CGILS Updated Results GCSS-BLCWG Meeting, September 29-30, 2010 KNMI Minghua Zhang (Stony Brook University) Julio Bacmeister, Sandrine Bony, Chris Bretherton, Florent Brient, Anning Cheng, Stephan de Roode,Tony Del Genio, Charmaine Franklin, Chris Golaz, Cecile Hanny, Francesco Isotta, In-Sik Kang, Hideaki Kawai, Martin Koehler, Suvarchal Kumar, Vince Larson, Adrian Lock, Ulrike Lohman, Marat Khairoutdinov, Andrea Molod, Roel Neggers, Sing-Bin Park, Ryan Senkbeil, Pier Siebesma, Colombe Siegenthaler-Le Drian, Bjorn Stevens, Max Suarez, Kuan-man Xu, Mark Webb, Audrey Wolfe, Ming Zhao,

GPCI S6 S11 s12

Purpose: To understand the causes of cloud feedbacks, and thus climate sensitivities of climate models. Objectives: 1.To understand the physical mechanisms of cloud feedbacks in SCMs 2.To interpret GCM cloud feedbacks by using SCM results 3.To Evaluate the SCM cloud feedbacks using LES simulations

(moist adiabat) T(z) RH Fixed Warm PoolCold Tongue T(z) (Zhang and Bretherton, 2008) CGILS (CFMIP-GCSS Intercomparison of Large-Eddy and Single-Column Models) Need to be relevant to observations and GCMs

SCM (16) CAM4* CAM5* CCC* CSIRO* ECHAM5* ECHAM6* ECMWF* GFDL GISS* GSFC* JMA* KNMI-RACMO* LMD* SNU UKMO* UWM* LES (5) DALES SAM UCLA UCLA/LaRC UKMO * Indicates SCMs that completed the revised runs

The second round of SCM simulations Forcing revised to be the same as in LES Control case temperature and relative humidity are now directly from ECMWF Interim Analysis for July 2003 from Martin Koehler. 1. Lower troposphere above the boundary layer is warmer, with more realistic inversion height and strength. 2. More moisture in the boundary layer for LES to start with. 3. Subsidence extended to below 1000 mb, thus slightly stronger than before below 900 mb.

Cloud feedbacks at S6

Round 1 versus Round 2

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12

Points to Note Some models maintain the same sign of feedbacks or negligible feedback at all three locations Negative feedback: CAM4, ECMWF, JMA, UWM Positive feedback: CAM5, CCC, CSIRO, ECHAM6, LMD, UKMO Flipped: GISS, GSFC, RACMO (in the same direction: positive at s6, negative at s12). Feedbacks at the two locations of S11 and S12 mostly show the same sign in the models

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12

Points to Note At S6, Two models show little feedback (CSIRO and UKMOL38) The small group show negative feedback (CAM4, ECMWF, JMA UWM) The majority show positive feedback (CAM5, CCC, ECHAM6, GISS, GSFC, LMD, RACMO) At S11, Two models show little feedback (ECHAM6 and LMD) About half of the models show positive feedback (CAM5, CCC, CSIRO, RACMO, UKMO) Half of the models show negative feedback (CAM4, ECMWF, GISS, GSFC, JMA, UWM) At S12, Two models show little feedback (ECHAM6 and LMD) Three models show positive feedback (CCC, CSIRO, UKMO) The majority show negative feedback (CAM4, CAM5, ECMWF, GISS, GSFC, JMA, RACOM,UWM)

Instead of discussing clouds at one location at a time (the GCSS perspective), in the following, we will show all cloud types for one model at a time (the GCM perspective)

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 CAM4

CAM4 s6 ctl CAM4 s6 p2k CAM4 s11 ctlCAM4 s11 p2k

CAM4 s12 ctlCAM4 s12 p2k CAM4 ql s6 CAM4 ql s11CAM4 ql s12

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 CAM5

CAM5 s6 ctl s6 p2k s11 ctl s11 p2k

CAM5 s12 ctls12 p2k ql at s6 ql at s11ql at s12

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 ECHAM6

ECHAM6 s6 ctl s6 p2k s11 ctl s11 p2k

ECHAM6 s12 ctls12 p2k ql at s6 ql at s11ql at s12

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 ECMWF

ECMWF s6 ctl s6 p2k s11 ctl s11 p2k

ECMWF s12 ctls12 p2k ql at s6 ql at s11ql at s12

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 ECMWF2

ECMWF2 s6 ctl s6 p2k s11 ctl s11 p2k

ECMWF2 s12 ctls12 p2k ql at s6 ql at s11ql at s12

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 LMD

LMD s6 ctl s6 p2k s11 ctl s11 p2k

LMD s12 ctls12 p2k ql at s6 ql at s11ql at s12

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 RACMO

RACMO s6 ctl s6 p2k s11 ctl s11 p2k

RACMO s12 ctls12 p2k ql at s6 ql at s11ql at s12

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 UKMOL38

UKMOL38 s6 ctl s6 p2k s11 ctl s11 p2k

UKMOL38 s12 ctls12 p2k ql at s6 ql at s11ql at s12

 CRF at S6  CRF at S11  CRF at S12 S6 S11 S12 UWM

UWM s6 ctl UWM s6 p2k UWM s11 ctlUWM s11 p2k

UWM s12 ctls12 p2k ql at s6 ql at s11ql at s12

Negative feedbacks Deepened mixed layer without much dilution at the cloud top: 1.no explicit cloud-top entrainment and no shallow convection 2.weak cloud-top entrainment and small moisture effect of shallow convection 1000mb 900mb 950mb 800mb 1010mb

Negative feedbacks Deepened mixed layer without much dilution at the cloud top: 1.no explicit cloud-top entrainment and no shallow convection 2.weak cloud-top entrainment and small moisture effect of shallow convection 1000mb 900mb 950mb 800mb 1010mb

1000mb 900mb 950mb 800mb 1010mb Positive feedbacks 1.More intermittent clouds due to cloud-top entrainment or shallow convection 2.Less cloud water due to cloud-top mixing

Summary 1.Evidences suggest that the CGILS results from some models can be used to interpret GCM cloud feedbacks. For other models, especially those displaying multiple equilibrium behavior, alternative configuration of running the SCMs may be insightful. This was demonstrated by the LMD group who added transient forcing to break up the multiple equilibriums. 2.Two groups of models have been identified to show the same sign or negligible cloud feedbacks at all three locations, displaying positive and negative feedbacks. It would be interesting to see if the corresponding GCMs show the same sign of cloud feedbacks. 3.The deciding physical process appears to be the mixing at the cloud top, carried out either by explicit cloud-top entrainment or shallow convection.

Summary 4.Comparisons with LES and with GCMs are yet to be carried out. SCMers are waiting for LES and CFMIP results.

Opposite: CSIRO ECMWF GISS KNMI Cloud feedbacks at S11 Round 1 versus Round 2

Opposite: ECMWF GISS Cloud feedbacks at S12 Round 1 versus Round 2

CGILS SCM Next Steps: 1. Understanding of processes in each model. 2. Re-configure simulations, such as adding transience, if results are suspicious. 3. Linking with GCMs 4. Use LES models to say something about the SCMs

CGILS LES Next Steps 1. At s11, reduce vertical resolution to 5 m for all models to seek consistent LES results 2. At s12, either the forcing data or the simulation setup will be revised to enable the LES models maintain stable cloud layer.

CGILS Future Plan 1. Use seasonal, interannual, and decadal variations to evaluate the models 2. Close coordination with CFMIP to conduct SCM&GCM sensitivity experiments 3. Use LES results to constrain the SCM parameterizations

CGILS Paper Options Wait for LES results Wait for some CFMIP results SCM results written up first Next meeting: A telecon before the end of 2010 A CGILS session at the next EUCLIPSE general meeting