!the!stochas+c!mul+cloud!model!(smcm)! as!partof!an ...€¦ ·...

17
The Stochas+c Mul+cloud Model (SMCM) as part of an opera+onal convec+on parameterisa+on in a comprehensive GCM Karsten Peters 1,2 , Chris+an Jakob 1 , Benjamin Möbis 2,1 EGU GA 2015, Session AS1.13, 17 Apr 2015 SMCM development: Boualem Khouider and Andrew J Majda 1 ARC Centre of Excellence for Climate System Science, Monash University, Australia 2 Max Planck Ins+tut für Meteorologie, Hamburg Germany

Upload: others

Post on 15-Nov-2020

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

 The  Stochas+c  Mul+cloud  Model  (SMCM)  as  part  of  an  opera+onal  convec+on  

parameterisa+on  in  a  comprehensive  GCM  

Karsten  Peters1,2,  Chris+an  Jakob1,  Benjamin  Möbis2,1    

EGU  GA  2015,  Session  AS1.13,  17  Apr  2015      

SMCM  development:  Boualem  Khouider  and  Andrew  J  Majda  

1ARC  Centre  of  Excellence  for  Climate  System  Science,  Monash  University,  Australia  2Max  Planck  Ins+tut  für  Meteorologie,  Hamburg  Germany  

Page 2: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

Rainfall in current generation GCMs (IPCC AR5 status) CMIP5  precip  vs  GPCP  (IPCC  AR5  (2013))  

Projected  precip  change  2081-­‐2100  w.r.t.  1986-­‐2005,  RCP8.5  scenario  (IPCC  AR5  (2013))  

Page 3: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

The future of convection schemes?

Our  sugges(on    

i.  stochas+c  framework,    ii.  which  incorporates  memory  effects  and    iii.  convec+ve  organisa+on/aggrega+on    

Such  a  tool  already  exists….  

Page 4: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

The Stochastic Multi-Cloud Model (SMCM) (Khouider  et  al  2010)

-  divides  a  large-­‐scale  domain  into  independent  sub-­‐domains  

-  predicts  an  ensemble  of  congestus,  deep  and  stra+form  cloud  

Ø transi+ons  determined  by  stochas(c  Markov-­‐Chain  process  

Ø area  frac(ons  per  cloud  type  -  driven  by  2  external  large-­‐scale  variables  

Ø C  as  “convec+on”  (e.g.  CAPE  or  ω500)  

Ø D  as  “dryness”  (e.g.  RH500)  

Page 5: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

The Stochastic Multi-Cloud Model (SMCM) (Khouider  et  al  2010)

SMCM  produces  sensible  tropical  convec+ve  features  in  idealized  theore(cal  studies  and  compared  to  observa(ons  

Khouider  et  al  (2010),  Frenkel  et  al  (2012,2013),  Peters  et  al  (2013),Khouider  (2014),  Deng  et  al  (2015)  

-  divides  a  large-­‐scale  domain  into  independent  sub-­‐domains  

-  predicts  an  ensemble  of  congestus,  deep  and  stra+form  cloud  

Ø transi+ons  determined  by  stochas(c  Markov-­‐Chain  process  

Ø area  frac(ons  per  cloud  type  -  driven  by  2  external  large-­‐scale  variables  

Ø C  as  “convec+on”  (e.g.  CAPE  or  ω500)  

Ø D  as  “dryness”  (e.g.  RH500)  

Page 6: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

C:  “convec+ve    propensity”  

D:  500  hPa    “dryness”  

   Sta+s+cs  similar  between  model  vs.  observa+ons    

C  determined  by  ω  at  500  hPa    D  determined  by  2x(1  -­‐  RH)  at  500  hPa  

SMCM vs observations

cf.  Peters  et  al  (2013)  

 mean            variability  

             SMCM

   

                   observa+o

ns  

Page 7: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

C:  “convec+ve    propensity”  

D:  500  hPa    “dryness”  

   Sta+s+cs  similar  between  model  vs.  observa+ons    

✔  C  determined  by  ω  at  500  hPa    D  determined  by  2x(1  -­‐  RH)  at  500  hPa   cf.  Peters  et  al  (2013)  

 mean            variability  

             SMCM

   

                   observa+o

ns  

SMCM vs observations

Page 8: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

The SMCM as part of an existing scheme (ECHAM6.1)

Host  model  Dynamics  

Physics  

convec(on  scheme    

diagnosis    

parcel  ascent    

closure  

+mestep  +  1  

ECHAM  uses  a  mass-­‐flux  convec+on  parameterisa+on  based  on  Tiedtke  (1989)  and  Nordeng  (1994),  all  simula+ons  performed  at  T63L47  resolu+on  

Page 9: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

The SMCM as part of an existing scheme (ECHAM6.1)

ECHAM  uses  a  mass-­‐flux  convec+on  parameterisa+on  based  on  Tiedtke  (1989)  and  Nordeng  (1994),  all  simula+ons  performed  at  T63L47  resolu+on  

Host  model  Dynamics  

Physics  

+mestep  +  1  

SMCM  RH500    w500  

fd,  used  for  deep  convec+ve  cloud  base  mass  flux  Ø  fd  *  1  m/s  

The  currently  opera+onal  CAPE  closure  is  not  used!    Through  the  SMCM,  we  know  how  much  deep  convec+on  there  is  given  the  large  scale  environment    

convec(on  scheme    

diagnosis    

parcel  ascent,  closure  

Page 10: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

“How does the SMCM couple to the convection scheme?” Experiment  setup:  AMIP,  one  year  spinup  (2002),  then  one  week  output  @(mesteps.    Here:  4  January  2004,  loca+on:  148E,1S  (random  pick)    

SMCM    

REF  

Page 11: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

How  does  the  SMCM  version  of  ECHAM6.1  behave  in  a  climate  simula+on?  

Page 12: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

The SMCM in 30yr AMIP simulations Total  precipita+on  

Convec+ve  precipita+on  

Large-­‐scale  precipita+on  

SMCM

 –  REF  

ECHAM(REF)  -­‐  GPCP  

ECHAM(SMCM)  -­‐  GPCP  

Page 13: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

ECHAM(SMCM)  –  ECHAM(REF)   ECHAM(SMCM)  -­‐  ERAI   ECHAM(REF)  -­‐  ERAI  

The SMCM in 30yr AMIP simulations

Weaker  convec+on  in  SMCM  -­‐>  moister  BL  and  at  the  same  +me  drier  lower  and  middle  tropical  troposphere  Slightly  reduces  bias  compared  to  ERA-­‐Interim  

Specific  humidity  

RMSE:  2.5E-­‐4   RMSE:  2.8E-­‐4  

Page 14: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

Hovmoeller  diagrams  of  total  precipita+on,  May  –  August  2003,  10S-­‐10N    

The SMCM in 30yr AMIP simulations

Page 15: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

Summary ECHAM  SMCM  setup  results  in  

Ø more  temporally  coherent  deep  convec+ve  rainfall  (tropics)  Ø more  consistent  triggering  of  shallow  vs  deep  convec+on  

Ø  an  overall  reduc(on  of  global  mean  precipita(on    Ø  reduced  bias  w.r.t.  GPCP  

Ø  reduced  moist  bias  in  the  lower  troposphere  (w.r.t.  ERA-­‐I)  

Ø  increased  convec(ve  organiza(on  in  the  tropics    

Outlook Ø  implementa+on  into  most  recent  ECHAM  (vn6.3)  Ø  include  further  SMCM  simulated  cloud  area  frac+ons    Ø  include  convec+ve  interac+ons  (Khouider  (2014))  

 

Page 16: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

The  SMCM  has  the  poten+al  to  provide  the  backbone  of  next  genera+on  convec+on  

parameterisa+ons  

Page 17: !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an ...€¦ · !The!Stochas+c!Mul+cloud!Model!(SMCM)! as!partof!an!operaonal!convec+on! parameterisaon !in!acomprehensive!GCM ! Karsten!Peters1,2,!Chris+an!Jakob1,!Benjamin!Möbis2

References

-  Deng,  Q.,  B.  Khouider,  and  A.  Majda,  2015:  The  MJO  in  a  Coarse-­‐Resolu+on  GCM  with  a  Stochas+c  Mul+cloud  Parameteriza+on,  J.  Atmos.  Sci.,  DOI:  10.1175/JAS-­‐D-­‐14-­‐0120.1  

-  Frenkel,  Y.,  A.  Majda,  and  B.  Khouider,  2012:  Using  the  stochas+c  mul+cloud  model  to  improve  tropical  convec+ve  parameteriza+on:  A  paradigm  example.  J.  Atmos.  Sci.,  69,  1080–1105.  

-  Frenkel,  Y.,  A.  Majda,  and  B.  Khouider,  2013:  Stochas+c  and  determinis+c  mul+cloud  parameteriza+ons  for  tropical  convec+on.  Clim.  Dyn.,  41:1527–1551  DOI  10.1007/s00382-­‐013-­‐1678-­‐z  

-  Khouider  B,  Biello  J,  Majda  A.  2010.  A  stochas+c  mul+cloud  model  for  tropical  convec+on.  Commun.  Math.  Sci.  8(1):  187–216.  

-  Khouider,  B.,  2014:  A  coarse  grained  stochas+c  mul+-­‐type  par+cle  interac+ng  model  for  tropical  convec+on:  Nearest  neighbour  interac+ons.  Commun.  Math.  Sci.,  12,  1379-­‐1407  

-  Nordeng  T.  1994.  Extended  versions  of  the  convec+ve  parametriza+on  scheme  at  ecmwf  and  their  impact  on  the  mean  and  transient  ac+vity  of  the  model  in  the  tropics.  In:  Tech.  Memo.,  206,  European  Centre  for  Medium-­‐Range  Weather  Forecasts,  Reading,  U.K.  

-  Peters  K,  Jakob  C,  Davies  L,  Khouider  B,  Majda  AJ.  2013.  Stochas+c  Behavior  of  Tropical  Convec+on  in  Observa+ons  and  a  Mul+cloud  Model.  J.  Atmos.  Sci.  70(11):  3556–3575,  doi:10.1175/JAS-­‐D-­‐13-­‐031.1  

-  Tiedtke  M.  1989.  A  comprehensive  mass  flux  scheme  for  cumulus  parameteriza+on  in  large-­‐  scale  models.  Mon.  Wea.  Rev.  117(8):  1779–1800.