CAM6/CESM2 development simulations

CAM6/CESM2
development simulations
Cécile Hannay, Rich Neale, Pete Bogenschutz,
Julio Bacmeister, Andrew Gettelman, and Joe Tribbia
Thanks to: David Bailey, Cheryl Craig, Gokhan Danabasoglu, Brian Eaton,
Jim Edwards, Marika Holland, Jean-Francois Lamarque, David Lawrence,
Keith Lindsay, Keith Oleson, Bill Sacks, John Truesdale, Mariana Vertenstein,
and gazillions of others.
CESM 2: Development simulations
http://www.cesm.ucar.edu/working_groups/Atmosphere/development/cesm1_5/
At a glance
• Huge team effort started
in Mid November 2015
• 2 co-chair meetings/week
Feb 2016: Mini Breck
• 34 experiments (“cases”)
• 1300+ years of simulations + diagnostics
June 2016: Breckenridge workshop
• 94 experiments (“cases”)
• 2890+ years of simulations + diagnostics
• 271 T
And also
• Many standalone simulations in
individual working groups
AMWG homework
Following mini-Breck, we identified biases to be targeted
• Taylor score lower than in LENS
• Cold SSTs
=> Adjust tuning to reduce SWCF
• Underestimated Amazon precipitation => ZM changes
• Insufficient Tropical Sub-Seasonal Variability => ZM changes
• Weak Surface winds => Beljaars and ridge scheme (TMS off)
• Strong indirect effects => New autoconversion
• Excessive humidity (RHliq > 105%) => Fix in CLUBB
• Poor polar stratospheric clouds => New ice microphysics
How did we do on our homework ?
CESM 2: Development simulations
In this talk, we will compare:
• LENS: Large Ensemble
• CESM1.5: Mini Breck - Feb 206 (This is 28 or 36)
• CESM2.0-: Now – June 2016 (This is 79).
Are you lost in
translation ?
Taylor Diagram
• LENS
• Now
• Mini-Breck
Significant improvement in Taylor score since mini-Breck
Current simulation is better than LENS
Sea Surface Temperature (SST) bias (ANN)
LENS
Bias = -0.24K
RMSE = 0.91
Mini-Breck
Bias = -0.62K
RMSE = 1.12
Now
Bias = -0.58K
RMSE = 1.19
Sea Surface Temperature (SST) bias (ANN)
LENS
Bias = -0.24K
RMSE = 0.91
Mini Breck
Bias = -0.62K
RMSE = 1.12
Now
Bias = -0.58K
RMSE = 1.19
Not much improvement in SST bias.
Cold bias in Labrador sea (covered by sea-ice).
Sea ice thickness (ANN)
LENS (yrs 475-499)
Mini-Breck (yrs 75-99)
Now (yrs 72-91)
Sea-ice thicker over Central Arctic
Ice-covered Labrador sea
Sea ice thickness in Labrador sea
New atm+ocn+lnd (79)
Now
Cause for changes still uncertain,
but appears related to coupling of
new ocean and new atmosphere
New atm+ocn (66)
New atm + ocn (64)
New atm only (63) New ocn Only (MP3)
Run from Feb (36)
Courtesy Marika Holland
Mini-Breck
Precipitation bias versus GPCP (ANN)
LENS
Bias = 0.37
RMSE = 1.13
(mm/day)
Mini-Breck
Bias = 0.19
RMSE = 1.12
(mm/day)
Now
Bias = 0.19
RMSE = 0.91
(mm/day)
Precipitation bias versus GPCP (ANN)
LENS
Bias = 0.37
RMSE = 1.13
(mm/day)
Mini-Breck
Bias = 0.19
RMSE = 1.12
(mm/day)
Now
Bias = 0.19
RMSE = 0.91
(mm/day)
Improved precip bias
Better precip over Amazon
Improvement in land precipitation
Color code
Green means better
Red means worse
Between Mini Breck and Now: 20% reduction in RMSE
Courtesy Dave Lawrence
Equatorial Waves (rainfall)
Wavenumber frequency decomposition of precipitation (10S-10N), annual mean
Now
Obs
Kelvin
Improved
Kelvin waves
MJO
LENS
Mini-breck
Lack of
Kelvin waves
Courtesy Rich Neale
Sea-level pressure versus MERRA (ANN)
LENS
Bias = 0.29
RMSE = 1.61
(mbar)
Mini Breck
Bias = 0.09
RMSE = 3.02
(mbar)
Now
Bias = 0.28
RMSE = 1.76
(mbar)
Sea-level pressure versus MERRA (ANN)
LENS
Bias = 0.29
RMSE = 1.61
(mbar)
Mini-Breck
Bias = 0.09
RMSE = 3.02
(mbar)
Now
Bias = 0.28
RMSE = 1.76
(mbar)
Improved SLP
in Southern Ocean
Change in surface winds (ANN)
Weak surface winds in LENS and in mini-breck simulation
New beljaars/ridge scheme increase surface winds
Increase of the surface winds between mini-Breck and Now
Greenland and Antartica surface winds
Mini-Breck
Greenland
Antarctica
Courtesy Lenaerts/Lipscomb
Now
Obs
Aerosol effect
Direct (W/m2)
IPCC values
Indirect (W/m2)
-0.5 [-0.9 to -0.1] -0.7 [-1.8 to -0.3]
LENS
-0.2
-1.4
Mini Breck
-0.4
-1.8
Now
-0.4
-1.5
Mini-Breck:
indirect effect too strong
Now:
New autoconversion reduced it
20th Century warming
Mini-Breck:
not enough warming over 20thC
Now:
No 20th century simulation yet
Expecting impact of reducing
indirect effect
How did we do on our homework ?
Following mini-Breck, we identified biases to be targeted
• Taylor score lower than in LENS => Improved
• Cold SSTs => Not improved (degraded Labrador sea)
• Underestimated Amazon precipitation => Improved
• Insufficient Tropical Sub-Seasonal Variability => Improved
• Weak Surface winds => Improved
• Strong indirect effects => Improved
• Excessive humidity (RHliq > 105%) in CLUBB => Improved
• Poor representation of polar stratospheric clouds=> Improved
Remaining issues in current simulations
• Ice too thick over central Arctic and ice-covered labrador sea
This has been 8 months of intense work
We had good days
We had bad days
We always find the cause of the problem