LES cold-air outbreak Case


The proposed cold-air outbreak case is a quasi-lagrangian large eddy simulation (LES), in which domain is advected south-south east from 66 lat -11 lon to 60 lat -8.7 lon over the warmer seas to simulate the Sc to Cu transition.

Initial conditions for the case are based on high resolution limited area model (LAM) simulations performed with the Met Office Unified Model (UM). These simulations have been undertaken by Paul Field and are described in Pan-GASS constrain presentation .

The case is run for 14.5 hours, in which the first 1.5 hours represents a spin-up using a fixed surface condition. Following "spin-up", the surface is forced using prescribed time varying sea surface temperature (SST). The start time and date for all simulations is 0Z on January 31st 2010.

The change in SST represents the advection of the domain over warmer water. This is used as the main set-up as this permits a feedback depending on the atmospheric thermodynamics. (See end of page for other options)

The case includes interactive radiation, forcing of large scale vertical velocity to simulate subsidence and geostrophic forcing of horizontal winds. It is recommended that the simulation be performed with no ice microphysics and where possible ice microphysics.

CONTROL set-up


The grid set-up for the CONTROL case is as follows:

********Example code for vertical grid stretching******

dz_stretched = dz ! dz at 3000 m

alpha = 1.02

while z_les(k-1) lt z(Nz-1) do begin

dz_stretched = dz_stretched * alpha

z_les(k) = z_les (k-1) + dz_stretched




As alpha = 1.02 this results in 47 extra levels above 3000 m, so a total of 167 levels.

Both the horizontal resolution and vertical resolution are quite coarse, which is a trade off to permit the large domain.

Initial conditions

The initial conditions for the case are the mean profiles for 66 lat -11 lon derived from high resolution UM LAM simulation "u_66_11" (see Field et al, 2012, for description of LAM simulation).

Initial thermodynamic profiles are the adjusted liquid water potential temperature (&thetal_adj) and total water (qt_adj). &thetal is derived using the potential temperature and the cloud and ice water from the u_66_11 results, i.e. the beginning of a lagrangian track from that simulation. qt_adj is calculated by summing the vapour (qv_adj), cloud and ice water. The moisture variables are also from u_66_11 at the beginning of the lagrangian track, however, qv_adj is qv that is modified so that relative humidity is 100 % in the cloud region (the cloud region from the LAM simulations were close to 100 %, i.e. around 95 to 97 % RH w.r.t water). Thus, the initial profile is supersaturated, so cloud forms on the first time step of the simulation.

The initial horizontal wind components are also based mean wind fields derived from u_66_11 at the beginning of the lagrangian track. The wind fields are characterised by a strong N-S component (12 to 17 m s-1 in the boundary layer) and a substantially weaker E-W component (3 to 4 m s-1).

The thermodynamic and wind profile data (text format) can be downloaded from:

The initial thermodynamic and wind profile data is also available in netcdf format in the following file:

In constrain_setup_forcing.nc initial profiles are the first timestep of U, V, qt and theta_l. Also, note that these profiles contain 70 levels (up to 200 mbar). The first 37 levels represent the recommended vertical domain for this case. Finally, note that U, V, qt and theta_l contain 14.5 hours of direct data from the LAM simulation and so does not include constant conditions for the first 1.5 hours as in with the forcing (see later). Thus, this data is just provided to give a basic idea of the results from the LAM simulation.

The simulations are also all initialised with the following standard settings:


In order for the simulation to represent the advection of the air over warmer water, the surface is forced to simulate the warming sea surface temperature (or the increasing surface fluxes). Below is the SST forcing (text format):

The case is also forced with large scale vertical velocity forcing derived from the u_66_11 LAM simulation:

For convenience the time varying sensible and latent surface heat fluxes (SHF and LHF, respectively), time-varying sea surface temperatures (SST) and vertical velocity forcing (wsubs) are also supplied in netcdf format:

Note that the SHF, LHF, SST and wsubs in the netcdf file include the constant conditions for the first 1.5 hours for spin-up.

Finally, the N-S horizontal velocity (v) is forced using a geostrophic wind with the following values:

Ozone and Radiation profiles

Below is a file that contains the vapour and temperature field upto 37 km (derived from the the UM LAM simulation). This file also contains a standard McClatchy ozone profile which has been interpolated onto the LAM vertical grid. Note that the profile is set-up so that the first value is top-of-atmosphere.

Microphysics settings

One of the main areas for this case to understand is the role of ice microphysics in the evolution of the boundary layer during the cold air outbreak. In order to start to understand this we propose running at least two simulations of this case:

In both cases the cloud drop number concentration is set to 50 cm-3.

If institutions are not able to run their LES with ice microphysics, then just perform the simulation with warm micropphysics.

Sensitivity simulations

Resolution Testing - "Grey zone" experiments

To compare LES with the LAM output and understand the "grey zone", it is proposed that a series of horizontal resolution tests are undertaken with the LES, in which the horizontal domain of the CONTROL case the same as above is reduced. In addition to the CONTROL we ask participants to re-run the CONTROL at 4 coarser resolutions, i.e.

For all these extra simulations the vertical resolution kept the same as that in the standard simulation. As a minimum, in all these simulations microphysics should be set to warm. If participants have the time, could they also perform the simulations with ice microphysics switched on.

Finally, for institutions that have the computing power, it is proposed that they run extra simulations in which the horizontal resolution is increased. If provided these extra simulations will investigate whether 250m resolution is resolving the processes and therefore resolving the "grey zone". If necessary these tests can be performed on a domain (50 km by 50 km).

Surface forcing

In addition to the providing the prescribed SST, the prescribed surface fluxes are also provided. These are also derived from the high resolution LAM and they may be used to remove the uncertainty in the behaviour of the surface model, particularly when the horizontal resolution is reduced. If the participants have the computational resources and the time it would be very useful to receive results from the CONTROL with prescribed surface fluxes, in addition to the above simulations. The prescribed surface fluxes can either be downloaded from the link below or they are available in in the netcdf file.

Microphysics sensitivity

In addition to this standard simulation it would be interesting redo these simulations with initial cloud drop number concentration is set to 30 and/or 100 cm-3. At present, there are no observations of the cloud drop number concentrations in the Sc stage of the case, hence it would be useful to understand how sensitive the simulations are to this