Multi-year evaluations of a cloud model using ARM data
Peter W. Henderson and Robert Pincus
Journal of the Atmospheric Sciences, Vol. 66, pages 2925-2936. doi:10.1175/2009JAS2957.1

Abstract

This work uses long-term lidar and radar retrievals of the vertical structure of cloud at the Atmospheric Radiation Measurement (ARM) program's South- ern Great Plains (SGP) site to evaluate cloud occurrence in multi-year runs of a cloud system resolving model (CSRM ) in three configurations of varying reso- lutions and sophistications. The model is nudged to remain near the observed thermodynamic state and model fields are processed to mimic the operation of the observing system. We evaluate the model's skill in predicting cloud occur- rence using both traditional performance measures that assume ergodicity and probabilistic measures which do not require temporal averaging of the observa- tions.

The model shows considerable skill in predicting cloud occurrence when its thermodynamic state is close to that observed. The overall bias in modeled 1Cloud-scale models are known to be more skillful at predicting the inter- actions between clouds and their environment than are the parameterizations used in global models. This skill has motivated the development of multi-scale models, in which the parameterizations in the global scale model are replaced with a cloud-scale model running in each column of the global model. Though conceptually attractive, these models rely on the skill of the cloud-scale models running at coarse resolution in a range of thermodynamic states well beyond those in which the cloud-scale models have been tested. This motivates the de- velopment of techniques for evaluating model skill across a similarly wide range of circumstances.

Download a reprint (PDF format) or read the article on-line at the journal web site.
Further reproduction or electronic distribution is not permitted.