WMOD00309 (2) - Laserfiche WebLink

Home Browse Search

2.1.2 Meteorological setting Four convective situations were proposed for one experiment: (a) all warm cloud with rain, (b) maritime (mixed phase precipitation process), (c) continental (all ice precipitation process), and (d) hail. The second experiment would be primarily to test the ability of a model to simu- late the evolution of the microphysical properties of a cloud with detailed and parameterized formulations. The emphasis would be on ice-phase microphysics and an orographic situation was recommended in 'fjlhich the flow field is dominated by forced rather than convective lifting. Flow fields will be provided either from field data or from the output of an appropriate model. The number of cases will be reduced in order to maintain a focus that will per- mit useful intercomparison of models in a few data sets. In addition, at least one case involving cloud seeding was recommended. 2.2 Sensitivity Tests Sensitivity tests provide information on the dependence of a model to changes in input values or in the ways in which the model is internally constructed. The sensitivity of a perfect model would mirror the sensitivity of the atmosphere to the change made. Two types of sensitivity studies were proposed: (a) fluid dynamics, and (b) topics related to the mathematical architecture of the models. In all these experiments open data sets were recommended. Comparison of model behaviours would be carried out at the workshop. 2.2.1 Fl ui d dynami c tests The first series of these experiments deal with the effect of the initialization technique on the outcome of the simulation. Many models of convective clouds impose a somewhat arbitrary energy perturbation to initiate convection. Experience has shown that the evolution of the simulated cloud is initially dependent on the perturbation. In some cases this memory of the perturbation remains throughout the simulation; in other cases, the characteristics of the initiating perturbation disappear. The purpose of this experiment, is to exa- mine the effects of the perturbations of various sizes. The sensitivity of a given model to a range of perturbations should be established. These sen- sitivities could then be used as a basis for inter-model comparisons. It would be particularly useful to examine the perfonmances of several models initialized in identical manners. Three situations were proposed: (a) initialization by boundary layer convergence (forced lifting, the assumed reference), (b) bubble in the form of energy or its equivalent (heat, moisture, momentum, etc.), and (c) a combination of (a) and (b) . The second experiment in this series deals with information that becomes available if a finer resolution grid were employed. Commonly, numerical models need large passive domains and have low resolution within the cloud itself. The entire width of the cloud may occupy at most five grid intervals. Finer resolu- tion (about 30 grid intervals) should permit greater fidelity in the model, but 4.