Documentation of the Changes in INT2LM
Version 2.08

30.10.2020

Contents:

  1. Renamed S_ORO to S_ORO_MAX
  2. Corrections for computing difference between initial date and boundary start date
  3. Corrections for using HadGEM data and new vertical interpolation
  4. Technical Changes and Bug Fixes
  5. Changes in the Namelists
  6. Changes of the Results

1. Renamed S_ORO to S_ORO_MAX

(by Ulrich Schättler)

The new hydrology scheme with the modified ground water runoff now has been implemented to the official COSMO code. In this connection an official GRIB code entry has been defined for the slope of the orography and an official shortname has been given, which now is S_ORO_MAX. Therefore the name S_ORO, used up to now in INT2LM, has been changed to S_ORO_MAX.

Back to Contents


2. Corrections for computing difference between initial date and boundary start date

(by Denis Blinov)

There were problems when running INT2LM with driving data that have an older reference date than the actual start date, if the difference between these dates is less than an hour (e.g. 15 minutes).

In src_read_coarse_grid.f90, the difference in minutes is calculated by the routine diff_minutes. Then the difference is computed in units of 15 minutes (shortest time interval that could be used in practice in GRIB1) with the command:

ndiff_ini_bd = - NINT( (imindif * 60 + (ibdsec - inisec)) / dt )
but a difference in seconds will be "deleted" by the integer arithmetic used. So it is sufficient to have
ndiff_ini_bd = - NINT( (imindif * 60) / dt ).

Then the dubious correction was added "to account for seconds"
ndiff_ini_bdref = NINT((inimin * 60) / dt)
ndiff_ini_bd = ndiff_ini_bd - ndiff_ini_bdref

Reason for that once was, to be able to have an offset between initial date and reference date of the boundary fields of less than 15 minutes. But the way it is done here is not correct:

For a reference date for the boundary fields, inimin is always 0, so ndiff_ini_bdref is 0, and ndiff_ini_bd is not changed. But for a shorter offset (e.g. only 15 minute), wrong values are computed here. These two lines just have to be removed.

Note, that this kind of counting in steps of 15 minutes does not allow to have an offset between initial date and reference date of the boundary fields of less than 15 minutes.

The same corrections are done in files src_icon_interpol.f90 and src_gme_interpol.f90.

Back to Contents


3. Corrections for using HadGEM data and new vertical interpolation

(by Burkhardt Rockel)

INT2LM crashed when using HadGEM global data as input in case of lcm2lm=.TRUE. and itype_profiles_vert_interp=2. The reason was that wrong values of two parameters for setting height levels have been used for vertical gradient computations in the call to vert_interp in src_vert_inter_lm.f90. This has been corrected now.

The resetting of itype_profiles_vert_interp to 1 for lcm2lm in src_namelists.f90 has been removed.

Back to Contents


4. Technical Changes and Bug Fixes

Back to Contents


5. Changes of the Namelists

NONE

Back to Contents


6. Changes of the Results

NONE. If not using affected namelist settings described above.

Back to Contents