“…Nevertheless, there are important advantages that a 3‐D MC treatment has over 1‐D parametrizations: - full separation of the RT model from descriptions of attenuating media: descriptions of subgrid‐scale variations of the Earth‐atmosphere system rest with stochastic generators that are relatively easy to alter and adjust [cf. Räisänen et al ., ];
- account of 3‐D RT effects: assuming that regime‐dependent 3‐D RT effects are important, their biases and accuracies are limited only by the credibility of the stochastic generator;
- handling of scattering properties: MC RT solutions are effectively infinite‐stream models that can faithfully include arbitrarily detailed descriptions of particle scattering phase functions that vary throughout arbitrarily complex atmospheres [see Barker et al ., ];
- measurement simulation: if the local estimation method is used for the MC solution [ Marchuk et al ., ], fluxes and radiances can be computed straightforwardly, and efficiently, thereby ensuring consistency between modeled and diagnostic radiative quantities;
- scale independence I: MC RT models are scale independent so as the sizes of a GCM's grid‐spacings change, only the stochastic generator needs to be altered, not the RT model;
- scale independence II: as GCM horizontal grid‐spacings rival typical cloud cell sizes, such as in high‐resolution domains of weather prediction models [e.g., Leroyer et al ., ; Vionnet et al ., ], subgrid‐scale cloud generators can be jettisoned leaving the 3‐D MC RT model that operated on stochastically generated atmospheres to operate on domains provided by the host GCM.
…”