Forecast Settings

This page gives some details and advice on appropriate settings for the HARMONIE-AROME forecast

Microphysics

ICE-T

Switch ICE-T on by setting LICE_T=.TRUE. in harmonie_namelist.pm under &NAM_PARAM_ICEN in %arome. When using ICE-T (LICE_T), LOCND2 should be set to True, and LMODICEDEP preferably to False. LICE_T will override LKOGAN, so by default LKOGAN=F. Documentation: (Engdahl et al., 2020)

Description: ICE-T is a modified cloud microphysics scheme that builds upon ICE3 and OCN2D, with elements from the Thompson scheme from WRF. ICE-T was developed in cy40h1.1 for the purpose of better representation of supercooled liquid water, and downstream forecasts of atmospheric icing. The changes include stricter conditions for ice nucleation, less efficient collection of liquid water by snow and graupel, and variable rain size distribution depending on the source of the rain. (Rain originating from melting snow or graupel have larger drops, than rain originating from warm processes.)

Shallow Convection

LDUALMF activates (.TRUE.) or de-activates (.FALSE.) the DUAL (dry and moist) mass flux shallow convection parameterisation. Note that with LDUALMF=.FALSE. the mass flux activity as a source term for TKE in the turbulence scheme (energy cascade) will be also eliminated. Also the moist updraft transport contribution to the cloud scheme is eliminated with LDUALMF=.FALSE.. See for details of the convection scheme and links to the cloud and turbulence scheme: https://doi.org/10.5194/gmd-15-1513-2022.

The scale-aware convection scheme is activated by setting LSCAWAREMF=.TRUE.. Setting this reduces the dry and moist (if present) mass flux using a tangent function scaled with the dry boundary layer height $h$ for the dry updraft and sub-cloud height plus cloud layer depth $h+h_c$ for the moist updraft:

\[f = \tanh\left(1.86 \frac{\Delta x}{h+h_c}\right)\]

NOTE: this option can only be used when LDUALMF=.TRUE..

To support the model when it is trying to build up convection itself, the setting LWTHRESH=.TRUE. can be used. Depending on the gridsize, a vertical velocity threshold is defined. If the absolute value of the vertical velocity in a grid column exceeds this threshold the shallow convection is shut down.

The LWTHRESH(MOIST) option has been updated to only consider the vertical velocity in the lowest 6km in order to prevent high vertical velocities not related to convection from being used.

LWTHRESHMOIST option works similarly to option LWTHRESH but now only the parameterized moist convection is shut down whereas the dry convection is reduced (depending on grid size) if the threshold in vertical velocity is met. Note that the grid size dependent reduction of the dry convection (if triggering is met) is only tested at gridsizes 500m, 2km and 2.5km. However, no problems are expected for other resolutions. For resolutions from 100m and higher dry convection is reduced to 0.

LSUBEVAP Like LWTHRESH(MOIST) this option is active if a threshold is met that indicates meso-scale organisation. This option checks if the total column sub-grid evaporation and melting exceeds 1mm/hr. If this threshold is met, the same reduction to the parameterized convection is made as when LWTHRESHMOIST is being triggered.

If LMIXUV is TRUE the momentum mixing by the convection scheme will be reduced (50%) in comparison with scalar (temperature and humidity) mixing by the convection scheme

If LDRYMFREDUC is TRUE the parameterized dry convection is reduced in a similar way as when LWTHRESMOIST or LSUBEVAP are triggered (currently to 70%) in case the diagnosed cloud depth in the convection scheme (DUALMF) exceeds the ZCLDDEPTHDP threshold (i.e. the cloud depth exceeds 4km and the model is suppose to resolve the moist convection)

Turbulence scheme

HARATU

HARATU (HArmonie with RAcmo TUrbulence scheme) is the default (HARATU=yes in config_exp.h) turbulence scheme in HARMONIE-AROME originally developed for RACMO (Regional Atmospheric Climate MOdel). The length scale of this turbulence scheme is described by @(Lenderink and Holtslag, 2004). Note that HARATU is only tested in combination with LDUALMF=.TRUE. and CMF_UPDRAFT='DUAL'. The later convection scheme provides input to the HARATU turbulence scheme to present the important energy cascade (from large to small scales), see https://doi.org/10.5194/gmd-15-1513-2022

Statistical cloud scheme

The statistical cloud scheme can use different sources of variance in the variable s (the distance to the saturation curve). Apart from the sources related to turbulence and convection, an extra variance term is used, proportional to variable VSIGQSAT. Although the turbulence and the convection scheme can be resolution dependent, this extra variance term is not (but should be). Therefore, the logical LVARRESDEP (VARiance RESolution DEPendent) is introduced which describes a linear dependence of the extra variance from 0 at grid size 0 (Note that LES already use 0 at e.g. 100m resolution) and 0.02 at 2km, which is also the default value. For resolutions different from 2km this variable needs more testing.

With the option LTAUCONV the convection time scale used in the calculation of the contribution to the variance is not constant but a function of height. The formulation is derived from the reynolds averaged prognostic equation in mass flux variables (see de Roode et al., 2000 https://journals.ametsoc.org/view/journals/atsc/57/10/1520-046920000571585abmfar2.0.co2.pdf) and assumes (as usual) a balance between production and dissipation. In this way the convection time scale is consistent with the convection scheme. The result is a much stronger increase of the variance in the cloud layer in better correspondence with LES. Option LTAUCONV can only be active with option STATNW.