ICON model parameters suitable for model tuning and perturbation

last updated: 23 February 2021
get the pdf document

Purpose

This page lists the most important tuning variables for the ICON model, which may also be suitable for model perturbation.

It is based on the document available as pdf by Schlemmer et al (largely based on Reinert et al., 2020, chapter 12.2), which resides in the icon model git repository.

In addition to the description of the parameters, their range of variability, and some comments from the developers, this page aims at including also the experience matured in the COSMO Consortium on the usage of these parameters for model tuning (5th column of the table) and for model perturbation (6th column of the table). In addition, which parameters are operationally perturbed in the ensembles of the COSMO members is also documented in the 7th column of the table.

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.

This page will be kept updated with respect to the original list in the icon repository updated by the developers. The COSMO scientists are invited to fill the last three columns of the table based on their experience with ICON, referring to the appropriate Working Group Coordinator.

Tunable parameters (related to )


parameter

description
meaningful
range
comment from
the developers
model
tuning
EPS
perturbation
EPS perturbation
in production

SSO tuning

gkwake
low level wake drag constant Cd for blocking
1.5 ±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 ±0.04
Should be zero (turned off) at convection-permitting resolutions
gfrcrit
critical Froude number determining depth of blocked layer Hncrit
0.4 ±0.1
grcrit
critical Richardson number
0.25

GWD tuning

gfluxlaun
variability range for non-orographic gravity wave launch momentum flux
2.50·10-3 ±0.75·10-3
[Pa] relevant for global applications only

grid scale microphysics

zvz0i
terminal fall velocity of ice
0.85-1.45 [m/s]
allows temperature bias tuning in the upper tropical troposphere as well as TOA long-wave fluxes
mostly relevant for simulations covering the tropics
ICON-D2-EPS
zceff_min
minimum value for sticking efficiency
0.01-0.075 [m/s]
tropics
mostly relevant for simulations covering the tropics
v0snow
factor in the terminal velocity for snow
10.0-30.0
recommended value 25.0
identified as "sensitive" for COSMO in [1]
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
0.25-4.0
multiplicative perturbation
identified as "sensitive" for COSMO in [1]
ICON-D2-EPS

cloud cover

box_liq
Box width for liquid cloud diagnostic in cloud cover scheme
0.05 ±0.02
ICON-D2-EPS
box_liq asy
Asymmetry factor for liquid cloud cover diagnostic
2.0-4.0
sensitive to TOA solar fluxes and to a lesser degree long-wave fluxes
ICON-D2-EPS
thicklayfac
factor for increasing the box width for layer thicknesses exceeding 150 m
0.005 ±0.005 [m-1]
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

turbulence

q_crit
critical value for normalised super-saturation
1.6-4.0
identified as "sensitive" for COSMO in [1]
ICON-D2-EPS
rlam_heat
scaling factor of the laminar bound- ary 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 (con- trolled by rlam heat·rat sea) the same.
0.333-3.0
multiplicative perturbation
identified as "sensitive" for COSMO in [1]
ICON-D2-EPS
rat_sea
controls latent and heat fluxes over water
0.5-10.0
lower values increase latent and sensible fluxes over water
identified as "sensitive" for COSMO in [1]
a_hshr
Length scale factor for the separated horizontal shear mode
1.0 ±1.0
ICON-D2-EPS
a_stab
factor for stability correction of tur- bulent length scale
0.0
turned off by default because it degrades global skill scores
identified as "sensitive" for COSMO in [1]
ICON-D2-EPS
c_diff
length scale factor for vertical diffu- sion of TKE
0.1-0.4
ICON-D2-EPS
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 la- tent and sensible fluxes over water, particularly at low wind speeds.
tur_len
asymptotic maximal turbulent distance
500 alpha ±150 [m]
identified as "sensitive" for COSMO in [1]
ICON-D2-EPS
tkhmin
scaling factor for minimum vertical diffusion coefficient for heat and moisture
0.75 ±0.2
identified as "sensitive" for COSMO in [1]
ICON-D2-EPS
tkmmin
scaling factor for minimum vertical diffusion coefficient for momentum
0.75 ±0.2
identified as "sensitive" for COSMO in [1]
ICON-D2-EPS
tkred_sfc
multiplicative change of reduction of minimum diffusion coefficients near the surface
0.25-4.0
multiplicative perturbation

TERRA

c_soil
evaporating fraction of soil
1.0 ±0.25
identified as "sensitive" for COSMO in [1]
TLE-MVE
cwimax_ml
scaling parameter for maximum interception storage
5·10-7-5·10-4
low values (10-6) turn off interception layer

snow cover diagnosis

minsnowfrac
Lower limit of snow cover fraction to which melting snow is artificially reduced in the context of the snow-tile approach
0.2 ±0.1

radiation

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

entrorg
Entrainment parameter in convection scheme valid for dx=20km
1.95·10-3 ±0.2·10-3
corresponds to entr_sc in the shallow convection part of COSMO Tiedtke scheme
ICON-D2-EPS
rdepths
maximum allowed shallow convection depth
2.0·104 ±5.0·103 [Pa]
ICON-D2-EPS
rprcon
coefficient for conversion of cloud wa- ter into precipitation
1.4·10-3 ±0.25·10-3
capdcfac_et
fraction of CAPE diurnal cycle correction applied in the extratropics
0.5 ±0.75
capdcfac_tr
fraction of CAPE diurnal cycle correction applied in the tropics
0.5 ±0.75
lowcapefac
Tuning parameter for diurnal-cycle correction in convection scheme: re- duction factor for low-cape situations
1.0 ±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
0.75 ±0.05 [m/s]
tropics
rhebc_ocean
RH threshold for onset of evaporation below cloud base over sea
0.85 ±0.05 [m/s]
rhebc_land_trop
RH threshold ... over tropical land
0.70 ±0.05
tropics
mostly relevant for simulations covering the tropics
rhebc_ocean_trop
RH threshold ... over tropical sea
0.76 ±0.05 [m/sec]
tropics
identified as "sensitive" for COSMO in [1]
rcucov
Convective area fraction used for computing evaporation below cloud base
0.05
rcucov_trop
Convective area fraction used for computing evaporation below cloud base
0.03
tropics
identified as "sensitive" for COSMO in [1]
texc
Excess value for temperature used in test parcel ascent
0.125 ±0.05 [K]
ICON-D2-EPS
qexc
Excess fraction of grid-scale QV used in test parcel ascent
0.0125 ±0.005 [kg/kg]
ICON-D2-EPS

References

[1] Avgoustoglou E., Voudouri A., Carmona I., Bucchignani E., Levy Y., and Bettems J.M., 2020:
COSMO technical report 42. A methodology towards the hierarchy of COSMO parameter calibration tests via the domain sensitivity over the Mediterranean area
COSMO TechReport No42, doi:10.5676/ DWD pub/nwv/cosmo-tr 42.

[2] Reinert D., and co-authors, 2020:
DWD Database Reference for the Global and Regional ICON and ICON-EPS Forecasting System, Version 2.1.1.
Research and Development at DWD, Offenbach am Main, icon_dbbeschr_aktuell, doi:10.5676/DWD pub/nwv/icon 2.1.1.