# Inverse Problems and Data Assimilation Wiki and International Community

## Get observation data

• Make sure you have access to the database (sky). (There should be a folder '.sky' in your home-directory with all the configurations.)
• Check what kind of observation data is provided in your 'obs'-directory (e.g. /e/uwork/USER/cosmo_letkf/obs). If possible set links to obtain missing but necessary (e.g. conventional) observation data.
• In order to obtain the rest of observation data (in particular the Radar data) modify the script 'get_obs_KENDA' (insert correct date/time and data type) and run it.

## switch on/off

open 'config_cycle.sh' and choose an appropriate value for 'RADAR = …' (e.g.: 'RADAR = 1')

• open 'namelist_letkf_template' and set 'use_radvel = 7'

## Modifications

### lmrun_sx9_de_template

open 'lmrun_sx9_de_template' and:

• change “llookup_mie” for:
  llookup_mie_glob
• make sure you have the following setting:
%fl%  ldebug_radsim = .false.,
%fl%  lout_geom = .true.,
%fl%  loutdbz = .false.,

%fl%  lfdbk_output  = .true.,

%fl%luse_radarfwo=.TRUE.,

### namelist_letkf_template

Open 'namelist_letkf_template' and check if in the section 'rescale ps observation error'

radvel_obs_error = 5

where '5' is just an example. Some other value might be more suitable.

Check whether the list of radar stations (at the end of file 'namelist_letkf_template') is up to date. Currently (17.12.2015) you should use

%fl%             'fof_radar_id-010089 #OPTIONAL#'
%fl%             'fof_radar_id-010950 #OPTIONAL#'

### superobing or thinning

(Remark: if we don't use superobing or thinning we probably run out of memory)

### What might be useful

• You might want to switch of the adaptive localisation. Therefore open 'namelist_letkf_template' and set
adap_loc  = F
• If you only want to use the data of one specific radar station for assimilation, uncomment the others in the list above.
• If you want to use conventional data only passively open 'namelist_letkf_template' and set
&REPORT check='NONE'       use='passive'   / ! active or passive use of conventional data
• Using the parameters 'par_trans' and 'par_fce' (in 'namelist_letkf_template') you can decide which variables are going to be transformed by LETKF and which are not (and hence taken from the forecast ensemble). The default for 'par_trans' is 'pp t q u v w qcl qci'. Make sure that opposites of par_trans are found in par_fce. E.g., if you don't want to have a modification of humidity variables by LETKF, open 'namelist_letkf_template' and insert
par_trans = '- q - qcl - qci'
par_fce   = 'q qcl qci'