The CONVECTion submodel calculates the
process of convection.
Via namelist one of the implemented
parametrisations can be chosen and the necessary calculations are
performed automatically. At the momentary state there are three similar
schemes available. Those all are based on the Tiedtke convection scheme
developed at ECWMF, but have different closures: the default value is
the Nordeng closure, a second option the original Tiedtke closure and a
third option, a hybrid closure.
All of the Tiedtke schemes
calculate after initialisation and determination of the cloud base a
first ascent with entrainment and detrainment in absence of downdrafts.
In a second step now the downdrafts and changes of temperature, humidity
and tracer concentrations due to these effects are estimated. As next
step the final ascent, now including the changes of the downdrafts, is
performed. Finally the fluxes are adjusted and the values of the larger
scaled fields get changed due to the convection.
Because for
chemistry it is very important that the tracer concentrations are
positive definite an update to the code by S.Brinkop from the DLR is
implemented that guarantees this condition, but at the momentry state
this only works for the Nordeng closure. The other closures as well as
other schemes that do not explicitly calculate tracer transport should
therefor better be used with the related
CVTRANS module. As the tracer fluxes can be
changed by wet deposition in the convective cloud and/or precipitation,
there is a link to the SCAVenging module.
All the necessary parameters can be easily stored and afterwards used by
other processes.
In addition to this there is a small routine that estimates a convective
cloud cover. The original Tiedtke scheme uses a fixed, grid size
independent fraction of 5%. This might be a good estimate, but is not
only dependent on the grid size, it also does not give any information
of the strength of the convective event. Therefor a new convective cloud
cover is estimated from the strength of the updraft mass flux.
In an experimental, not publically available version, more convection
schemes are implemented. These are:
The operational ECMWF convection
scheme (IFS-cycle 29r1)
The convection scheme of Zhang / Hack /
McFarlane as it is implemented in the CTM MATCH with two different
treatments of cloud water.
The convection scheme of Bechtold, that
is / was used at ECMWF and in the French MESO-NH.
The Zhang / Hack / McFarlane Scheme starts after initialisation with
calculation of the buoyancy strength by determining the convective
available potential energy (CAPE). In a second step the microphysical
part is calculated followed by the closure. At the end the changes of
temperature and humidity due to the convection are determined as well as
the rain parameters. This scheme does not change the winds in our
version and yields quite different temperatures and humidity fields
compared to the Tiedtke scheme, but the effect on climate calculations
has not been determined yet. The scheme is not implemented with its own
tracer transport mechanism, so the CVTRANS
module should be used for this task.
The Bechtold Scheme is also a
scheme following the mass flux approach. First it is determined by a
trigger function wether convection will take place in that column or
not. Afterwards the LCL for the column will be calculated and from this
level the updraft, downdraft and condensation. It is closed by a CAPE
closure.
The scheme has the capability of calculating ensembles for
the deep convection process, i.e. it can calculate the process of deep
convection with 3 additional perturbed starting conditions.