Fast Cloud Adjustment to Aerosols

The goal of project S1 "Fast cloud adjustment to aerosols" is an improved understanding of the effect of anthropogenic pollution (aerosols) on clouds, radiation, precipitation and climate. Questions that need to be answered are

  • To what extend will changes in aerosols have an impact on clouds and precipitation?
  • Which are the most important variables to take into account to be able to quantify this based on measurements?

These are not easy questions, because there are competing effects: on one hand, drizzle formation rate by the autoconversion process may be delayed so the cloud lifetime might be affected, on the other hand smaller cloud droplets can lead to more supercooled water freezing at higher altitudes, resulting in an increased surface precipitation. Even a strong effect of CCN and IN load can directly relate to precipitation.

To this, dedicated simulations with the high-resolved HD(CP)² model are conceived, performed, and analysed. The model is enhanced by the representation of essential processes. Moreover, HD(CP)² observational data are used to evaluate and further improve the model, and to work towards a detection and attribution of anthropogenic changes in observations.

Revising the cloud scheme

Cloud droplets in the atmosphere usually form on aerosols. In the early stage of droplet formation, water vapor attaches on the dry aerosol and forms a wetted aerosol or haze particle. If this particle grows further above a critical mass, it is called a cloud droplet. The required critical mass is predicted by Köhler-theory and depends on the size and chemical properties of the aerosol. Transition from a haze particle into a cloud droplet is called "activation". The next stage of droplet growth in a supersaturated environment is governed by the diffusional or condensation growth, where water vapour diffuses towards the cloud droplet and condenses. The aim of work package 1, is a physically correct implementation of the activation and condensation process for high-resolution numerical models, as developed within the first phase of the HD(CP)² project.
The process of droplet activation is crucial for cloud development since cloud droplets form on aerosol particles. If the cloud forms precipitation, large cloud droplets fall out of the cloud and therefore remove the aerosol mass where those droplets initially formed. This wash-out of aerosol is an important cloud-aerosol interaction and commonly called wet scavenging of aerosol particles. Wet scavenging of aerosol modifies the concentration and spatio-temporal distribution of aerosol and therefore influences formation of other clouds as well as air quality. Within this work package, wet scavenging of aerosol particles is also considered for a better description of this process.

Contrails

Air traffic kilometer within one hour for 17th September 2006. Air traffic for 2013 need to be scaled up from air traffic inventory for 2006.

Anthropogenic aerosol have an impact on cloud formation, properties and life times. Comparing those indirect aerosol effects on clouds, or the clouds themselves, in simulations and observations is generally challenging and made even more difficult by the changes in cloudiness due to air traffic. Persistent contrails form in ice supersaturated areas when temperatures are cold enough. Young contrails consist of many very small ice crystals and are microphysically significantly different to natural cirrus clouds. Work package 2 introduces a contrail cirrus parameterization within the ICON-LEM that will allow the simulation of direct cloud changes due to air traffic at the same time as indirect aerosol cloud effects. We adapt an air traffic inventory for use in the high resolution ICON model transforming an air traffic waypoint data set into the ICON icosahedral grid. We then extend the microphysical scheme in ICON to resolve also ice crystal formation within aircraft plumes. A simple contrail cirrus parameterization will be developed covering the main processes of the contrail cirrus life cycle such as ice crystal formation, loss of ice crystals in the aircraft vortices, growth of ice crystals due to deposition, dilution of contrails due to mixing and increase of contrail cross sectional area due to diffusion and sedimentation. We aim to analyze the impact of air traffic on the atmospheric state, such as the humidity and cloud field.

Revised aerosol representation in ICON_LEM and cloud adjustments to aerosol absorption

A prerequisite for realistically simulating the cloud adjustment to aerosol-radiation interactions is a realistic representation of aerosol radiative properties in the model. The aim of work package 3 are new time-varying 3D distributions of cloud condensation nuclei (CCN) and aerosol radiative properties, which will be used in dedicated simulations with the ICON_GCM (global circulation model) and the high-resolved HD(CP)² model to analyze the cloud adjustment effects.

For sensitivity simulations, the available CCN and ice nuclei (IN) climatologies will be merged to one temporally and horizontally constant profile first. A scenario for 2013 (present day) aerosol emissions, and one for 1985 (peak aerosol over Europe) will be developed as basis for CCN scenarios. Simulation results will be evaluated with different kinds of measurements. 3D radiative properties (optical depth, asymmetry parameter and single scattering albedo) will be developed and evaluated. In collaboration with work package 4, additional IN types will be considered. These may include biological aerosols, biomass-burning aerosol and metallic aerosols.


Response of mixed-phase clouds to aerosol perturbations

Atmospheric aerosol particles influence cloud properties by acting as the seeds for cloud droplet and ice particle formation. Work package 4 uses the ICON_LEM to investigate how changes in the concentration of cloud condensing and ice nucleating particles affect the microphysical characteristics of mixed-phase clouds, which comprise both liquid droplets and ice particles. The possible outcomes from perturbing the aerosol size distribution include changes in the thermodynamic phase of the cloud and precipitation formation which affect the cloud lifetime and radiative properties. In addition, the importance of different ice nucleation mechanisms will be estimated.

Modelling of mixed-phase clouds and cloud ice in general comprises many significant challenges. One of these is related to how the models distinguish between different cloud ice categories, such as ice particles, snow, graupel and hail. This task is difficult for traditional microphysical parameterizations that are currently used in the ICON_LEM. Therefore, in the beginning of this work package, a revised microphysical scheme is implemented into the model. The new scheme aims to alleviate the above mentioned challenges by reducing the number of ice categories, but introducing new prognostic equations, which allow a more flexible representation of the ice particle properties.

Statistical relationships between ice clouds and aerosols for model evalutaion

In work package 5 the aerosol and CCN (cloud condensation nuclei) profiles derived in work package 3 are evaluated using measurements from the supersites. Multiwavelength lidar is the main instrument for this purpose. Vertical profiles of the aerosol backscatter coefficient are obtained from continuous lidar measurements and are used for the evaluation. The Figure below shows a 48-h scenario from Leipzig with Saharan dust intrusion at altitudes between 2 and 6 km. The attenuated backscatter coefficient is shown as a time-height-plot. With black lines 30-min averaged profiles of the particle backscatter coefficient are overlaid whenever atmospheric conditions allowed the retrieval. These profiles are used for evaluation.

Fig. 1: Source: Baars, ACP, 2016. Lidar observation of the aerosol conditions above Leipzig on 19 and 20 August 2012.

Furthermore, a mixed-phase cloud data set will be derived from the HD(CP)² supersites to investigate the statistical relationship between ice containing clouds (mixed-phase) and aerosol. For that purpose, an automatic categorization scheme for clouds will be applied to finally derive micro-physical properties of the observed clouds in addition to aerosol profiles and vertical wind velocities. An example for the categorization scheme is shown below. The Cloudnet target categorization using cloud radar, microwave radiometer and lidar is first used to categorize the atmospheric targets. Then, the categorization scheme separates pure liquid clouds (blue rectangle) and mixed-phase clouds (reddish rectangle).

Fig 2.: Source: Bühl, ACP, 2016. Example of the categorization of mixed-phase cloud layers (red boxes) and supercooled liquid cloud layers (blue box) from a Cloudnet dataset.

Cloud process scale dependencies will be investigated with a unique dual-viel-of-view lidar in Leipzig. From that measurements, droplet size is derived within the lowest part of the clouds and linked to the aerosol observation directly below the cloud at different temporal and spatial scales.

Scale- and regime-dependency of cloud-aerosol relationships

Work package 6 deals with the dependencies of clouds and aerosols. A set of cloud properties and their distributions can be retrieved from observations of passive imagers on polar-orbiting satellites. Their global coverage within several days and the use of instruments from multiple platforms allows for an evaluation of modelled spatial and temporal distributions of cloud properties and their sensitivity to meteorological and aerosol parameters by comparing model output to observations, focusing on a statistical approach.
Probability density functions (PDFs) of these observed cloud properties over the HD(CP)² domains will be constructed and compared to PDFs constructed from model output. The impact of spatial resolution on the resulting statistics will be analyzed. Since cloud-aerosol interactions are sensitive to cloud regimes, the PDFs will be stratified with respect to meteorological parameters and the sensitivity of model physics to atmospheric parameters, with respect to observed cloud properties, will be analyzed. Additionally, efforts will be made to relate deviations between modelled and observed cloud properties to either modelled or observed aerosol parameters.

Modelling and Observations

Work package 7 in particular works on the evaluation of the ICON-LES model, we compare the control run and perturbed simulations with observations. It uses a full-domain region (Germany) and performs comparison between the ICON cloud vertical structure (lowest cloud base and higher cloud top) and the retrievals of ground-based ceilometers (DWD network) which yield the lowest cloud base in combination with satellite cloud tops derived from SEVIRI satellite observations. With the ICON-LEM meteogram output available for 39 supersites, it will be possible to derive other products from the simulated variables and analyze these in more detail.

The image shows observed and modeled cloud base heights normalized distributions on 02/05/2013. In blue are shown the ceilometer observations (DWD ceilometer network) in 141 stations around Germany. In orange are displayed modeled cloud bases heights from ICON-LES simulations done at domain 1 level, and in yellow the perturbed simulations doubling the CCN (cloud condensation nuclei) concentrations.

The image shows observed and modeled cloud base heights normalized distributions on 02/05/2013. In blue are shown the ceilometer observations (DWD ceilometer network) in 141 stations around Germany. In orange are displayed modeled cloud bases heights from ICON-LES simulations done at domain 1 level, and in yellow the perturbed simulations doubling the CCN (cloud condensation nuclei) concentrations.

Work package 8 "Detectability in observations" in particular conceives and assesses the simulations. On the basis of the model results it will be examined which observations are useful - or, if not available, which ones would be useful - to detect and attribute aerosol-cloud-precipitation effects.

The figure shows the weekly cycle over Europe of aerosol optical depth (top), cloud droplet number concentration (middle) and cloud fraction (bottom) from satellite observations (left, on Terra in blue and on Aqua satellite in red) and in simulations with the HadGEM and ECHAM5 models (middle and right, respectively) for a simulation with imposed weekly cycle in anthropogenic aerosol emissions (red) and a control simulation sampling only natural variability without a weekly cycle (grey). Despite the use of five years of data, the weekly cycle can be detected only for cloud droplet number concentration, but not for cloud fraction, illustrating the challenge for this project.