Status of numerical models

An overview of current numerical models

Clouds in the operational Met Office mesoscale model
Clouds in the operational ECMWF model
Clouds in the operational METEO FRANCE model
Clouds in the operational DWD Lokal Modell

Model data


An overview of current numerical models

Numerical models will never be able to represent individual clouds, so the atmosphere is split up into boxes, of typical size 50 to 200 km in the horizontal and 1 km in the vertical, and clouds within the box are represented by prognostic variables such as:

and within a decade

Size is currently prescribed, with different liquid droplet size for marine and continental clouds, and an ice particle size which varies with temperature. Cloud overlap affects radiation and precipitation efficiency. A vertical stack of grid boxes partially filled with cloud which is continuous in the vertical is assumed to be maximally overlapped, but if they are separated by cloud free layers then overlap is assumed to be random. A further issue not yet addressed is the degree of overlap of cloud inhomogeneities within a grid box.

In the project the following variables will be observed and compared with values held in the operational models or values diagnosed from the model variables:

Key issues

At the ECMWF seminar on Key Sub-grid parameterisation issues in NWP, 3-8 September 2001, the issue of sub-grid cloud variability was identified as the major difficulty needing attention if forecasts are to be improved.

To this we may add the representation of layers of supercooled clouds. These are quite common, have an important radiative effect and yet most models cannot represent them. For example, the ECMWF model represents the liquid water fraction in clouds by a simple temperature function, so that all clouds at 0°C, and the fraction falls monotonically to zero at -22°C.

Currently we have virtually no observations of ice water content of clouds and yet this variable is crucial for the earth's radiative balance and for the production of precipitation. Very little is known about the frequency or characteristics of mixed phase clouds.

Each site is equipped with radar, lidar and a suite of passive instrumentation. The observations will be used in the evaluation of four operational numerical models, and to demonstrate the role that could be played by an operational network of cloud remote sensing stations. We will also be exploiting the extensive existing datasets from numerous observational campaigns that have already taken place.