Uncertainties in future climate predictions due to convection parameterisations

Abstract

In the last decades several convection parameterisations have been developed to consider the impact of small-scale unresolved processes in Earth System Models associated with convective clouds. Global model simulations, which have been performed under current climate conditions with different convection schemes, significantly differ among each other in the simulated transport of trace gases and precipitation patterns due to the parameterisation assumptions and formulations, e.g. the simplified treatment of the cloud microphysics. Here we address sensitivity studies comparing four different convection schemes under alternative climate conditions (doubling of the CO₂ concentrations) to identify uncertainties related to convective processes. The increase in surface temperature reveals regional differences up to 4 K dependent on the chosen convection parameterisation. The increase in upper tropospheric temperature affects the amount of water vapour transported to the lower stratosphere. Furthermore, the change in transporting short-lived pollutants within the atmosphere is highly ambiguous for the lower and upper troposphere. Finally, cloud radiative effects have been analysed uncovering a shift in different cloud types in the tropics

Future changes in regional precipitation simulated by a half‐degree coupled climate model: Sensitivity to horizontal resolution

“…Simulation quality related to time step is a model development issue and beyond the scope of the present paper, as is a more extensive intercomparison among different convective schemes for simulation characteristics related to model resolution. However, model uncertainties due to convective parameterization often revolve around the computation of mass fluxes, detrainment levels, and trigger functions, [ Rybka and Tost , ; Donner et al ., ].…”

Improved Climate Simulation by Using a Double‐Plume Convection Scheme in a Global Model

“…General circulation models (GCMs) are widely used for global weather forecasting and climate modeling. In a GCM, the convective parameterization, which represents the bulk effects of convection, is typically regarded as a large source of model uncertainty (Arakawa, 2004; Rybka & Tost, 2014; Tost et al., 2006). Cumulus convection interacts with other processes in complex ways.…”

Fighting big data and ensemble fatigue in climate change impact studies: Can we turn the ensemble cascade upside down?

“…Various comparative studies have analysed climate models and concluded that a single best climate model does not exist (Dai, 2006; Stocker et al ., 2013; Jacob et al ., 2014; Rajczak and Schär, 2017; Vanden Broucke et al ., 2018). For a specific location, variable and season, the skill of different climate models and the climate change signals deviate among the models due to differences in spatial resolution (Anstey et al ., 2013; Watterson et al ., 2014; Mendoza et al ., 2016; Davini et al ., 2017; Hartung et al ., 2017), initial climatic conditions representing the internal variability of the climate system (Deser et al ., 2012; Fadhel et al ., 2017; Hosseinzadehtalaei et al ., 2017), parametrization schemes (Arakawa, 2004; Rybka and Tost, 2014; Prein et al ., 2015) and model components (Pitman, 2003; Lorenz et al ., 2012; Alexander and Easterbrook, 2015). Besides climate model differences, similarities arise from shared coding, the process of continuous model improvements and the fact that they all represent the same climate system (Masson and Knutti, 2011; Pennell and Reichler, 2011; Knutti et al ., 2013; Sunyer et al ., 2013; Leduc et al ., 2016).…”