A Neural-Network based Cloud-Fraction Parameterization Scheme with Adaptivity for Different Resolutions
Abstract
Clouds are in the sub-grid scale for most general circulation models. Therefore, the parameterization of cloud fraction poses a significant challenge in climate modeling. Here we present a neural-network based cloud fraction scheme that implicitly considers the effects of both horizontal and vertical grid sizes. The scheme is trained, tested, and evaluated using the CloudSat cloud data and the associated auxiliary meteorology data. Primary results show that the scheme can simulate realistic spatial distribution of total cloud fraction and cloud vertical structure while possessing good computational efficiency and adaptivity for varying grid sizes. Details on the scheme development and its performance will be presented.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFM.A16D..06C