ACCLAIM

Aerosol effects on Convective CLouds And clIMate (ACCLAIM)

Despite the potential importance of aerosol effects on convective clouds, sign and magnitude of the introduced global radiative perturbations are entirely unclear.

In the European Research Council funded project ACCLAIM, we consistently combine global models, cloud resolving models and remote sensing data to advance our understanding of the physical interactions of aerosols and convective clouds:

Motivation

Strong correlations of satellite retrieved aerosol properties and indicators for convective clouds exist, for example in the observed statistical correlation of satellite retrieved aerosol optical depth and cloud top pressure:


Figure: MODIS satellite retrieved relationship between cloud top pressure (CTP) and aerosol optical depth (AOD) expressed as linear regression of dln(CTP)/dln(AOD) of 10 years of 1x1 gridded level 3 retrieval products [Benjamin Grandey].

Similar analyses have been the basis of the "invigoration hypothesis" put forward in Rosenfeld et al. (2008). However, sole analysis of satellite correlations does not necessarily establish causality – synoptic covariability, retrieval errors and sampling artefacts will contribute to or even dominate the relationships between satellite retrieved aerosol and cloud properties.


Figure: Pyrocumulus cloud triggered by wildfires in the Yellowstone National Park on 1 August 2008 (Wikipedia).

Team

The funding provided by the ERC allows supporting a focused team within one research group, fostering exchange across methodologies.

Principal Investigator Researchers Doctoral Students
Philip Stier Zak Kipling Sarah Taylor
A bit of everything Global modelling Satellite remote sensing
Bethan White
Cloud resolving modelling

Global Modelling: ECHAM–HAM–CCFM

Our global modelling strategy focuses on aerosol–convection interactions in the global aerosol–climate model ECHAM–HAM (Stier et al., 2005; Zhang et al., 2012) coupled to the Convective Cloud Field Model (CCFM) (Wagner and Graf, 2010). HAM provides size-resolved aerosol composition (two-moment modal microphysics scheme M7) online in the general circulation model ECHAM.

CCFM is based on a new formulation of the quasi-equilibrium closure hypothesis of Arakawa and Schubert (1974) and simulates a full spectrum of different cumulus clouds in each global model column. Each cloud is simulated using a one-dimensional Lagrangian entraining parcel model with its own radius and updraught velocity, and an embedded parameterisation of mixed-phase microphysics (Zhang et al, 2005; we are carrying out further work on this as part of the BACCHUS project). The spectrum is determined dynamically based on the grid-scale environment via competition between cloud types for convective available potential energy (CAPE); this leads to a set of Lotka–Volterra equations, similar to those found in population ecology.

To facilitate the consideration of aerosol effects on convection in a climate model, we have extended CCFM in ACCLAIM with a sub-cloud model based on dry entraining parcel theory (Kipling et al., submitted) This allows the model to determine the updraught velocity at cloud base, not just its subsequent evolution, making it possible to explicitly simulate the activation of aerosols into cloud droplets via the same parameterisation (Abdul-Razzak and Ghan, 2000) used for stratiform clouds.

Figure: Global joint distribution of convective cloud-base radius and updraught velocity from ECHAM–HAM–CCFM (Kipling et al., submitted)

Our baseline evaluation (Kipling et al., submitted) shows CCFM performing well compared to the standard Tiedtke–Nordeng scheme used in ECHAM–HAM in terms of the global distribution of precipitation and cloud cover.

Figure: Taylor diagram of monthly precipitation from ECHAM(–HAM)(–CCFM) against GPCPFigure: Taylor diagram of monthly cloud cover from ECHAM(–HAM)(–CCFM) against CALIPSO–GOCCP

High Resolution Aerosol–Cloud Modelling: WRF

For detailed process studies we use the Advanced Research Weather Research and Forecasting (WRF) model in real-data Cloud Resolving Modelling (CRM) and Large Eddy Simulation (LES) configurations as well as idealised configurations [Skamarock et al.,2008]. WRF is a three-dimensional, nonhydrostatic, compressible model. The limited area model is initialised using global analysis files and can be nested down to very high horizontal resolution (1km and beyond).

WRF is used in large-domain, convection-permitting high-resolution (4km, 3km, 1km and sub-1km horizontal grid lengths) configurations. Our real-data simulations focus on regions of intense deep convective activity and sources of biomass burning aerosol (the Congo basin, the Amazon). The following video shows 48 hours of a real-data simulation of deep convection over the Congo basin. The model domain covers an area of about 2000 km2 and has a horizontal grid length of 4 km. The blue surface shows cloud, and the grey surface shows rain. This video was produced using VAPOR (vapor.ucar.edu), a product of the Computational Information Systems Laboratory at the National Center for Atmospheric Research

We explore the the effects of aerosols on convection through investigating the response of different microphysics representations to aerosol, cloud or ice condensation nuclei (CCN, IN) or cloud droplet number concentration CDNC perturbations. Work from our group shows that uncertainty in the response of two-moment bulk microphysics parameterisations to perturbations in prescribed CDNC between different microphysics schemes, model configurations and cases of convection far outweighs any convective invigoration response in the model (White et al., in prep).

Remote Sensing: SEVIRI

In order to capture the full range of spatial and temporal variability of convective cloud, high time-resolution observations are required across a large area. We use remote sensing data from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) instrument aboard the geostationary Meteosat Second Generation (MSG) satellite to observe the lifecycle of individual convective clouds and to understand interactions between convection, aerosols and meteorology.

SEVIRI provides continuous, high time-resolution observations over a large area, including the Congo Basin region. Unlike low earth orbit satellites, SEVIRI is therefore able to sample the entire lifecycle of individual clouds. SEVIRI observations are combined with the Cumulonimbus Tracking and Monitoring (Cb-TRAM) algorithm [Zinner et al. 2008] to identify and track individual convective cloud cores across Sub-Saharan Africa and the southern Atlantic Ocean. We associate an anvil with each core by applying an image processing algorithm to SEVIRI brightness temperatures.

The following video shows SEVIRI brightness temperatures (greyscale background), tracked convective cloud cores (solid colours), their associated anvils (transparent colours) and MODIS collection 6 aerosol optical thickness (green-blue overlay) in the Congo Basin for the period 1-10 August 2007.

Tracked cores and their associated anvils are used to quantify spatial and temporal variability in various metrics of convective cloud lifecycles including the time and location of convective initiation and dissipation, degree of organization, cloud lifetime, cloud top height and the co-evolution of cores and anvils.

In the Congo Basin region, tracked clouds are collocated with aerosol and cloud properties from the SEVIRI and the low earth orbit satellites MODIS (Moderate Resolution Imaging Spectroradiometer) and TRMM (Tropical Rainfall Measuring Mission). Tracked clouds and collocated variables are rotated onto a common direction of motion, creating composite images of clouds at various stages of development. By clustering composites according to their meteorological environment, we are able to take advantage of SEVIRI’s high time-resolution observations to investigate the contribution of meteorology to observed correlations in aerosol and cloud retrievals.

Publications

Most of the Climate Processes Group's publications are relevant for this project. Examples of directly resulting work include: