Workpackage 1 focuses on the formation, evolution and growth of ice particles in deep convective clouds and the formation of convectively-generated cirrus (anvils) and convectively-generated mid-level stratiform clouds (stratiform region of convective systems). The questions addressed in this work package are:
- Which ice nucleation processes (homogeneous versus heterogeneous freezing) are dominant for the formation of anvils and the stratiform region of convective systems?
- How important is riming (the transition from snow to graupel) for the formation and the size of the stratiform region?
- How can we better parameterize these processes in large-scale models used for climate research and numerical weather prediction?
To answer these questions, a particle-based microphysics model has been developed .
As atmospheric ice particles can pass through several states, e.g.
In our model the multidimensional phase space is spanned by the microphysical properties: ice mass mi, monomer number Nm, rime mass mr, rime volume Vr, and liquid mass mw. These properties are influenced by processes such as nucleation, vapor diffusion, sedimentation, aggregation, riming, rime-splintering, melting, shedding, and break-up.
To sample the multidimensional phase space, super-particles are used in a Monte-Carlo approach. A super-ice-particle represents ξ similar real ice particles at the vertical position z.
Currently the model is being implemented into the ICON-LEM framework. This will allow to show what trajectories and pathways individual snowflakes and graupel particles take in a dynamic, three-dimensional convective cloud. Such 3D simulations will need to use very advanced simulation techniques to follow billions of particles in a cloud in a single run. It then can be explicitly shown how these particles grow, what makes one snowflake different from another and explain why some end up in the anvil, some in mid-levels, and some reach the ground as precipitation. The results in Figure 1 are taken from the following publication.
 S. Brdar & A. Seifert 2017, “McSnow – A Monte-Carlo particle model for riming and aggregation of ice particles in a multidimensional microphysical phase space”, submitted to Journal of Advances in Modeling Earth Systems.
Work package 2 investigates the occurrence as well as the microphysical and macrophysical properties of cirrus outflow of Warm Conveyor Belts (WCB). The use of a newly developed microphysics scheme allows us to easily differentiate between the contribution of multiple ice formation processes such as homogeneous and heterogeneous nucleation, freezing of cloud droplets and ice multiplication. With the help of this method we contribute to the leading research question if cirrus outflow of WCBs is formed in situ or if it is of liquid origin.
WCBs are identified by trajectory calculations and simulated with ICON in order to investigate the dependence of the formation pathways on environmental conditions and embedded convection and to characterise their physical properties.
Work packages 3 and 4 focus on the convective outflow and its representation in NWP (numerical weather prediction) and GCM (global circulation) models. The goal is to achieve a better understanding and ultimately better representation of processes which influence the development of convection and its outflow when parameterized. An ensemble of low model runs (ICON-NWP and ICON-GCM) covering spatial horizontal resolutions between 13 and 160 km and different initial conditions have been performed to identify shortcomings while simulating an explosive convective event (4./5.7.2015). These simulations are compared against ICON-LEM runs (156m - 625m resolution) and observations. Currently one specific convective day is chosen but further analysis of multiple days will follow.
- The ice cloud cover and cloud water content/path (including different types of hydrometeors) will be evaluated in these simulations.
- An analysis of mean profiles of moisture and stability, detrainment volume and ice content will be performed.
- We will attempt to improve moisture transports and equilibrium stability profiles within the parameterizations of turbulent and convective transports to provide accurate inputs for the cloud parameterization.
Work packages 5 and 3 study the impact of convection on the water budget and variability in the upper troposphere, anvil top temperatures, ice supersaturation and the macro-scale distribution (e.g. cover and ice water) of ice clouds and their temporal evolution in the ICON-GCM (global circulation model). For this reason a parameterization for ice cloud macrophysics was developed and implemented in ICON-GCM, which allows for supersaturation and accounts for the ice cloud cover hysteresis. Simulations with the LEM [hier bitte link auf die entsprechende Unterseite] model can be used to study the variability of total water and its tendencies due to specific processes (in particular convective processes) in order to improve the statistical cloud parameterization. Of special interest is the investigation of the prevalence of bimodal total water distributions in convective meteorological situations, as found in the LEM and the significance of their representation in the GCM parameterization.
The climate impact of convection depends to a large degree on the cloud top heights and on anvil cloud coverage, life time and properties. Therefore, we concentrate on those variables in the model evaluation. We compare to observations as computed by the CiPS algorithm (SEVIRI) which is trained to be sensitive to thin ice clouds. This is important for the determination of cloud top heights.
Workpackage 6: To provide more process-oriented validation of convective systems in the model system, we apply a combination of on-line feature identification and a trajectory tool. In this approach, “massless” particles are started in regions identified to be favoured to convection initiation. These particles are transported in the model system in a Lagrangian framework, resulting in on-line trajectories.
The Lagrangian viewpoint will provide more insight in the development of convective cells and will enable a more process-oriented validation of the model system. This diagnostic will be applied to study the life cycle of convective systems, the characteristics of convection and the outflow of convective systems. In future, statistical data, like the distribution of mass fluxes, freezing levels, microphysical conversion rates, detrainment, anvil top heights and temperature within convective systems should be aggregated automatically during the simulation using this novel diagnostic tool.
In workpackage 8 four goals are specified:
- the statistical evaluation of anvil properties
- the evaluation of the ice-microphysical scheme
- a validation of the satellite retrieval products, and
- the determination of the retrieval accuracy of the Ice Water Content (IWC) Cloudnet product.
In order to do so, cloud microphysical remote measurement products from the Richard Aßmann Observatory in Lindenberg are used. Since the first three tasks depend heavily on the retrieval accuracy of the IWC Cloudnet product, the latter has to be determined first. The product is therefore compared to the IWC measured by the Raman Lidar RAMSES, which is the only instrument worldwide capable of measuring the IWC. Yet, RAMSES IWC depends on one spectroscopic quantity, which is currently poorly known - on the efficiency of the Raman backscattering of small droplets compared to big droplets. For this reason, an experiment is currently set up at the observatory to measure the Raman efficiency directly. As another step the other three tasks are performed.
This workpackage brings observational scientists and model developers together to identify model biases and develop hypotheses for the model behaviour of ICON-LEM which can be tested and which lead us to improved parametrizations. We study the microphysical properties of convectively generated ice clouds and their dependence on background temperatures in the ICON-LEM model. These results can be evaluated with satellite data. We perform sensitivity studies to gain insights in the cause of the deviations
between model and observations. The simulations are executed on JUQUEEN, the high performance computing machine at the Research Centre in Jülich or on the subsequent machine JURECA. respectively the successor at Research Centre Jülich.