A Missing Link
Coupling clouds to radiation
Greg Thompson, NCAR-RAL & DTC
25 Jun 2013
With excellent assistance from:
Mukul Tewari, NCAR-RAL
Shaowu Bao, Ligia Bernadet NOAA-ESRL/GSD
Sam Trahan, NCEP-EMC
Supported by
STEP
HWRF
Short Term Explicit Prediction
DTC/NOAA/NCAR
• OU-CAPS ARW ensembles
• SPC/NSSL Hazardous Weather Testbed
• old MCS decaying cloud mostly composed
of snow was essentially transparent
• next day’s convection triggered much too
early since minimal cloud cover seen by
radiation
• caused by cloud ice versus snow
categorization
Motivation…
• Implement/test Thompson microphysics
scheme in HWRF
• existed but essentially untested in HWRF
• scheme was recently implemented in
COAMPS (Yi Jin, NRL)
• initial tests in COAMPS improved
hurricane forecasts compared to legacy
scheme
Case: large winter cyclone
Visible satellite image 17:45UTC 01 Feb 2011
Clear-sky, snow
cover
Clear-sky, forest, snow
Lake-effect snow
Deep, snowing cloud
Thin/partial cloud,
mostly ice-phase
Deep, mixed-phase
cloud
Shortwave reaching ground
HWRF Test0: Ferrier microphysics & GFDL radiation
broad, thin ice cloud
minimal lake-effect
missing all clouds
Clouds too opaque?
0-15 W/m2 ?
Shortwave reaching ground
HWRF Test1: Thompson microphysics & GFDL radiation
GFDL/Goddard
radiation schemes
essentially ignore
snow, only care
about cloud ice.
There is near
blizzard-like
conditions here
correctly predicted in
the model and yet no
reduction of
shortwave reaching
the ground.
Shortwave reaching ground
HWRF Test2: Thompson microphysics & RRTMG radiation-uncoupled
RRTMG has internal
assumptions about
size of cloud
droplets, ice, and
snow; NOT coupled
with what is known in
microphysics
scheme(s).
RRTMG does
significantly better
with all clouds, but
still can be improved.
Shortwave reaching ground
HWRF Test3: Thompson microphysics & RRTMG radiation-coupled
Properly connecting
effective radii of
cloud water, ice, and
snow (passing from
microphysics to
RRTMG).
Thin, mostly ice
clouds become
slightly more
opaque?
Deepest clouds
become slightly less
opaque?
Current code issues
RRTMG: Combined cloud ice and snow variables
GFDL & Goddard: only Qice, neglect Qsnow
Ramifications of adding ice and snow!
Look at Slide#10, table of look-up values of assumed ice radius. Dependence on
temperature. Therefore, if there exists 0.5g/kg of snow (typical in deep synoptic winter
storm) plus 0.1-0.2 g/kg of cloud ice, up near tropopause/anvil level, then this combined
mass will have a very small diameter (massive impact to radiation) as compared to using
larger ice crystal size or lower mass of the small crystals.
Current code issues
Cloud water radii
Current code issues
Cloud ice radii
New treatment ice/snow path
For ice, go back to ice content only; for snow, reduce mass by inversely scaling with diameter
Modifications to WRF (v3.4.1)
Should be part of v3.5.1 release (late Summer 2013)
Registry (.EM_COMMON, NMM, NMM_NEST,
HWRF)
dyn_em/start_em.F
dyn_em/solve_em.F
dyn_nmm/module_PHYSICS_CALLS.F
phys/module_physics_init.F
phys/module_microphysics_driver.F
phys/module_mp_thompson.F
dyn_em/module_first_rk_step_part1.F
phys/module_radiation_driver.F
phys/module_ra_rrtmgsw.F
phys/module_ra_rrtmglw.F
added new 3D variables: re_cloud, re_ice, re_snow
added new switch variables: has_reqc, has_reqi, has_reqs
pass new top-level variables to physics_init routine
pass the top-level variables to micro_driver
pass variables into one or more microphys routines
calculate effective radii for cloud, ice, snow
pass the top-level variables to radiation_driver
calculate cloud optical depth from new radii
only if has_reqX=1, otherwise, unaltered code!
Also possible to do similar for nearly all other microphysics choices, especially Morrison,
Milbrandt, WSM6, WDM6, etc.
Thus far (May 2013), fixed code coupling Thompson MP & RRTMG schemes only
submitted to MMM for version 3.5.1 release ~Aug2013. This altered code is a member
of OU-CAPS spring experiment ensembles (“arw_25”).
Proof it is working (cloud drop size)
Radiative effective radius: cloud droplets
(k=16 from bottom)
Proof it is working (cloud ice size)
Radiative effective radius: cloud ice
(k=44 from bottom)
Proof it is working (snow size)
Radiative effective radius: snow
(k=37 from bottom)
Shortwave reaching ground
HWRF Test2: Thompson microphysics & RRTMG radiation-uncoupled
RRTMG has internal
assumptions about
size of cloud
droplets, ice, and
snow; NOT coupled
with what is known in
microphysics
scheme(s).
RRTMG does
significantly better
with all clouds, but
still can be improved.
Shortwave reaching ground
HWRF Test3: Thompson microphysics & RRTMG radiation-coupled
Properly connecting
effective radii of
cloud water, ice, and
snow (passing from
microphysics to
RRTMG).
Thin, mostly ice
clouds become
slightly more
opaque?
Deepest clouds
become slightly less
opaque?
Microphysics & Radiation
Test3 minus Test2
showing RRTMG coupled versus uncoupled
Previous slides
showed sensitivity
with WRF-NMM, this
graphic created from
WRF-ARW with 4-km
grid spacing
changing only the
coupling of RRTMG
with Thompson
microphysics.
Thin, mostly ice
clouds become
slightly more
opaque?
Deepest clouds
become slightly less
opaque?
Aerosols, Microphysics & Radiation
Preview of next steps
Cloud droplet number concentration difference from run with 10X aerosols minus 1X aerosols
Higher aerosol
concentration leads
to more numerous
cloud droplets.
[Single model
vertical level, k=10,
shown as example.]
Aerosols, Microphysics & Radiation
Preview of next steps
Cloud droplet radiative effective size difference from run with 10X aerosols minus 1X aerosols
Higher aerosol
concentration leads
to more numerous
cloud droplets that
causes overall
decrease in effective
size. [Single model
vertical level, k=10,
shown as example.]
Aerosols, Microphysics & Radiation
Preview of next steps
Cloud albedo (shortwave, upward, top-of-atmos difference from run with 10X aerosols minus 1X aerosols
Smaller cloud droplet
size produced by
having more
aerosols leads to
larger cloud albedo
(Twomey, 1974)
illustrating the first
aerosol indirect
effect.
See
Poster#84
Testing:
• Hurricane Earl (2010), Hurricane Sandy (2012)
• ~28 days May/Jun 2013 in OU-CAPS ensemble members
• DTC project to test in HFIP re-runs
Next steps:
• Compute water/ice sizes in other microphysics schemes to
couple directly with RRTMG
• Compare WRF output to SurfRad (other) radiation
measurements
• Using coupled “aerosol-aware” Thompson microphysics,
investigate aerosol indirect effects
Thank you, especially:
Mukul, Ligia, Laurie, Shaowu