[University home]

Centre for Atmospheric Science

ACTIVE Project-Modelling

Output from a Hector simulation using the CRM model.
Output from a Hector simulation using the CRM model.

Large scale modelling: TOMCAT
p-TOMCAT is a new version of the three-dimensional global TOMCAT Chemical Transport Model (CTM). It has been parallelised using the MPI message-passing library and can be run on many tens of processors to achieve much improved performance. The model includes a detailed tropospheric chemistry scheme and includes photolysis, wet and dry deposition and a comprehensive set of emissions. The model also includes treatment of physical processes in the troposphere with parameterisations of shallow and deep convective transport, boundary layer mixing and vertical diffusion. The model has been verified by several comparisons against observations from established datasets and during measurements campaigns. Recently, the improvements to the performance of the model have enabled some preliminary integrations at very high resolutions (0.5°x0.5°) to be performed. The model will be used, along with trajectory studies (see next), to diagnose transport into and out of the TTL by a variety of processes. p-TOMCAT will provide information on the expected variability of the TTL and possible differences between the two campaign periods.

Trajectory modelling
The use of trajectory calculations based on 3-D winds from global meteorological datasets is well established as an approach to interpreting and complementing in-situ chemical data. Recent studies in the extratropical troposphere indicate very good consistency between calculated trajectories and structure observed in chemical fields, supporting the idea that the wind fields are sufficiently accurate for this purpose. Dynamical phenomena in the tropics are probably not so well resolved by the meteorological datasets and such trajectories cannot capture all aspects of non-convective transport. However agreement between structure measured in water vapour, for example, and calculated trajectories is often very good, suggesting that when used with due care, taking account of the known presence of convective systems, they can be a valuable complement to observations.

Large Eddy Modelling
A combination of the Met Office Cloud Resolving Model (CRM) and the Manchester explicit microphysics model (EMM) are being used to simulate individual storms. This approach has previously been used to interpret the EMERALD-II dataset and was successfully used by Phillips et al (2001) to model storms forming over New Mexico. This model provides the main theoretical tool to link the aerosol and microphysical measurements by the two aircraft.

The CRM includes parameterised microphysics with a double-moment scheme representing the number density and mixing ratios of the ice phase as described by Swann (1998). The Centre for Atmospheric Science will run this model for each of the observational case studies using aircraft and radiosonde ascent data as input. The cloud structure predicted by the model as the storm develops will be compared to the radar image, i.e. the model will predict the radar reflectivity. Sensitivity tests will be performed with the model, varying input parameters over the range of their uncertainty e.g. wind shear, temperature profile and ice nucleus concentration. This will enable us to understand the sensitivity of the model to each of these parameters and hence determine the reason for differences between the predictions and observations.

The 3-D cloud dynamics and cloud water content from the CRM simulations provide the input parameters for the EMM. This model treats the following microphysics explicitly within this dynamical framework:
a. The activation of cloud droplets from the aerosol entering the cloud.
b. The growth of raindrops by collision coalescence, the production of ice particles by primary nucleation.
c. The homogeneous nucleation of ice is treated at temperatures below –35ºC.
d. The growth of ice particles, of habit determined by temperature and supersaturation (including ventilation effects) aggregation and riming.
e. Secondary ice particle production by riming splintering, raindrop freezing and ice crystal evaporation.
f. The melting of precipitation below the freezing level is treated along with the recirculation of hydrometeors between neighbouring updraughts and downdraughts.

In this way the model is able to predict the microphysical properties of the anvil region for direct comparison with the in situ measurements. It is also able to predict the fluxes of water vapour, particles and passive tracers out of the cloud through the anvil region. The detailed microphysical model will also enable direct comparisons between the modelled evolution of the cloud and measurements of reflectivity from the radar both as the cloud develops and in its mature state.

Microphysics, Aerosol and Chemistry (MAC) model
The specific advantage of MAC in this project is that it simulates a size-resolved aerosol interactively with the cloud microphysics and dynamics, which is not done in the EMM. MAC currently uses a 2-D axisymmetric/slab-symmetric configuration for the dynamics with bin-resolved microphysics for aerosol, drops, ice, graupel and aggregates (Yin et al 2001, 2002, 2004). It also includes a kinetic treatment of gas scavenging and bin-resolved aqueous-phase chemistry. Therefore, size-resolved aerosol particles processed by clouds and detrained from the cloud outflow region can be simulated. Extension of the model to a 3-D version and to include cloud electricity is underway and will be complete before the field campaign. This model will be the main theoretical tool for investigating chemical transport and generation (NOx/O3) in the storms. It will also be used to interpret the black carbon measurements. For the lightning scheme, measurements from the Osaka University lightning interferometer at Darwin will be used to constrain the simulations.

Simulations with MAC can include the effects of variable aerosol composition and externally mixed aerosol distributions, which can influence cloud droplet spectra. On the other hand, the formation and diffusional growth of ice particles, and the production of ice by secondary processes such as the Hallett-Mossop process, are calculated in more detail in the EMM. Simulation of the same case studies using the two models, with regular intercomparisons, will enable us to reach more robust conclusions regarding the microphysical and dynamical response to aerosols.

In the second (monsoon) campaign a much greater variety of convective clouds were encountered than in the pre-Christmas campaign. It is not feasible, nor useful, to simulate them all - it is the ensemble properties of these clouds that will be of interest. The Dornier measurements provided the range of aerosol concentrations and types entering the storms (bearing in mind that Darwin, being a coastal location, is within reach of convection over land and ocean). Thermodynamic profiles are available from the network of radiosondes as well as the Dornier, Egrett and other TWP-ICE aircraft. These will be used as input to the model to examine the range of anvil properties (aerosol and cirrus and chemicals) that it produces; these will be compared with the observations and used to improve the model simulations.