Documentation of the Changes in INT2LM
Versions 1.9 / 1.10

After more than a year of stability, we had 2 new versions for the interpolation program INT2LM in the last months. The changes and new features are described below.

Contents:

1. Modifications in Versions 1.9 and 1.10
• New (alternative) Reference Atmosphere
• Processing Additional External Parameters
• Processing of Soil Humidity
• Vertical Interpolation of Humidity Variables
• Processing of Unified Model Data
• Optional Interpolation of GME Sea Ice Variables
• Optional Computation of FLake Variables for a Coldstart
• Modifications in the Initialization of Perturbed Pressure
• Modifications in the Vertical Interpolation for llm2lm
• Decomposition Independent Computation of the Interpolation Weights
2. Changes in the Namelists
3. Changes of the Results

1. Modifications in Versions 1.9 and 1.10

New (alternative) Reference Atmosphere
(by Günther Zängl; Version 1.9)

A new (alternative) reference atmosphere has been introduced, which is based on the temperature profile

Thus, the reference atmosphere approaches an isothermal profile in the stratosphere, whereas the existing reference profile has an increasingly negative vertical temperature gradient in the stratosphere. The vertical extent of the model domain is no longer limited with the new reference atmosphere.

To choose the reference atmosphere, a new Namelist switch irefatm has been introduced:

1. old reference atmosphere with additional specifications p0sl, t0sl, dt0lp, vcflat
2. new reference atmosphere with additional specifications p0sl, t0sl, dt0lp, vcflat, delta_t, h_scal where
   • delta_t: temperature difference between sea level and stratospere
     (reasonable values are between 50.0 and 100.0 Kelvin)
   • h_scal: scale height (reasonable values are between 7000.0 and 12000.0 meters)
   are also 2 new Namelist variables.

All 3 new Namelist variables are in the group /LMGRID/.

Also the input grid can have the new reference atmosphere (if input data are coming from the COSMO-Model). New variables irefatm_in, delta_t_in and h_scal_in are used for that. But these variables are no Namelist variables, because they are read from the header information (vertical coordinate parameters) of the input data.

NOTE:
Reading of the vertical coordinate parameters of the input model (and therefore also computation of the input reference atmosphere) was moved from routine external_data to the module src_read_coarse_grid, because only then real atmospheric fields are read, that definitely contain the vertical grid information. Twodimensional fields, that are read in external_data, need not carry these informations.

The CLM Community implemented the new reference atmosphere also for the NetCDF I/O.

Back to Contents

Processing Additional External Parameters

New external parameters are provided for the COSMO-Model, which can be optionally read from the external parameter file, if present. Please note, that these fields are not yet distributed regularly.

• New field EMIS_RAD for the radiation: Thermal radiative surface emissivity. To read this field, a new Namelist parameter lemiss has been introduced in group /CONTRL/ (Default: .FALSE.).
• New field PRS_MIN for the soil model: Minimum stomata resistance of plants. To read this field, a new Namelist parameter lstomata has been introduced in group /CONTRL/ (Default: .FALSE.).
• For plant characteristics:
  Further options have been implemented to compute the fields plcov_lm, lai_lm (and z0_lm). This is controlled by the Namelist parameter itype_ndvi:
  0. Default and existing option:
     plcov_lm and lai_lm are computed using max (vegetation) and min (rest) values.
  1. Up to now this option was used to read GME external parameters for the ratio of monthly mean of normalized differential vegetation indices to the annual maximum. This has been eliminated.
     Now the corresponding external parameters are provided on the COSMO-Model grid and these are used instead to compute plcov_lm and lai_lm. The fields are NDVI_MRAT (provided by DWD in Grib).
  2. This has been introduced by the CLM Community:
     12 monthly climatological mean values are read for plcov, lai and z0 and actual values for the given day are computed by linear interpolation. The fields read are plcov12_lm, lai12_lm, z012_lm (provided by the CLM Community in NetCDF).
• For aerosols that can be used in the radiation scheme:
  Up to now the radiation scheme in the COSMO-Model uses constant aerosol values for desert, sea, land and urban areas. Now, monthly mean of the following aerosol types can be read from the external parameters:
  • AER_SO4: Tegen (1997) aerosol type sulfate drops
  • AER_DUST: Tegen (1997) aerosol type mineral dust
  • AER_ORG: Tegen (1997) aerosol type organic
  • AER_BC: Tegen (1997) aerosol type black carbon
  • AER_SS: Tegen (1997) aerosol type sea salt
  Which aerosol scheme is used can be controlled by the new Namelist Switch itype_aerosol:
  1. Default and present option: no external parameters are read; the COSMO-Model uses constant values.
  2. 12 monthly mean values for the above fields are read and the INT2LM computes an actual value for the given day and gives that to the COSMO-Model for further processing.
• For topographic corrections in the radiation, NL Switch lradtopo:
  This has been implemented before, but the field HORIZON is now (really) treated as a 3D variable. In Grib it is coded with leveltyp 110 and the levels 1 ... nhori. nhori is a new Namelist parameter in group /CONTRL/ (Default: 24)
  (added by MCH)
• For the FLake Model, NL switch llake: the mask fr_lake_lm is already read, but also the depth of the lakes is needed. This is now also read (DEPTH_LK), if llake = .TRUE. (added by CLM)

Back to Contents

Processing of Soil Humidity

An optional interpolation of soil humidity has been introduced by MeteoSwiss following the introduction of the new hydrological IFS model on Nov. 6, 2007. This option is also used for other input models.

It can be controlled by the new Namelist Switch: l_smi (default: .FALSE.) in the group /CONTRL/.

2 new options have been implemented by the CLM Community:

3. The soil moisture profile in relation to the pore volume is kept constant below the deepest layer of the input soil model.
4. The soil moisture profile related to field capacity is kept constant there.

Back to Contents

Vertical Interpolation of Humidity Variables

An optional treatment of the humidity has been implemented by MeteoSwiss. Per default, cloud water and specific humidity are added to get the relative humidity, which is used for vertical interpolation to COSMO levels. Afterwards, the relative humidity is splitted again to the two compontents. Cloud ice, however, is treated separately.

When setting the new Namelist switch lmixcld to .TRUE., (Default: .FALSE.), the general relative humidity is computed as the sum of specific humidity, cloud water and cloud ice. After vertical interpolation, these three components are derived again.

NOTE:
This new treatment has not been fully implemented in the official version. It calls a routine moist_split, which is written only for one grid point. Therefore the corresponding loops do not vectorize on the NEC. The code lines calling this routine are commented, so that lmixcld=.TRUE. will not work without modifications to the Source Code!

Back to Contents

Processing of Unified Model Data

A new option has been implemented into the INT2LM, to interpolate the data from the Unified Model (from the MetOffice) to the COSMO-Model grid. Up to now only the atmospheric variables are treated. To choose this option, set lum2lm=.TRUE.

The Unified Model data are provided on an unrotated lat-lon horizontal grid and hybrid height levels in the vertical. They are treated similar to COSMO-Model data:

• Wind speeds u and v are horizontally interpolated from the unrotated lat-lon grid to the rotated COSMO-grid. Then they are interpolated vertically.
• Temperature t is also interpolated horizontally and vertically similar to the option llm2lm.
• Full pressure is interpolated horizontally, but for the vertical interpolation the deviation from a reference pressure is used. The reference pressure is calculated on the levels of the Unified Model data as on the COSMO-Model levels.
• The humidity variables are treated in the same way as in the COSMO-Model.

NOTE:
For this option, the Unified Model data have to be provided in terms of DWD grib code regarding element numbers and table codes. For NWP, the UM data is usually provided in the british PP-format. DWD provides a transformation program that converts the PP-format to (DWD) grib code.

Back to Contents

Optional Interpolation of GME Sea Ice Variables

An option to interpolate the variables t_ice and h_ice from the GME sea ice model to the COSMO-grid has been implemented. To switch on this option, set lseaice=.TRUE. (Default: .FALSE.).

Back to Contents

Optional Computation of FLake Variables for a Coldstart

An option to initialize the prognostic variables of the FLake model has been implemented. To switch on this option, set llake_coldstart=.TRUE. (Default: .FALSE.).

NOTE:
This option should be used with great care: For case studies it is of NO sense to run the lake model, because it takes several weeks until the lake variables have reached a useful content. Therefore it is NOT recommended to use the FLake model in case studies.

Back to Contents

Modifications in the Initialization of Perturbed Pressure

The recalculation of a hydrostatic balanced pressure deviation in the module src_read_coarse_grid.f90 (Section 6) has been improved. This changes the results in case of llm2lm and lbalance_pp.

Back to Contents

Modifications in the Vertical Interpolation for llm2lm

There were bugs in the vertical interpolation in the lm2lm-case, which were corrected by Anne Roches, MCH:

• The use of kgr in the computation of the linear regression was wrong: kgr is the index of the PBL height, refering to the input grid, i.e. k counting from top to bottom. But it was used in the variables for the intermediate profiles, which serve as input for the spline interpolation, which are counting k from bottom to top!
• The linear regression is now done only for temperature. The horizontal wind components are treated now as all other variables.
• The extrapolation of the input profile below the surface is now done only to 50 m below the surface of each grid point (Before, a minimum height over the whole domain was computed).

Back to Contents

Decomposition Independent Computation of the Interpolation Weights

There are reproducibility problems with the INT2LM, option llm2lm. One reason was, that the interpolation weights were computed for every coarse grid subdomain. And with differing subdomains also the interpolation weights could be different, because of numerical rounding errors. This has now been fixed by computing the interpolation weights always by using the information for the total coarse grid domain.

Another problem still remains for the soil variables that are horizontally interpolated using the "match" type of horizontal interpolation. If none of the surrounding grid points has the same land-sea characteristics as the point in the fine grid, a search in the whole coarse grid subdomain is done. Depending on the subdomains, the result of this search can be different.

There is no solution for that up to now!

Back to Contents

2. Changes of the Namelists

There were the following changes for the Namelist variables:

Back to Contents

3. Changes of the Results

If using the default settings of all the new Namelist switches, the results of the interpolation GME ⇒ COSMO are not changed at all.

For the interpolation COSMO ⇒ COSMO, the results are changed due to some changes of algorithms.

Back to Contents