Documentation of the Changes in the COSMO-Model
Version 4.22

31.01.2012

This version brings in the new modular nudging code for the data assimilation. And also the new zenith angle dependency of the solar radiation for every time step has been introduced.

Contents:

  1. New modular Nudging Code
  2. Zenith Angle Dependency
  3. Changes of Results
  4. Changes to the Namelists


1. New modular Nudging Code

(by Christoph Schraff)

The Nudging Code has been overhauled and has been more modularized:

Interdependency with LMSTAT

The new version will write VOF Version 3 instead of Version 1. If the program LMSTAT is used to read and complete the VOF (YUVERIF) files produced by the current COSMO version (for the purpose of the upper-air verification), then the version of LMSTAT must be at least V1_7 .

Back to Contents


2. Zenith Angle Dependency

(by Thorsten Reinhardt)

Introduce zenith angle dependency of solar radiation for each timestep. This can be switched on by an internal parameter l_zenith_update. Internal parameter means, that the change now is hard-coded, and if somebody wants to get the old behaviour, this internal parameter has to be changed and the code has to be recompiled.

Back to Contents


3. Changes of Results

The changes of the radiation changes the results.

Most code changes of the nudging do not change the results or change them only very slightly as long as the new options for the new balancing of near-surface wind, the weighting of multiple observations, or the use of new observation types are not used actively. Some changes will occur due to the modified station selection of surface pressure observations, and due to the bug correction at the rejection of aircraft reports very close to the ground. However, the impact by these changes on the results usually remain small too. On very rare occasions, very large improvements have been found due to a more successful rejection of data from a single buoy station where the observations gradually drift away from the 'truth'.

In the tropics near the equator, however, the results will be strongly changed due to the latitude dependent reduction of geostrophic wind correction (provided that this correction is used actively). This change can be regarded as a bug correction.

Back to Contents


4. Changes to the Namelists

There were several changes for the Namelist group NUDGING:

New Namelist Variables

Group Name Meaning Default
/NUDGING/ mveripr switch for writing VOF and/or feedobs files
  1. no file written, equivalent to lverif=.FALSE.
  2. NetCDF (feedobs/feedbk) file for LETKF or verification
  3. ASCII file VOF (YUVERIF)
  4. both NetCDF and ASCII VOF
3
mpsgcor switch for applying geostrophic pressure increments balancing near-surface wind analysis increments:
  1. no pressure correction
  2. correction. balacing wind from scatterometer only
  3. correction balacing scatt. + in-situ 10-m wind obs
1
qgeops reduction factor to the geostrophic pressure increments 0.9
lscatt for active use of observation type scatterometer .TRUE.
maxmlv max. number of observation levels in multi-level report
(The default value 100 is equal to the fixed value of the old version)
100
mxfrep max. number of reports in NetCDF feedobs file -1
mxfobs max. number of observations in NetCDF feedobs file
(if mxfrep, mxfobs ≤ 0 then reasonable values are computed dynamically from maxmlo, maxsgo, hverend etc.)
-1
lcd037 for active use of new observation code type TEMP MOBILE .TRUE.
lcd038 for active use of new observation code type PILOT MOBILE .TRUE.
lcd123 for active use of new observation code type scatterometer ASCAT .TRUE.
lcd122 for active use of new observation code type scatterometer OSCAT .TRUE.
igpscen GPS processing centres assigned as active obs code types, ordered according to their preference in the redundancy check
(The default of -1 means that no GPS processing centre is used actively)
-1
For new weighting of multiple observations and observation systems:
(default values equivalent to the hitherto used weighting)
nwtyp if > 1 then compute net observations increments for nwtyp different sets of observing systems (obs types or code types) separately 1
niwtyp number of observing systems for each set of obs systems 1,0,0,...
iwtyp defines these sets of observation systems 0,0,0,...
kwtyp defines the multiple weighting function for each of these sets 1

The default settings of the new namelist variables imply that the previous settings of the nudging scheme are used as long as there are no observation input files for scatterometer data (except that a NetCDF feedobs file is written additionally).

Other Changes to the Namelist Variables

Group Name Meaning Default
/NUDGING/ lseparw This switch has been eliminated.
kmultw This switch has been eliminated.
doromx Default values and meaning have been modified due to modified meaning of 'fdoro': 100.,150.,150.,150.; positive height differences (z-obs > z-model) between station height and model orography are scaled by a factor of 1/4 for station pressure and GPS IWV obs, since extrapolation is replaced by inter- polation in this case (i.e. max. height differences of 600 m are allowed)
This implies that if the values for doromx have been set explicitly in the namelist for the previous version, these values should be adapted with the new COSMO version! If the values have not been set in the namelist explicitly and the default values have been used, no changes are required.
100.,150.,150.,150.

Back to Contents