Assessing the response of several GCMs to the McICA radiative transfer methodology
| Barker, Howard | Meteorological Service of Canada |
| Cole, Jason | Meteorological Service of Canada |
| Clothiaux, Eugene | The Pennsylvania State University |
| Li, Jaingnan | |
| Morcrette, Jean-Jacques | European Centre for Medium-Range Weather Forecasts |
| Pincus, Robert | NOAA-CIRES Climate Diagnostics Center |
| Raisanen, Petri | Finnish Meteoroligical Institute |
| Stephens, Graeme | Colorado State University |
Recently, the Monte Carlo Independent Column Approximation (McICA) was introduced as a new approach for parametrizing radiative transfer within global climate models (GCMs). The McICA computes domain-averaged, spectrally-integrated radiative fluxes by randomly sampling stochastically-generated subgrid-scale columns during spectral integration. The McICA is unbiased with respect to the full ICA, and because it removes the description of cloud structure from the radiative transfer code, it is flexible and computationally efficient. However, since the McICA is a Monte Carlo procedure, it generates conditional (unbiased) random noise. As an example of the impact of McICA noise, results are presented for an ensemble of annual-cycle simulations with varying levels of McICA noise performed by the NCAR Community Atmospheric Model version 1.8 (CAM-1.8). Given the range of nonlinear parametrizations in GCMs and their interactions with radiation, the impact of McICA noise may be model dependent. Indeed, McICA's success depends on these impacts being 'negligible'. Therefore, we are inviting GCM groups to participate in an experiment that aims to assess the impacts of McICA's noise. To date there are six GCMs involved in this intercomparison. In addition to results from CAM-1.8, this poster documents the intercomparison agenda.
This poster will be displayed at the ARM Science Team Meeting.