Dynamical Prediction of Indian Monsoon Rainfall and the Role of Indian Ocean K. Krishna Kumar CIRES Visiting Fellow, University of Colorado, Boulder [e-mail] Martin P. Hoerling Climate Diagnostics Center, Boulder and Balaji

1
Dynamical Prediction of Indian Monsoon Rainfall
and the Role ofIndian OceanK. Krishna
KumarCIRES Visiting Fellow, University of
Colorado, Boulderkkrishna_at_colorado.edu Martin
P. HoerlingClimate Diagnostics Center,
BoulderandBalaji RajagopalanUniversity of
Colorado, Boulder
2
Current Practices of Dynamical Monsoon Rainfall
Prediction

• 2-tiered approach wherein SSTs are predicted
  first using a coupled model and then the AGCMs
  are forced using these SST fields
• Use persistent SSTs to run AGCMs
• Dynamical Downscaling using Regional Climate
  Models taking lateral boundary values from AGCM
  Simulations

3
Skills of the Present Generation of
AGCMs(Reproduced from the IRI Website)
4

• We set out to examine the skills of monsoon
  rainfall in detail by involving long simulations
  made using observed SSTs with a suite of
  multi-model, multi-member ensemble runs.

5
Research Questions..?

• How skillful are the AGCMs in simulating Monsoon
  Rainfall over the Indian region?
• Is specifying SSTs a constraint on realistic
  monsoon simulations?
• How sensitive are monsoon simulations to initial
  conditions?
• What is the impact of coupling on Monsoon-ENSO
  relationships?
• Are the ENSO related western Indian Ocean SSTs
  acting as negative feed-back on Monsoon-ENSO
  relations?

6
Details of AGCMs Used
S.No. Model Resolution Ens. Size Run Length
1 ECHAM4 2.8x2.8 24 1950-2002
2 ECHAM3 2.8x2.8 10 1950-1999
3 GFDL 2.5x2.0 10 1951-2002
4 NASA 2.8x2.8 9 1950-2002
5 ECPC 1.8x1.8 7 1950-2001
6 MRF (NCEP) 2.8x2.8 13 1951-1994
7 ARPEGE 2.8x2.8 8 1948-1997
8 CCM3 2.8x2.8 12 1950-1999
9 CAM2 2.8x2.8 15 1950-2001
7
Climatology of Monsoon Rainfall
8
(No Transcript)
9
(No Transcript)
10
Monsoon-ENSO Relation in AGCM Simulations
11
PDFs of Correlations(1) Obs. Vs. Model ENS (2)
PERPROG
12
Impact of Initial Conditions on Monsoon
Simulations
13
Monsoon-ENSO Teleconnections Coupled vs.
Uncoupled Models
14
GOGA Obs SSTs globallyDTEPOGA Obs SSTs
in Deep Tropical East Pacific and Climatological
SSTs elsewhereDTEPOGA_MLM Same as DTEPOGA
but a Mixed Layer Model used in the Indian Ocean
15
Progressive Improvement in Monsoon Rainfall
Simulation Skills 1. Un-coupled AMIP
2. Un-coupled AMIP only in eastern tropical
Pacific and Climatological SSTs elsewhere
3. AMIP in the Pacific and Mixed Layer Model
in the Indian Ocean
16
(No Transcript)
17
Summary

• The skills of current generation AGCMs in
  simulating monsoon rainfall in India even when
  forced with observed SSTs are very low.
• However, there appears to be much higher
  predictive potential as evidenced by the large
  PERPROG skills.
• No clear hint of higher skills either for models
  with better monsoon climatology or when
  multi-model-super ensembles are involved.
• Specification of SSTs in the Indian Ocean appears
  to be the main reason for the low-skills.
• An interactive ocean-atmosphere in the Indian
  Ocean (using even a simple mixed layer ocean
  model) produces more realistic monsoon
  simulations compared to specifying actual or
  climatological SSTs.
• General belief that the ENSO related SSTs in the
  Indian Ocean (particularly the western Indian
  Ocean and the Arabian Sea) might act as a
  negative feedback on Monsoon-ENSO teleconnections
  appears to be wrong based on the above
  observations.
• In general the monsoon-ENSO links are much
  stronger in fully coupled models compared to the
  AGCMs forced with observed/predicted SSTs.
• The 2-tiered approach currently being pursued in
  seasonal forecasting needs immediate revision to
  achieve higher forecast skills for the Indian
  region. We also believe, this might be true for
  some other countries located in the warm pool
  region in the west Pacific and the Indian Ocean.

18
(No Transcript)
19
Thank You!!!