Cumulus convection plays a major role in the vertical transport of heat, moisture, momentum and chemical tracers in the atmosphere. Because of the small horizontal size, cumulus clouds can not be explicitly resolved in climate and numerical weather prediction models. Hence, a parameterization is needed.
The convection scheme of the ECMWF model was equipped with a new trigger function, proposed by Siebesma and Jakob (2003). This new formulation originates from the idea that the equations that describe the updraft properties in the cloud layer can be extended to the subcloud layer. This gives opportunities to aim for a unified formulation for the cumulus-topped boundary layer as a whole. The updraft equations consist of assumptions about lateral mixing of thermodynamical properties and the vertical velocity. Furthermore, the roots of the clouds (thermals) are assumed to have a certain excess in heat and moisture with respect to their environment, at their prescribed release height.
First, this formulation is tested in an offline model and sensitivity tests for various parameters are performed. The model input is obtained from data gathered at the BBC2 cloud campaign at Cabauw (May 2003) and consists mainly of mast, radiosonde, aircraft and remote sensing measurements. The evaluation of model outcome is as well carried out with data from certain shallow cumulus days in this period. For the studied cases, we find that the offline model simulates the observed cloud convective properties considerably well, as well as the presence of clouds during the day. Cloud tops are overestimated in multiple layer situations of boundary layer clouds. For the sensitivity, we notice that the new formulation is less sensitive for changes in parcel release parameters than the formulation that can be associated with the old trigger function. This is mainly caused by the assumption of lateral exchange of the updraft with its environment in the subcloud layer.
In the second part of the study, coherent structures along the Cabauw mast are identified with help of high-frequent sonic anemometer measurements. Wavelet analysis is used to determine prevailing time scales, whereafter the characteristics of thermals during typical convective conditions are studied. For the 5 m level in Cabauw, we find dominant time scales that agree well with values found in Krusche and de Oliveira (2004), i.e. 19 – 39 s. This derived dominant time scale grows with height, as well as the uncertainty in it. The feedback of this part of the project on the offline model lies in the excess factors for temperature and humidity that can be obtained from mean thermal properties. We find excess factors bT = 2.68 0.08 ± and bq = ± 3.32 0.10 for sampling on simultaneous positive temperature and humidity anomalies, while sampling on positive anomalies in vertical velocity delivers 1.51 0.23 bT = ± and 1.59 0.21 bq = ± .
MJ de Haij. Evaluation of a new trigger function for cumulus convection
KNMI number: TR-276, Year: 2005, Pages: 117