are described by prognostic equations for cloud liquid water, cloud ice, rain, snow and a grid box fractional cloud cover.   The IFS predicts the three-dimensional cloud field with three variables for each grid box; cloud fraction, cloud liquid water and cloud ice.  Cloud processes such as condensation, evaporation, glaciation and precipitation formation in convective and stratiform clouds are all taken into account with physically-based equations.

There are some meteorological situations that are more challenging to forecast than others.  One of these is stratiform cloud beneath an inversion, especially subsidence inversions in high pressure situations, which can be difficult for atmospheric models to both analyse and forecast.  Often there is uncertainty regarding the cloud extent, phase, thickness or persistence.  This has a corresponding effect on radiation balance at the surface and consequently also upon near-surface temperatures.  Users should check the analysed cloud against observations as far as possible in these circumstances (see also model boundary layer).


Fig2.1.5.2-1: Schematic diagram to illustrate the parameterised processes for precipitation and clouds within a single grid box.  A cloud-overlap algorithm calculates the relative placement of clouds across IFS model levels.  This is important for the “life history” of falling precipitation (from level-with-cloud to level-with-clear-sky and vice-versa) and this process may occur several times during the descent of the IFS model precipitation. 


Cloud processes

The convective and stratiform processes are modelled by complex interactions and changes between the phases of water (vapour, liquid, supercooled water, and solid) and include precipitation.  Convection processes (due to subgrid-scale convective updraughts) are calculated separately from larger scale cloud processes (e.g. due to large-scale ascent or radiative cooling), but the two schemes are connected and represent different parts of the cloud and precipitation in a grid column.  Cloud is parameterised within a grid box as stratiform or convective according to the stability or instability of the IFS model atmosphere but it is possible to have both types where convection does not extend throughout the IFS model troposphere (e.g. convection limited to lower troposphere with stable moist layer above).

Only rain or snow is produced by the precipitation scheme; hail is not considered nor developed in the IFS model convection scheme, no matter how unstable is the IFS model atmosphere.  Although cloud phase changes, including super-cooled water, are modelled in the clouds, freezing rain at the ground is forecast when rain from cloud falls through a lower layer of the atmosphere with sub-zero temperatures.

Large scale cloud processes are calculated using a complex suite of programs dealing with the interactions between the phases of water.  The precipitation from the large scale cloud scheme has a finite fall speed which is different for rain and snow, and can be blown laterally by the wind across grid boxes during descent.

The convection processes are similarly calculated using a complex suite of programs dealing with the interactions between the phases of water.  The convection process "detrains" cloud and precipitation into the large-scale cloud scheme.  This allows representation of convective anvils and also the precipitation associated with the more stratiform part of the convective cell.  However, the main part of the precipitation from the core of the updraught is treated diagnostically in the convection scheme, and assumes all the precipitation falls out within the grid column in a timescale less than the model timestep.  Convective precipitation in the model falls vertically and immediately; it is not advected with winds underlying the model cloud.  Thus, model showers are not advected with the wind during their life-cycle.  In particular, any showers that the model develops over the sea do not penetrate beyond the coast while in reality, showers normally advect with the wind during their life-cycle.


Cloud Fraction

Cloud fraction is a prognostic model variable for each grid box.  This means at each timestep cloud fraction is advected into the grid box by the model wind and then acted upon by sources and sinks of moisture.   

Changes in cloud fraction depend upon:

  •                rate of change of cloud area due to transport through the boundaries of the grid volume.
  • Plus       the rate of change of cloud area by convective processes.
  • Plus       the rate of change of cloud area by stratiform condensation processes.
  • Minus    the rate of decrease of cloud area due to evaporation.

Total cloud cover

Total cloud cover is calculated from the predicted cloud fraction within each layer for each grid column.  The layers used are: low layer (surface to 800hPa), medium layer (800hPa to 450hPa) and high layer (450hPa to model top) for each grid column.  The model clouds are assumed to extend vertically over the whole of depth of each model layer.  The calculation uses assumptions about the overlap between the subgrid clouds in the vertical (whether the layers of cloud are stacked above one another in the vertical, or whether they are displaced relative to one another.

Fig2.1.5.2-2:  Example of Assessment of Cloud Cover. 
IFS model cloud layers are assigned as:

  • High-level cloud cover (HCC). - Cloud cover at each model layer integrated from top of the atmosphere down to 450hPa*.
  • Medium-level cloud cover (MCC). - Cloud cover at each model layer integrated from 450hPa* down to 800hPa*.
  • Low-level cloud cover (LCC). - Cloud cover at each model layer integrated from 800hPa* down to the surface.

But note: the Total Cloud Cover (TCC) is cloud layers integrated from the top of the atmosphere down to the surface with overlap assumptions based upon global observations.  The degree of randomness in the overlap is dependant upon distance between layers.   Hence TCC  ≤  HCC + MCC + LCC.

* strictly pressure levels are not used, but actually the IFS model levels that correspond to the given pressure levels in a standard atmosphere.  This means that one can get low cloud over the Tibetan plateau, for example, because we are using there the same IFS model levels to divide up cloud layers that we use the over open ocean.

Cloud base height

Cloud base height is derived by upwardly searching the structure of the model atmosphere to find the altitude where the cloud fraction with a covering of more than 1% of the model grid box and condensate content greater than 10-6  Kg Kg-1.  The upward search starts from the surface.

If no cloud base is found, but the convection scheme diagnoses a convective cloud base, then the cloud base height is set to the convective cloud base.

Ceiling

Cloud ceiling is a measurement used in the aviation industry to indicate airport landing conditions.

Ceiling height is derived by upwardly searching the structure of the model atmosphere to find the altitude where the base of the lowest layer of cloud with a covering of more than 50% of the model grid box and condensate content greater than 10-6 Kg Kg-1.

The upward search starts from the second lowest model level (Level 136, 31m).  Fog, or equivalently cloud in the lowest layer (Level 137, 10m), is not considered when deriving the ceiling.

Fig2.1.5.2-3: Schematic representation of difference assessing cloud base and ceiling.  Cloud base is derived by upwardly searching from the surface; the presence of fog implies cloud base at the surface, or possibly a base near 30m in the case of very low Stratus.  Ceiling is derived by upwardly searching from Level136 (about 31m) to avoid allocation of ceiling to the surface in case of fog. This can result in ceiling being quite high (or even there being no ceiling at all) even above surface fog. Cloud base only requires >1% cloud cover while ceiling requires 50% cloud cover.

 

Fig2.1.5.2-4: Schematic representation of difference assessing cloud base and ceiling showing the difference between derivation of Ceiling and Cloud Base.  

Fig2.1.5.2-5: Example of differing cloud base and ceiling heights where fog is forecast.  Forecasts are for fog (visibility 345m), cloud base (height 29.6m), ceiling (height 8726m) and the corresponding vertical profile.  Forecasts VT 00UTC Fri 28 Feb 2025, DT 00UTC Sun 23 Feb 2025.  The cloud base identified when the cloud cover is greater than 1% and the moisture greater than 10-6 Kg Kg-1.  Cloud cover overnight forecast less than 1% at 10m (Level 137) and separate from forecast of fog. The ceiling identified when the cloud cover is greater than 50% and the moisture greater than 10-6 Kg Kg-1.  This corresponds to the moist layer shown on the ENS control (ex HRES) vertical profile (Temperature in Red, Dew point in black). Shading represents the range of temperatures and dew points among the ensemble members.

Convective cloud top height

The height of convective cloud top is output from the convection scheme as the height where the convective updraught (velocity) vanishes.  This is not necessarily the same as where the model parcel temperature becomes the same as the model environment temperature.


   

 

(FUG Associated with Cy49r1)