Investigating the Variability in Cumulus Cloud Number as a Function of Subdomain Size and Organization using large-domain LES
Abstract
Recent advances in supercomputing have introduced a "grey zone" in the representation of cumulus convection in general circulation models, in which this process is partially resolved. Cumulus parameterizations need to be made scale-aware and scale-adaptive to be able to conceptually and practically deal with this situation. A potential way forward are schemes formulated in terms of discretized Cloud Size Densities, or CSDs. Advantages include i) the introduction of scale-awareness at the foundation of the scheme, and ii) the possibility to apply size-filtering of parameterized convective transport and clouds. The CSD is a new variable that requires closure; this concerns its shape, its range, but also variability in cloud number that can appear due to i) subsampling effects and ii) organization in a cloud field. The goal of this study is to gain insight by means of sub-domain analyses of various large-domain LES realizations of cumulus cloud populations. For a series of three-dimensional snapshots, each with a different degree of organization, the cloud size distribution is calculated in all subdomains, for a range of subdomain sizes. The standard deviation of the number of clouds of a certain size is found to decrease with the subdomain size, following a powerlaw scaling corresponding to an inverse-linear dependence. Cloud number variability also increases with cloud size; this reflects that subsampling affects the largest clouds first, due to their typically larger neighbor spacing. Rewriting this dependence in terms of two dimensionless groups, by dividing by cloud number and cloud size respectively, yields a data collapse. Organization in the cloud field is found to act on top of this primary dependence, by enhancing the cloud number variability at the smaller sizes. This behavior reflects that small clouds start to "live" on top of larger structures such as cold pools, favoring or inhibiting their formation (as illustrated by the attached figure of cloud mask). Powerlaw scaling is still evident, but with a reduced exponent, suggesting that this behavior could be parameterized.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2017
- Bibcode:
- 2017AGUFM.A23L..05N
- Keywords:
-
- 3307 Boundary layer processes;
- ATMOSPHERIC PROCESSES;
- 3310 Clouds and cloud feedbacks;
- ATMOSPHERIC PROCESSES;
- 3311 Clouds and aerosols;
- ATMOSPHERIC PROCESSES;
- 3314 Convective processes;
- ATMOSPHERIC PROCESSES