I ran the UCLA large eddy simulation (LES) model to generate a wide variety of marine boundary layer cloud structure for developing cloud remote sensing methods using 3D radiative transfer. Bjorn Stevens, the author of the UCLA LES model, helped me with running the model. I initialed the LES model with atmospheric profiles from random columns at specific times from the ECMWF ERA-Interim reanalysis.

Description of LES model:

- UCLA LES version 2.0; model developed by Bjorn Stevens and distributed here.
- My modifications: read in the assumed constant large scale forcing (vertical velocity and temperature and moisture advection profiles) from a file; modified the Monte Carlo spectral integration radiation scheme to allow calling the radiation code for a column with a specified probability to speed up the calculations; added a netcdf LWC output file.

Description of the simulation procedure:

- Data source: ERA-Interim reanalysis; 1.5
^{o}x 25 mb x 6 hrs resolution; surface and pressure level parameters from 600 to 1000 mb. - Location: North east Pacific: 0N to 45N, 150W to 105W, oceanic columns
- Times: 18Z on 2007-05-21, 2007-06-21, 2007-07-22, 2007-08-22, 2007-09-23.
- ERA column selection:
- CAPE: 0 to 500 J/kg
- BL max cloud fraction: 0.10 to 1.00
- Mid-layer max cloud fraction: 0.0 to 0.01

- Simulation time: 8:00 to 14:00 local time
- Horizontal grid: 160x160 at 50 m spacing
- Domain top: 500 m above first ERA level with potential temperature more than 10 K above that at 1000 mb.
- Vertical grid: Variable grid; 40 m default spacing; 20 m spacing at surface; 10 m spacing around imposed jumps or 20 m spacing around the max dtheta/dz ERA layer if dtheta/dz>10 K/km; 400 m wide high-resolution region (75% above jump) for imposed jumps, otherwise 250 m wide region centered on layer.
- Vertical jump in sounding: 10 m wide jump imposed at ERA boundary layer height if the two ERA layer (50 mb) potential temperature difference is > 5 K and the max BL cloud fraction > 0.80.
- Large scale forcing: zero temperature and total water advection; large scale vertical velocity is from ERA.
- Surface fluxes: LES water surface parameterization
- Sea surface temperature: Initial SST set ERA SST; SST changes linearly with time according to ERA SST gradient and mean wind speed for a Lagragian-type simulation.
- Microphysics: fixed cloud droplet concentration from random CCN
with an exponential distribution between 40 and 400 cm
^{-3}; two-moment drizzle/rain bulk microsphysics scheme (see Stevens and Seifert 2008). - Radiation: Fu/McISA independent column radiation code (Pincus and Stevens, 2009) , randomly sampled 1/3 of the time for speed up, for longwave and shortwave radiative transfer.
- Subgrid scale mixing: Smagorinsky closure; turned off SGS mixing for scalars.

Click on image for a web page with graphs for each set of five simulations.

At the end of each web page below there are links to netcdf files with the 3D fields of liquid water content. These files may be read with les2lwc.f90 .