Process-based evaluation of LES and SCM simulations using in-situ observations and 
satellite retrievals and implications for GCM cloud parameterizations.

Jasmine Remillard, Ann Fridlind, George Tselioudis, Andy Ackerman - NASA GISS
Pavlos Kollias -McGill University 
Ed Luke  - BNL

Two low cloud periods from the 19-month deployment of the ARM AMF at the 
Azores are simulated with the DAHRMA LES and the GISS SCM. The two cases are 
selected through a cluster analysis of ISCCP cloud property matrices, so as to 
represent the two low cloud Weather States that the GISS GCM severely 
underpredicts in that region and, as it turns out, in the global domain.  The two 
selected cases represent shallow cumulus clouds occurring in cold air outbreaks 
behind cold frontal passages, and stratocumulus clouds occurring when the region 
is dominated by the Azores high-pressure system. The LES simulation captures well 
the difference in the horizontal structure of the two low cloud fields but requires 
improvement in the simulation of the microphysical cloud properties. The SCM 
simulation, when run at the native vertical resolution, fails to simulate the shallow 
cumulus case and to maintain the stratocumulus case, providing insight into the 
reasons for the GCM cloud deficiencies. Sensitivity tests with increased SCM vertical 
resolution at the boundary layer show improvement in the representation of cloud 
amount for both cases but still include severe errors in the simulation of cloud 
microphysical properties. Further tests are under way to examine the sensitivity of 
both the LES and SCM results to changes in the parameterization of cloud 
microphysical properties. In parallel, GCM runs are performed to test the effects of 
any positive changes to the global cloud field. The methodology followed in this 
study, with the process-based examination of different time and space scales in both 
models and observations, is tested as a prototype for GCM cloud parameterization 
improvements.