Superparameterised cloud effects in the EMAC general circulation model (v2.50) – influences of model configuration

Abstract

Abstract. A new module has been implemented in the fifth generation of the ECMWF/Hamburg (ECHAM5)/Modular Earth Submodel System (MESSy) Atmospheric Chemistry (EMAC) model that simulates cloud-related processes on a much smaller grid. This so-called superparameterisation acts as a replacement for the convection parameterisation and large-scale cloud scheme. The concept of embedding a cloud-resolving model (CRM) inside of each grid box of a general circulation model leads to an explicit representation of cloud dynamics. The new model component is evaluated against observations and the conventional usage of EMAC using a convection parameterisation. In particular, effects of applying different configurations of the superparameterisation are analysed in a systematical way. Consequences of changing the CRM's orientation, cell size and number of cells range from regional differences in cloud amount up to global impacts on precipitation distribution and its variability. For some edge case setups, the analysed climate state of superparameterised simulations even deteriorates from the mean observed energy budget. In the current model configuration, different climate regimes can be formed that are mainly driven by some of the parameters of the CRM. Presently, the simulated total cloud cover is at the lower edge of the CMIP5 model ensemble. However, certain “tuning” of the current model configuration could improve the slightly underestimated cloud cover, which will result in a shift of the simulated climate. The simulation results show that especially tropical precipitation is better represented with the superparameterisation in the EMAC model configuration. Furthermore, the diurnal cycle of precipitation is heavily affected by the choice of the CRM parameters. However, despite an improvement of the representation of the continental diurnal cycle in some configurations, other parameter choices result in a deterioration compared to the reference simulation using a conventional convection parameterisation. The ability of the superparameterisation to represent latent and sensible heat flux climatology is independent of the chosen CRM setup. Evaluation of in-atmosphere cloud amounts depending on the chosen CRM setup shows that cloud development can significantly be influenced on the large scale using a too-small CRM domain size. Therefore, a careful selection of the CRM setup is recommended using 32 or more CRM cells to compensate for computational expenses.