General description of the HARMONIE-AROME model

HARMONIE-AROME

The non-hydrostatic convection-permitting HARMONIE-AROME model is developed in a code cooperation with Météo-France and ALADIN, and builds upon model components that have largely initially been developed in these two communities. The forecast model and analysis of HARMONIE-AROME are originally based on the AROME-France model from Météo-France (Seity et al, 2011, Brousseau et al, 2011) , but differ from the AROME-France configuration in various respects. A detailed description of the HARMONIE-AROME forecast model setup and its similarities and differences with respect tot AROME-France can be found in <Bengtsson et al. 2017>. 

 

Model equations and discretization

The HARMONIE-ARMOME model is a non-hydrostatic spectral model, of which the dynamical core (developed by ALADIN) is based on a two-time level semi-implicit Semi-Lagrangian discretisation and in this presentation by P. Bénard). The main differences with the AROME-France dynamics and nesting setup are:
- HARMONIE-AROME is generally nested in the ECMWF global model rather than in Arpege,
- uses the SETTLS 2-time level scheme (Hortal 2002), and
- applies vertical nesting through Davies relaxation.
Full details of the HARMONI-ARMOME dynamics configuration can be found in Bentsson et al. 2017. In practice normally a rotated lat-lon grid projection is adopted by HARMONIE for the output grid.

 

Physical parametrizations

A variety of sub-gridscale physical processes are taken into account by parametrization schemes. Some of the parametrizations adopted by default in HARMONIE-ARMOME are the same as those used in the AROME-France model (Y. Seity et al., 2011), developed by the meso-NH community. , particularly the radiation scheme by <Morcrette> and most of the ICE-3 microphysics package (see the Meso-NH scientific documentation Physics ). One deviation from the Morcrette radiation scheme as described in Siety et al. 2011, is that the cloud liquid optical properties scheme by Nielsen et al. 2014 is used. For the parametrization of shallow convection, the eddy-diffusivity mass-flux scheme EDMF-M developed by De Rooij and Siebesma (2010) and De Rooij et al. (2013) is applied. For turbulence, the HARATU TKE scheme is adopted (<Lenderink and Holtslag 2004>, <De Rooij 2014>). Various adaptiations have been introduced to the original ICE3 microphysics: the OCND2 scheme (Muller et al. 2016) for improving model performance under cold conditions, and the Kogan autoconversion scheme (Khairoutdinov and Kogan, 2000). There is no separate paramitriation for gravity wave drag. A mean and sub-grid-scale orograhphic parametrezation for radiation is available (<ref>). All of these schemes are described in more detail in <Bengtsson et al. 2017>.

The following table gives a short overview of the upper air physics parameterizations used operationally and of additional options implemented so far:

Radiation scheme: Default: The Morcrette scheme with the Nielsen et al. cloud liquid optical properties scheme.
Optionally: ACRANEB scheme (ALARO) Saavijarvi scheme (Undèn et al, 2002)
Shallow convection:

Default: A combined eddy diffusivity- mass-flux scheme for shallow convection (EDMF-M, De Rooy and Siebesma, 2010, De Rooij 2014)

Optionally: Eddy diffusivity mass flux scheme EDKF (Pergaud et al, 2009)

   
Microphysics: The ICE3 package is used for the description of the microphysics of warm clouds, of atmospheric ice and sub-grid condensation, in combination with the OCND2 scheme (Muller et al. 2016) and Kogan autoconversion (Khairoutdinov and Kogan 2000), see Meso-NH Scientific Documentation on Physics
Turbulence:

Default: HARATU scheme (<Lenderink and Holtslag, De Rooij and De Vries 2017>)
Optionally: 1D prognostic Cuxart-Bougeault TKE scheme (Cuxart et al. 2000)

 

Surface and soil processes

The externalized surface scheme SURFEX (see SURFEX scientific documentation) is a set of models used for the description of the different types of surfaces: soil, sea and inland water bodies, and urban environments. It assumes a tile approach, distinguishing different surface types. Surface physiographies are prescribed using the 1km resolution ECOCLIMAP database. SURFEX consists of a number of components, which are all described in the SURFEX scientific documentation:

  • Soil:
    • a three-layer force-restore ISBA scheme (Noilhan and Planton, 1989); in the future, this may be replaced by an x-layer diffusion soil scheme ISBA-DIF, which has become available in <Surfex-v8>
    • A 1D high resolution column model CANOPY to describe the boundary layer with the canopy (Masson 2008);
    • For snow and vegetation the Multiple Energy Budget scheme by Gollvik (Gollvik 2002, Gollvik 2004);
    • Several parameterizations are available for snow
  • Sea surface fluxes (ECUME empirical formulae see SURFEX scientific documentation
  • A sea ice scheme (SICE, Batrak 2016)  
  • Urban areas: the Town Energy Budget (TEB) urban canyon model (Masson 2000)
  • Lakes: the Flake model (Mironov 2008, Kourzeneva 2008), including parametrizations to describe snow on lake ice (Semmler et al, 2012)

 

Data assimilation

The default upper air data assimilation scheme in HARMONIE is the 3DVAR scheme developed in ALADIN (Brousseau et al, 2011). Background error statistics are be calculated using the NMC method. An analytical balance condition is applied.
The observation data types which are assimilated by default presently are conventional observations (TEMP, SYNOP, AIREP, PILOT, SATOB, SHIP, DRIBU) and AMSU-A / ATOVS radiances over sea. Additionally, it is possible to assimilate AMSU-A over land and sea ice, AMSU-B, IASI, geostationaryatmospheric motion vectors,  MODIS aircraft observations, SEVIRI cloud-cleared radiances, GNSS zenith total delay, wind profilers, radar radial winds and profiles and scatterometer data. The use of cloudy radiances is under investigation.

Variational bias correction is applied by default to all satellite data. Observation screening involves logical and representivity checks, background quality checks, black-or whitelisting, multi-level and station level checks, redundancy checks and moving platform checks. Thinning is used to reduce the amount of observations.

Optionally, 3D-VAR with FGAT can be applied for upper air data assimilation. A 4D-VAR scheme is under construction and far advanced (Gustafsson et al. 2017, in preparation). Two hybrid ensemble assimilation techniques are under development: EDA (Raynaud et al, 2012) and LETKF (Bojarova et al, 2011).

Analysis of surface variables is done within the spatial interpolation tool CANARI, applying optimum interpolation for the assimilation of screen level parameters T2m and RH2m, and sea surface temperature. This is to be extended with snow depth. Assimilation of other soil variables (soil moisture, leaf area index, etc) will be done in an extended Kalman filter which is expected to replace the OI scheme in Cy43h2.

Initial and boundary conditions

To reduce noise and spinup, analyses are initialized out by incremental digital filter initialisation (DFI), (Lynch et al. 1997). In HARMONIE-AROME, initial and boundary conditions are normally taken from a global model (ECMWF). Lateral boundaries are overspecified, all variables being externally prescribed by the nesting model. Normally a relaxation zone of 10 grid points is adopted. Boundary relaxation is performed after the horizontal diffusion. At the upper boundary, Davies relaxation is imposed.

The HARMONIE script system

To allow the model to be used for routine operational numerical weather forecasting, the model analysis and forecast code has been embedded in a system of scripts, executables, support libraries, documentation and tools. This overall HARMONIE system must be applicable in all HIRLAM institutes for both operational and research applications. As such, portability of the code and tools included is an important issue.
There is a standard version of HARMONIE, which is referred to as the Reference System. This Reference system (which consists of code of the HARMONIE-ARMOME Reference Model configuratuin and the scripts, libraries and tools used to build, run, visualize and verify the model) is maintained on the HIRLAM server and at ECMWF. More information on the status of the HARMONIE system is given in the system wiki (only for registered users).

To install HARMONIE-AROME and run it with the HARMONIE script system on a local computer, a copy of the Reference system should be obtained. Instructions on how to do this are given in the HIRLAM system wiki (only accessible for registered users).

 ...............................................

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a non-hydrostatic spectral model, of which the dynamical core (developed by ALADIN) is based on a two-time level semi-implicit Semi-Lagrangian discretization of the fully elastic equations, using a hybrid coordinate in the vertical (see description of the ARMOME model in Seity et al, 2011 an d this presentation by P. Bénard). The main differences with the AROME-France dynamics and nesting setup are: HARMONIE-AROME is generally nested in the ECMWF global model rather than in Arpege, uses the SETTLS 2-time level scheme (Hortal 2002), and appies vertical nesting through Davies relaxation. Full details of the HARMONI-ARMOME dynamics configuration can be found in Bentsson et al. 2017. In practice normally a rotated lat-lon grid projection is adopted by HARMONIE for the output grid.

 

 

...At the default horizontal resolutions <= 2.5 km, the forecast model and analysis system are basically those of the AROME model from Météo-France (Seity et al, 2011, Brousseau et al, 2011). At coarser resolutions the ALARO or ECMWF physics parameterisations can be used, and/or the hydrostatic dynamics of ALADIN. ...

    of the fully elastic equations, using a hybrid coordinate in the vertical (see description of the AROME model in Seity et al, 2011) . Optionally, for larger domains and coarser resolutions the hydrostatic version of this semi-Lagrangian scheme can be used. An Eulerian dynamics core is available, but has been little used in recent years.

A description of the model equations are also presented in this presentation by P. Bénard.

In practice normally a rotated lat-lon grid projection is adopted by HARMONIE for the output grid.

 a Extensive scientific documentation is available from the Meso-NH scientific documentation on the upper air physics and on the surface module SURFEX.   

 

Deep convection (inactive now):  Mass-flux convection scheme with a moist convergence closure (Bechtold et al, 2001)  

 

There is no separate parametrization for gravity wave drag.

Optionally three other upper air physics packages can be used:

 

 oude tekst: bij sea surface fluxes (paragraaf surface and soil processes: (HIGHTSI, Cheng et al, 2003, Cheng et al, 2008, and GELATO, GELATO information on CNRM/GAME website)