Opened 8 years ago

Last modified 8 years ago

#89 accepted task

Minimise effect of convection on surface convergence

Reported by: raw88 Owned by: raw88
Priority: normal Milestone:
Component: Atmosphere Version:
Keywords: Cc:

Description

The focus here is on a case of quasi-stationary convection over the southwest peninsula of the UK, which was forced by a narrow convergence line. A sensitivity test is required to investigate to what extent the convergence is due to low-level hydrostatic pressure decreases associated with the column warming through latent heat release. An inital experiement will set the latent heat of condensation to be very small; however, this may have significant undesirable effects on other aspects of the simulation (e.g. surface heat budget). Alternative approaches may be required.

Change History (2)

comment:1 Changed 8 years ago by raw88

  • Owner set to raw88
  • Status changed from new to accepted

comment:2 Changed 8 years ago by raw88

As expected reducing the latent heat of condensation to a value of 1.0 J/kg had dramatic effects on the simulation, with increased low-level cloud, weakening of the synoptic scale low-pressure system and large changes in the surface heat budget. Clearly this approach does not make for a good sensitivity test and therefore it will be abandoned.

As an alternative I will now attempt to create a 'dry run', i.e. one in which the water vapour and liquid water quantities are close to zero throughout. While this will likely have equally significant effects on the large scale flow, the idea is to carry out this and other sensitivity tests on a smaller grid centred over the area of interest (Cornwall) nested within the UKV model but with the same resolution.

To obtain a dry run it will be necessary to do 2 things: (1) Set the initial moisture fields to zero, and (2) Set the lateral boundary moisture fields to zero. The former of these will be achieved by editting the control routine ATM_STEP.F90, so that the variables Q, QCL, QCF AND QRAIN are set to zero (or very small in the case of Q) at the first time step. The second code edit will be to the routine READ_ATMOS_LBCS.F90 which is in /src/atmosphere/lbc_input. Here each of the moisture fields will again be set (close) to zero. A third edit may be required to prevent surface evaporation, if this is able to provide sufficient mositure during the simulation for the formation of cloud; however, initially I will try just the 2 edits above.

Note: See TracTickets for help on using tickets.