stratocumulus clouds in a multiscale modeling framework

Stratocumulus Clouds in a Multiscale Modeling FrameworkCharlotte A. DeMott, David A. Randall, and Marat Khairoutdinov

Department of Atmospheric Science, Colorado State University

Summary: The cumulus parameterization in theCAM3 GCM has been replaced with an embeddedCRM. The relatively coarse 4-km CRM gridresolution in the MMF produces physically realisticmarine stratocumulus clouds with improved diurnalcycle over the traditional parameterization method,but low cloud fractions are too low in some areas.

1. IntroductionThe traditional cumulus parameterization in theCommunity Atmospheric Model (CAM3) wasreplaced with embedded 2D cloud resolvingmodels (CRMs) in each GCM grid column, a“multiscale modeling framework” (MMF).Computational constraints limit CRM resolution tojust 4-km in the x-direction, much coarser thantypical CRM resolutions used to simulatestratocumulus clouds.

Can the MMF adequately simulate thisimportant cloud type?

2. The Pacific Cross SectionSeasonal mean maps and cross sections alongthe GCSS Pacific Cross Section:

DJF OLR, 1998CAM MMF OBS

DJF Cloud Cover (%), 1998

DJF vertical velocity, 1998

DJF cloud fraction, 1998

MMF cloud fractions are lower than those inCAM and those derived from observations, butother large-scale fields generally agree wellwith observations.

3. High Resolution Analysis“Looking inside the model…”

Hourly CRM-scale output was saved for a 5-dayperiod (July 1-5, 1998) to examine the fine-scalestructure of simulated stratus clouds.

Mean diurnalcycle of thestratocumuluscloud field at29N, 227W: a)cloud liquidwater (green,every 0.1 g kg-1),longwave cooling(blue, 15, 20, and25 K/day), andvertical velocityvariance (black,every 0.01 m2/s2).b) rainfall(green) andnormalizedvertical velocityskewnessaveraged overthe sub-cloudlayer.

Diurnalcycle of theprobabilitydistributionfunction(PDF) forthe height oflapse ratesexceeding0K/km inthe PacificCrossSectiontransitionzone (20N,215E).Contourinterval is10%.

Mean zonalMMF resolved(left) and sub-grid scale (right)fluxes of non-precipitatingwater for July 1-5, 1998. Contourinterval is 0.5 x10^5 kg/m2/s. A map of the time-averaged ratio of the SGS non-

precipitating water vapor fluxes to the total (resolved + SGS)fluxes computed by the MMF’s CRM at ~975 mb level forJuly 1-5, 1998. The Pacific cross section is shown as a blackline going from the coast of California towards the equator.

Simulated clouds:

• driven by radiative cooling

• have realistic vertical velocity skewness profiles.

• improved diurnal variability compared to the CAM.

• non-precipitating water fluxes handled by the CRM,and not the sub-gridscale parameterization.

4. Global marine stratocumulusDo MMF-produced marine stratocumulus cloudsexhibit the same sensitivity to lower troposphericstability as seen in numerous observationalstudies?

Results from two 14-year AMIP runs let uscompare model output to the “Klein Line” (Kleinand Hartmann, 1993). CAM low cloud isparameterized using the Klein line. MMF lowcloud is produced according to the embeddedCRM dynamics and thermodynamics.

Above: Stratus cloud regions and seasons analyzed in Kleinand Hartmann, 1993.

Right: Scatter plot of monthly mean lower-tropospheric stability(θ700 - θsfc) versus low cloud fraction for the areas and seasonsdefined above. Klein line is shown in blue.

MMF stratus cloud sensitivity to lower troposphericstability is good in the Northern Hemisphere, butless so in the descending regions of the Hadleycell in the Southern Hemisphere.

References

Klein, S. A. and D. L. Hartmann, 1993: The seasonal cycle oflow stratiform clouds. J. Climate, 6, 1587-1606.

Acknowledgements

This work was funded by ARM Grant #DE-FG02-02ER63370.