GitHub

Surface Data Assimilation Scheme: Canari

Introduction

(by Alena.Trojakova)

CANARI stands for Code for the Analysis Necessary for ARPEGE for its Rejects and its Initialization. It is software (part of IFS/ARPEGE source code) to produce an ARPEGE/ALADIN analysis based on optimum interpolation method. The number of ARPEGE/ALADIN configuration is 701. CANARI has the two main components the quality control and an analysis. According to the type of observations the analysis can be:

  • 3D multivariate for U, V, T, Ps
  • 3D univariate for RH
  • 2D univariate for 2m/10m fields
  • soil parameters analysis is based on 2m increments

CANARI can handle following 10 types of observations:

  • SYNOP: Ps, T2m, RH2m, 10m Wind, RR, Snow depth, SST
  • AIREP: P ( or Z), Wind, T
  • SATOB: P, Wind, T - from geostationary satellite imagery
  • DRIBU: Ps, T2m, 10m Wind, SST
  • TEMP: P, Wind, T, Q
  • PILOT: Wind with the corresponding Z, (sometimes 10m Wind)
  • SATEM: Q, T retrieved from radiances- surface

Applications

  • Diagpack - diagnostic of mesoscale features via detailed analysis of PBL using high resolution surface observation with specific tunings:
REF_S_T2  = 3.0,
REF_S_H2  = 0.20,
REF_A_H2  = 40000.,
REF_A_T2  = 40000.,
...
  • Quality control and verification - VERAL package
  • Surface analysis

Detailed description of the method and technical documentation can be found in the References below. Here follows basic input/output summary and command line arguments.

CANARI procedure

All optional items are controlled by various keys in namelist.

INPUTs

  • The first guess file

    ln -s guess ICMSHANALINIT
ln -s guess ELSCFANALALBC000

  • The observation database, which requires special variables to be exported

    export ODB_STATIC_LINKING=1
export TO_ODB_ECMWF=0
export ODB_CMA=ECMA    ... Database type (ECMA extended CCMA compressed)
export ODB_SRCPATH_ECMA=...
export ODB_DATAPATH_ECMA=...
export IOASSIGN=
export ODB_MERGEODB_DIRECT= ... optional direct ODB merge, If your ODB was not merged previously use  1

  • Concerning the observation use, another file is necessary, but it is without any interest for CANARI (just part of variational analysis code is not controlled by a logical keyt !) The file can be obtained on "tori" via gget var.misc.rszcoef_fmt.01.

    ln -s rszcoef_fmt var.misc.rszcoef_fmt.01

  • The climatological files

    ln  -s  climfile_${mm}  ICMSHANALCLIM
ln  -s  climfile_${mm2} ICMSHANALCLI2

  • The namelist file

    ln -s namelist fort.4

  • The ISBA files

    • file used to derive soil moisture from 2m increment

      ln -s POLYNOMES_ISBA fort.61

    • OPTIONAL assimilated increments files to smooth the fields (ICMSHANALLISSEF file is created at the and of analysis )

      ln -s increment_file ICMSHANALLISSE

    • OPTIONAL The SST file - an interpolated NCEP SST analysis on ARPEGE grid and stored in FA file format) used for relaxation towards "up-dated" climatology

      ln -s SST_file ICMSHANALSST

    • OPTIONAL The error statistic file; OI allows to know the variance of the analysis error, which can be used to improve background error next cycle => an option to use "dynamics" statistics instead of fixed. Output file ICMSHANALSTA2 is produced with statistics for the current run

      ln -s statistics_file ICMSHANALSTA

    • OPTIONAL The incremental mode files (global option only); it is possible to read 3 input file to build non-classical init; the combination is done on spectral fields only: G1=G0+A-G

      ln -s G0_file ICMSHANAINIT
ln -s A_file ICMSHANALANIN
ln -s G_file ICMSHANALFGIN

run CANARI

MASTERODB -c701 -vmeteo -maladin -eANAL -t1. -ft0 -aeul

  • -c: configuration number (CANARI = 701)
  • -v: version of the code (always "meteo" for ARPEGE/ALADIN)
  • -m: LAM or global model ("aladin" or "arpege")
  • -e: experiment name (ANAL for instance)
  • -t: time-step length (no matter for CANARI, usually "1.", avoid 0.)
  • -f: duration of the integration (t0 or h0 for CANARI)
  • -a: dynamical scheme (does not matter for CANARI Eulerian = eul or semi-Lagrangian = sli (sli as usual))

OUTPUTs

  • OPTIONAL The analysis file
    ICMSHANAL+0000
  • OPTIONALLY updated observational database
    • OPTIONAL The error statistics file
      ICMSHANALSTA2
    • OPTIONAL The increment file
      ICMSHANALLISSEF
    • The output listing - enables checking of various parameters, e.g. number observation of given type (SYNOP,TEMP,..), number of used observation parameters (T2m, RH2m, T, geop., ...), some namelist variables, various control prints (O-G and O-A statistics, ...), grid-point and spectral norms.

NODE*

Sample of script is attached.

As a part of the system training in Copenhagen in 2008, Roger prepared an intoduction to CANARI, which is found in HarmonieSystemTraining2008/Lecture/SurfaceAssimilation on hirlam.org

References