Susceptibilities and Resilience in Aerosol-Cloud Interaction

The contribution of aerosols continues to pose an important uncertainty in anthropogenic climate forcing (IPCC Assessment Report 5). However, over the last decade and with increasingly detailed process representation in climate models, forcing estimates tend to decrease (Figure 1). Such a trend would back the notion of clouds as resilient or buffered systems, which has emerged from small-scale cloud modeling (Stevens & Feingold, 2009): Like chemical buffer solutions that keep a constant pH-value at the addition of an acid or base, the complex interplay of cloud processes and their dependence on the cloud regime might compensate aerosol perturbations on local to global scales.

In this project, we study cloud adjustments to aerosol perturbations and aim to quantify the degree of resilience observed for different cloud regimes and resolutions. Our approach is based on quantifying the link strengths in aerosol-cloud-radiation-precipitation interaction (Figure 2) by susceptibilities dln(effect)/dln(cause), i.e. relative sensitivities or effective exponents, respectively.

Connecting the process to the global scale, we compiled the regional climate model COSMO-ART-M7. To capture major processes from aerosol emission to radiation and precipitation, the model is coupled to a modal aerosol scheme and includes aerosol activation and heterogeneous freezing as well as two-moment cold and warm cloud microphysics.

Enlarged view: "effective" aerosol forcing
Figure 1: Model and satellite-based IPCC estimates for the "effective" aerosol forcing, i.e. including adjustments. On the x-axis, models are sorted by time (AR4: prior to 2007, AR5: 2007-2013, CMIP5/ACCMIP:2013) and increased complexity in cloud representation (IPCC Assessment Report 5, adaptations in red).
Enlarged view: Chain model of aerosol-cloud-radiation-precipitation interaction
Figure 2: Chain model of aerosol-cloud-radiation-precipitation interaction. Adjustment processes may weaken the strength of links as compared to the non-coupled parametrization. Some proposed adjustment effects are indicated next to the corresponding links.
JavaScript has been disabled in your browser