Documentation of the Changes in the COSMO-Model
Version 4.20

31.08.2011

This version summarizes some "small" bug fixes and changes, but most of them also do change the results.

1. Technical changes regarding unified COSMO-ICON physics

The organizational structure of calling the different schemes within the turbulence has been modified with regard to the future changes for the unified COSMO-ICON physics.

The most important changes are:

Introduced new utility module: turbulence_utilities.f90
contains subroutines: canopy_source, diag_level, init_volume_canopy, init_surface_canopy, stab_funct, turb_param
Introduced new interface module turbulence_interface.f90
contains subroutine organize_turbulence. This subroutine is called from organize_physics and contains the part IF (ltur) from there. In addition it does some presettings for TKE time stepping.
Introduced new module turbulence_tran.f90
contains the subroutine turbtran. This is the former include file turbtran.incf, now as an extra module. The subroutine is called with a bigger argument list, where all fields and some control variables are passed (as is required by the COSMO-ICON physics).
Introduced new module turbulence_diff.f90
contains the subroutine turbdiff. This is the former include file turbdiff.incf, now as an extra module. The subroutine is called with a bigger argument list, where all fields and some control variables are passed (as is required by the COSMO-ICON physics).
Moved some global data definitions to module data_turbulence.f90.
organize_physics.f90:
- Introduced call to new subroutine organize_turbulence which organizes the calls to the packages for turbulence parameterization.
- Re-organized 'init'-phase for the turbulence (now calls init_volume_canopy and init_surface_canopy.
The former module src_turbdiff.f90 now is obsolete.

Back to Contents

2. Digital Filter Initialization

(by Uli Schättler)

The backward-forward filtering (ndfi=2) has been disabled for the Runge-Kutta scheme. Because of the nature of this scheme, an adiabatic backward integration is not possible.

Back to Contents

3. Changes in the dynamics

Runge-Kutta substeps:
The sub-stepping in the Runge-Kutta scheme has been modified, so that an additional small step is done in the third Runge-Kutta step. Now, always the 2/3/6 or 4/6/12 ... small steps are done for the Runge-Kutta steps. This is slightly more stable but will also increase the run time a bit.
The modification has been implemented with an internal switch in the subroutine calc_small_timestep: icalc_version
- icalc_version = 0: use new method (this is now the default)
- icalc_version = 1: do the same as before
src_slow_tendencies_rk.f90, slow_tendencies.f90:
Limit the horizontal velocity to a maximal value of vel_min=0.01, to prevent 0 heat/moisture fluxes at 0 windspeed.

Back to Contents

4. Changes in the physics

Radiation: (by Jürgen Helmert)
Bug fix in sub grid scale cloudiness.
Microphysics: (by Axel Seifert, Thorsten Reinhardt)
Correction of size distribution for rain droplets (in hydci_pp, hydci_pp_gr). Generalization of implementation for arbitrary mu_rain. (by Axel Seifert)
Bugfix in density correction of fall speeds (by Thorsten Reinhardt). Added missing factor zrho1o2 for consideration of density in the sedimentation of rain and snow (in hydci_pp).

Turbulence: (by Matthias Raschendorfer)
Additional 3D arrays for additional TKE sources by the action of other sub grid scale flow patterns have been introduced:

tket_conv: TKE-tendency due to convective buoyancy
tket_hshr: TKE-tendency due to (seperated) horizontal shear
tket_sso: TKE-tendency due to SSO wake production

These fields are allocated in the long term storage (by src_allocation.f90) To consider the convective buoyancy production for TKE, a new namelist switch has been introduced in /PHYCTL/ (similar to ltkesso):

Group	Name	Meaning	Default
`/PHYCTL/`	`ltkecon`	to consider the convective buoyancy production for TKE	`.FALSE.`

The reference pressure p0ref, which is used in the Exner function, has been declared in data_constants.f90.

Convection: (by Matthias Raschendorfer)
- src_conv_shallow: Introducing calculation of convective buoyant TKE production tket_conv
- src_conv_tiedtke: Using tke-field only, if 'lctke=T'.
  Introducing calculation of convective buoyant TKE production tket_conv
Soil Model: (by Jürgen Helmert)
Removed the dependency on diagnostic quantity t_2m by replacing it with the lowest atmospheric temperature level (as is in GME)

Back to Contents

5. Idealized Test Cases

(by Uli Blahak)

Additional bug fixes and modifications
Possibility to specify theta instead of t in radiosondes files
Improved error messages for Namelist-errors by using IOMSG-mechanism (can be used by defining #define HAS_IOMSG, if the compiler supports it.

Back to Contents

6. Interfaces for coupling with OASIS

(by Edouard Davin, Eric Maisonnave, Andy Dobler, et al.)

An interface to the OASIS coupler has been implemented into the COSMO-Model. In the CLM community there have been already activities to couple the Community Land Model (CLM) and an ocean model to the COSMO-Model using such an interface.

The interface is implemented using conditional compilation with -DCOUP_OAS. Sources from applications that should be coupled with the COSMO-Model or the coupler itself are NOT part of the COSMO-Model package. Such sources have to be obtained elsewhere. The COSMO-Model only provides the interfaces.

Necessary changes in the COSMO-Model:

lmorg.f90 Call to OASIS routines for initialization (at the beginning) and for update (during the timeloop).
src_setup.f90 Added icomm_world to the argument list of subroutine init_procgrid (up to now this routine started with MPI_COMM_WORLD)
environment.f90 Changed definition of icomm_world: in case of coupling, this communicator is defined by OASIS; only in case of un-coupled runs, it is MPI_COMM_WORLD. Therefore some uses of MPI_COMM_WORLD have to be replaced by icomm_world.

Back to Contents

7. Bug Fixes

Usage of tgrlat: (reported by Andreas Will and Jack Ogaja)
tgrlat (tangens of transformed latitude) has been computed only on u-latitudes (same latitudes as mass grid points). But it has also to be considered on v-latitudes, which are different from the u-latitudes because of the staggering.
Modified modules: data_fields.f90, numeric_utilities.f90, src_2timelevel.f90, src_allocation.f90, src_advection_rk.f90, src_setup.f90
src_runge_kutta.f90 (reported by Oli Fuhrer)
The variable govcp was initialized only for itheta_adv >= 1, but was used in some idealized tests also for other cases. Therefore initialize this (and some other variables) in any case.
src_radiation.f90: (reported by Thorsten Reinhardt)
In Section 6: Heating rates and radiation budget at surface: Variables dp0 and zsmu0 must not be referenced with js, but with loop variable jz1
organize_satellites.f90: (reported by Oli Fuhrer)
The variable nsat_steps had only been allocated, if output of synthetic satellite images was switched on. But in the init- and compute-phase of organize_satellites it was accessed in all cases. Therefore allocate this variable also in all cases.
src_lheat_nudge.f90: (by Klaus Stephan)
Bug correction: To consider both, blacklist flag and bright band flag for calculation of pr_obs and the weight, it should be an .AND. instead of .OR. (around line number 2135).

Back to Contents

8. Some technical modifications

mpe_io.f90, dummy_mpi.f90
Modifications and extensions because of some compiler problems.
meteo_utilities.f90
Replaced arguments p0, pp of subroutine cloud_diag with full pressure p_tot.
organize_data.f90, src_output.f90
Replaced variable name namelist (which is a Fortran Keyword) by outblock.
src_allocation.f90
Introduction of additional 3D-arrays tket_(conv, sso, hshr) for additional TKE source by the action of other sub grid scale flow patterns.
Removed yaction=canopy fully and move allocation of canopy fields to init phase of organize_physics.
src_radiation.f90
Replaced arguments p0, pp of subroutine cloud_diag with full pressure p_tot.
src_setup.f90
Introduction of ifndef for SCLM. Introduction of p0ref and presetting of kcm
src_setup_vartab.f90
Added instantaneous values of swdir_s, swdifd_s, swdifu_s, lwd_s, lwu_s
src_soil.f90
Warning if SCLM mode is used with surface flux forcing (with ifdef)
src_turbulence.f90
Eliminated unnecessary MAX-construct for index computations

Back to Contents

9. Changes of Results

Most of the above modifications do change the results slightly. But no significant changes are expected.

The run times of the very high resolution runs (2-3 km) are increasing by about 2-4 percent. This is mainly (but not only) due to the additional small time step in the 3rd Runge-Kutta step.

Back to Contents