The dataset contains hourly outputs of five ICON-D2 ensemble simulations with combined initial and boundary conditions (IBC) and microphysical perturbations. One ensemble dataset consists of 180 ensemble members that are a full combination of 20 IBC, 3 CCN concentration parameters and 3 cloud droplet size distribution shape parameters. The combined impact of microphysical uncertainties can be investigated in the presence of initial and boundary conditions uncertainties in the state-of-the-art numerical weather prediction model.

ICON-D2 microphysically perturbed ensemble simulations including initial and boundary condition uncertainty

Takumi Matsunobu, Christian Keil (LMU MIM) and Christian Barthlott (KIT IMK-TRO)

The dataset contains hourly outputs of five ICON-D2 ensemble simulations with combined initial and boundary conditions (IBC) and microphysical perturbations. One ensemble dataset consists of 180 ensemble members that are a full combination of 20 IBC, 3 CCN concentration parameters and 3 cloud droplet size distribution shape parameters. The combined impact of microphysical uncertainties can be investigated in the presence of initial and boundary conditions uncertainties in the state-of-the-art numerical weather prediction model.

---

Experimental design

We perform numerical experiments using 20 different IBCs, 3 different CCN concentrations and 3 different shape parameters of the CDSD yielding in total a 180-member ICON-D2 ensemble (Fig. 1).

Fig. 1 (a) Design of microphysically perturbed ensemble experiments. (b) Example of cloud droplet size distribution with different shape parameters ν. See Matsunobu et al. (2022) for detail.

The initial conditions are from ICON-D2-KENDA (Kilometer-scale ENsemble Data Assimilation (Schraff et al., 2016)). The first 20 of 40 analyses are used as initial conditions as in operations at DWD. Lateral boundary conditions are based on ensemble ICON global and EU-nest simulations initialised 3 h before the initial time of the ICON-D2 ensembles. The lateral boundary conditions are updated hourly using data from the EU-nest forecasts at lead times from 3 to 27 h.

CCN concentrations are varied across pristine conditions and extremely polluted conditions: maritime (NCN = 100 cm−3), continental (NCN = 1700 cm−3) and polluted (NCN = 3200 cm−3) (Hande et al., 2016). The sub-ensemble using the same IBC and CDSD but sharing the same CCN concentration is named with suffixes m(aritime), c(ontinental) and p(olluted), as shown in Fig. 1a.

The size distribution of hydrometeors is approximated using the generalised gamma distribution (Seifert and Beheng, 2006). The shape parameter ν is varied between 0, 2 and 8 to cover a wide spectrum of the possible shape parameter values (Fig. 1b). The sub-ensemble using the same ν parameter is named with prefixes nu0, nu2 and nu8 (Fig. 1a).

In total, one ensemble consists of 20 IBC x 3 CCN x 3 CDSD = 180 ensemble members.

Case studies and data structure

Simulations for five case studies are performed. In the dataset, 5 top directories are containing each day of the case studies.

Date	Forcing	Directory name
11 Augst 2020	weak	D2KENDA_i2020081100
12 Augst 2020	weak	D2KENDA_i2020081200
13 Augst 2020	strong	D2KENDA_i2020081300
17 Augst 2020	strong	D2KENDA_i2020081700
18 Augst 2020	weak	D2KENDA_i2020081800

Outputs of one ensemble member are stored in a directory. The directory is uniquely identified by the three-layer directory structure; {case study}/{microphysical parameters}/{IBC ensemble member}. Output at a time step is contained only in one grib file. For example, the forecast of IBC member 1 of nu0c simulation at the lead time of 12 hours initialised at 00 UTC 11 August 2020 is: D2KENDA_i2020081100/nu0c/M001/idfrf00120000ML.grb

Directory structure:

Directory name
│
+--nu0c
│   |
│   +--M001
│   |   │   idfrf00000000ML.grb
│   |   │   idfrf00010000ML.grb
│   |   │   ...
|   |
│   +--M002
|   |   |   idfrf00000000ML.grb
|   |   |   ...
|   | ...
|
└───nu0m
│   |
│   +--M001
|   |  ...

---

Related publications

Matsunobu, T., Keil, C., and Barthlott, C. (2022) The impact of microphysical uncertainty conditional on initial and boundary condition uncertainty under varying synoptic control, Weather Clim. Dynam., 3(4), 1273–1289. https://doi.org/10.5194/wcd-3-1273-2022