Tuning parameters

Please note that this description applies to the NWP Physics Package.

To get the best performance out of a model for a specific application, or for a particular area on the globe, it can be useful to change certain parameters of the model specification within reasonable bounds.

ICON model parameters suitable for model tuning

The tables summarize the most important tuning variables for the ICON model, and are largely based on Reinert et al. 2025, chapter 12.2 and Avgoustoglou et al. 2020. Yet, the document and the list of variables should be handled with care. Purely varying some of the listed parameters blindly will most likely not give satisfactory results. A physical understanding of the identified model shortcomings/biases should be built up first, followed by a choice of the associated model parameters and a systematic variation and evaluation of simulations. The parameters of interest may strongly vary for the region of interest, the model resolution and the specific purpose. Please also keep in mind that the list is neither exhaustive, nor complete. There may well be further model parameters that are more suitable for individual applications.

SSO tuning

Tuning of the SSO and GWD parameters is dependent on the employed external parameters.

Parameter	Description	Meaningful Range	Comment
gkwake	low level wake drag constant \(C_d\) for blocking	\(1.5 \pm 0.5\)	Very strong dependency on raw data resolution: for ICON-D2 with ASTER data, we use 0.25
gkdrag	gravity wave drag constant \(G\), a function of mountain sharpness	\(0.075 \pm 0.04\)	Should be zero (turned off) at convection-permitting resolutions
gfrcrit	critical Froude number determining depth of blocked layer \(H_{n_{crit}}\)	\(0.4 \pm 0.1\)
grcrit	critical Richardson number	0.25
tune_minsso	minimal value of SSO-STDH (m) where SSO-effects are being considered	default 10	must also be adapted in extpar!
tune_blockred	multiples of the SSO-STDH, above which the SSO-blocking tendency is being reduced proportionally to STDH/z_AGL	15	default 100 = deactivated

GWD tuning

Parameter	Description	Meaningful Range	Comment
gfluxlaun	variability range for non-orographic gravity wave launch momentum flux	\(2.50 \cdot 10^{-3}\) \(\pm 0.75\cdot 10^{-3}\) [Pa]	relevant for global applications only

grid scale microphysics

Parameter	Description	Meaningful Range	Comment
zvz0i	terminal fall velocity of ice	\(0.85 \pm 0.25\) [m/s]	allows temperature bias tuning in the upper tropical troposphere as well as TOA long-wave fluxes
zceff_min	minimum value for sticking efficiency	0.01 - 0.075	tropics
zcsg	efficiency for cloud-graupel riming	0.5
v0snow	factor in the terminal velocity for snow	10.0 - 30.0	depending on microphysics scheme, see gscp_data.f90
icesedi_exp	exponent for density correction of cloud ice sedimentation	0.3 - 0.33	no perturbation recommended
rain_n0fac	multiplicative change of intercept parameter of raindrop size distribution	0.25 - 4.	multiplicative perturbation

cloud cover

Parameter	Description	Meaningful Range	Comment
box_liq	Box width for liquid clouds assumed in the cloud cover scheme	\(0.05 \pm 0.02\)
box_liq_asy	Asymmetry factor for liquid cloud cover diagnostic	2.0 - 4.0 (def. 3.0)	sensitive to TOA solar fluxes and to a lesser degree long-wave fluxes
box_liq_sfc_fac	Tuning factor for box_liq reduction near the surface	1.0
box_ice	Box width for ice clouds assumed in the cloud cover scheme	0.05
thicklayfac	factor for increasing the box width for layer thicknesses exceeding 150 m	\(0.005 \pm 0.005\) [1/m]	accounting for vertical sub-grid overlap
sgsclifac	Scaling factor for turbulence-induced subgrid-scale contribution to diagnosed cloud ice	0.0 - 1.0	0.0 turns this contribution off
supsat_limfac	Limiting factor for allowed supersaturation in satad	0.
allow_overcast	Tuning factor for steeper dependence CLC (RH)	\(\leq 1.0\)	setting allow_overcast \(\leq 1\) together with reduction of tune_box_liq_asy causes steeper CLC(RH) dependence. recommendation: allow overcast<1 should not be used in combination with lsgs_cond=.TRUE.

turbulence

Parameter	Description	Meaningful Range	Comment
q_crit	critical value for normalised super-saturation	\(1.6 \pm 1.0\)
rlam_heat	scaling factor of the laminar boundary layer for heat (scalars), the change in rlam_heat is accompanied by an inverse change of rat_sea in order to keep the evaporation over water (controlled by rlam_heat\(\cdot\)rat_sea) the same. recommendation: the product of rlam heat and rat sea should not be significantly larger than 10. Otherwise, there will be too little evaporation over the oceans.	\(10.0 \pm 8.0\)	additive perturbation
rat_sea	controls latent and heat fluxes over water	0.8 - 10.0	lower values increase latent and sensible fluxes over water
a_hshr	length scale factor for the separated horizontal shear mode	\(1.0 \pm 1.0\)
a_stab	factor for stability correction of turbulent length scale	\(0.0 \pm 1.0\)	turned off by default because it degrades global skill scores
c_diff	length scale factor for vertical diffusion of TKE	\(0.2 \pm 2.0\)
alpha0	lower bound of velocity-dependent Charnock parameter	0.0123-0.0335	additive ensemble perturbation of Charnock-parameter
alpha1	parameter scaling the molecular roughness of water waves	0.1-1.0	lower values increase latent and sensible fluxes over water, particularly at low wind speeds.
tur_len	asymptotic maximal turbulent distance	\(500. \pm 150.\) [m]	default is 150 m
tkhmin	scaling factor for minimum vertical diffusion coefficient for heat and moisture	\(0.75 \pm 0.2\)	0.75
tkmmin	scaling factor for minimum vertical diffusion coefficient for momentum	\(0.75 \pm 0.2\)
tkred_sfc	multiplicative change of reduction of minimum diffusion coefficients near the surface	0.25 - 4.0	multiplicative perturbation

TERRA

Parameter	Description	Meaningful Range	Comment
c_soil	evaporating fraction of soil	\(1.0 \pm 0.25\)
cwimax_ml	scaling parameter for maximum interception storage	\(5.\cdot 10^{-7} - 5.\cdot 10^{-4}\)	multiplicative perturbation, low values (\(< 10^{-6}\)) turn off interception layer
minsnowfrac	Lower limit of snow cover fraction to which melting snow is artificially reduced in the context of the snow-tile approach	\(0.2 \pm 0.1\)
dust_abs	Tuning factor for enhanced LW absorption of mineral dust in the Saharan region	0.0	Reduces bias over Sahara for the RRTM scheme but not necessary and implemented with ecRad and itype_lwemiss=2

convection

Parameter	Description	Meaningful Range	Comment
entrorg	entrainment parameter in convection scheme valid for dx=20km	\(1.95\cdot 10^{-3}\pm 0.2\cdot 10^{-3}\)	corresponds to entr_sc in the shallow convection part of COSMO Tiedtke scheme
rdepths	maximum allowed shallow convection depth	\(2.0 \cdot 10^{4}\) \(\pm 5.0\cdot 10^{3}\) Pa
rprcon	coefficient for conversion of cloud water into precipitation	\(1.4\cdot 10^{-3}\pm 0.25\cdot 10^{-3}\)
capdcfac_et	fraction of CAPE diurnal cycle correction applied in the extratropics	\(0.5 \pm 0.7\)
capdcfac_tr	fraction of CAPE diurnal cycle correction applied in the tropics	\(0.5 \pm 0.75\)
lowcapefac	tuning parameter for diurnal-cycle correction in convection scheme: reduction factor for low-cape situations	\(1.0 \pm 0.5\)
negpblcape	tuning parameter for diurnal-cycle correction in convection scheme: maximum negative PBL CAPE allowed in the modified CAPE closure	-500.- 0.
rhebc_land	RH threshold for onset of evaporation below cloud base over land	\(0.825 \pm 0.05\)	0.75 as default in code
rhebc_ocean	RH threshold for onset of evaporation below cloud base over sea	\(0.85 \pm 0.05\)
rhebc_land_trop	RH threshold … over tropical land	\(0.70 \pm 0.05\)	tropics
rhebc_ocean_trop	RH threshold …over tropical sea	\(0.76 \pm 0.05\)	tropics
rcucov	convective area fraction used for computing evaporation below cloud base	0.075	0.05 coded as default
rcucov_trop	convective area fraction used for computing evaporation below cloud base, tropics	0.03	tropics
texc	Excess value for temperature used in test parcel ascent	\(0.125 \pm 0.05\) [K]
qexc	Excess fraction of grid-scale QV used in test parcel ascent	\(0.0125 \pm 0.005\) [kg/kg]