Séminaire Vendredi 25 Nov 2005 11h00 / Seminar Friday Nov 18th 11h00 

Conférencier/Lecturer: Howard Barker and Jason Cole
                       Cloud Physics and Severe Weather Research Division

        Sujet/Subject: Assessing the Monte Carlo Independent Column
                       Approximation McICA) in Large-scale Atmospheric Models

         Presentation: Anglais/English
            Lieu/Room: Grande salle du premier étage CMC 

                 iweb: http://iweb.cmc.ec.gc.ca/rpn/SEM 
                  web: http://collaboration.cmc.ec.gc.ca/science/rpn/SEM/index.php

                  -------------------------------------------------- 
                                 RESUME / ABSTRACT 

This seminar reiterates briefly the rationale behind the Monte Carlo
Independent Column Approximation (McICA) method for computing broadband
radiative fluxes in large-scale global models. Unlike most radiative
transfer (RT) parametrizations used in large-scale atmospheric models,
the McICA approach separates the RT solver from the description of
unresolved fluctuations in optical properties. This separation results
in flexible and simple RT coding, and execution times that are similar
to conventional RT models. Moreover, it produces flux estimates that are
unbiased with respect to the full ICA. It does, however, produce highly
conditional random noise.

McICA has been implemented in the NCAR CAM-3 GCM as well as the CCCma
GCM. Versions of McICA with excessive random noise have locally
significant impacts on CAM-3. As the magnitude of McICA noise
diminishes, so too does McICA's impact. The CCCma GCM, on the other
hand, is essentially insensitive to McICA's random noise; even when the
magnitude of the noise is large.

We are beginning a GCM intercomparison study that aims to deduce the
GCM-dependence of McICA noise. The plan is to run ensembles of 15-day
simulations to see how various GCMs using various levels on McICA noise
diverge from near-noiseless reference simulations. Our hypothesis is: if
after 15 days there is no discernible sign that McICA-generated noise
has impacted the large-scale dynamics and other physical fields, then it
is unlikely that McICA noise will have an impact on longer timescales.
This approach resembles the methods used to test GCM parametrizations
employed by the U.S. DoE's Climate Change Prediction Program (CCPP) and
Atmospheric Radiation Measurement (ARM) Program CCPP-ARM
Parameterization Testbed (CAPT) study.