Impacts of the McICA method on the cloud-radiation interaction in the GEM model

Cloud processes and their interactions with radiation are the principal sources of uncertainty in projections made by climate models. This is due, in part, to the various spatio-temporal scales needed to properly simulate these processes and, in part, to an inability to represent subgrid-scale processes in general circulation models (GCMs).

In GCMs, many cloud processes and interactions are often simplified within the microphysics, convection and radiative transfer schemes. Within these schemes parameters are often derived from observations or physical/statistical relationships. Moreover, tuning of these parameters is often used to balance the global energy budget at the top of the atmosphere within the GCM, which may require that the simulated clouds and precipitation disagree with observations. This underlines the fact that the GCM may produce a reasonable radiative budget with the underlying fields being incorrect – this suggests there are compensating biases in the model.

To address these issues in the Global Environmental Multi-scale (GEM) model we use the Monte Carlo Independent Column Approximation (McICA) parameterization for radiative transfer. The McICA uses a stochastic cloud generator to represent subgrid-scale cloud structure which is then randomly sampled. This approach produces unbiased radiative fluxes and heating rates with respect to the full, and computationally expensive, independent column approximation but at the same time introduces conditional random noise which has the potential to adversely influence model simulations.

Within the context of the GEM model we evaluate the impact of using McICA, rather than a deterministic radiative transfer parameterization, on global numerical weather prediction (NWP) and climate simulations. Through NWP simulations the short-term impact of McICA can be evaluated whereas climate simulations allow evaluation of the effect of McICA on long-term feedbacks in GEM. Since the GEM may also be sensitive to the magnitude of McICA's noise, different renditions of McICA are used to test the impact of random noise on GEM. Part of the examination of GEM will include comparing surface and top of atmosphere radiative fluxes as well as the correlation between cloud variables, such as water content and surface radiative fluxes.