Transcript Slide 1

Is there one Indian Monsoon in IPCC
AR4 Coupled Models?
Massimo A. Bollasina – AOSC658N, 3 Dec 2007
Framework
The ASM is a challenging testbed for models, both coupled or run with observed
SST (e.g., CMIP, AMIP, CLIVAR/IMP, etc.), given the complex feedbacks among landocean-atmosphere
Several intercomparison studies focused around the AM region and the tropical
Pacific have been carried out recently with IPCC AR4 models (as part of the WCRPCMIP3), including:
 South Asian monsoon and ENSO in simulations (Annamalai et al., 2007)
 South Asian monsoon precipitation variability, simulated and projected (Kripalani
et al., 2007)
 Double ITCZ and ocean-atmosphere feedback analysis (Lin, 2007)
 Global precipitation characteristics (Dai, 2006)
 Agreements in the simulated global water cycle (Waliser et al., 2007)
 Atmospheric hydrological cycle in the Tropics (Wang and Lau, 2007)
 Enso evolution and teleconnections (Joseph and Nigam, 2006)
Motivation
Precipitation is one of the most important climate variables
It is also a key variable for the assessment of model skill, being the result of a
variety of physical processes. Errors in the simulated precipitation often reflect
deficiencies in the representation of these processes in the model
Precipitation, in particular summer rainfall (~75% of the total annual amount),
is also of major importance for the Indian monsoon region
This work aims at addressing these questions:
 How well do IPCC AR4 CGCMs simulate the spatial and temporal
patterns of the mean annual cycle of precipitation in the Indian
monsoon region?
 Are there sistematic biases and/or common deficiencies?
 Are there skills at regional scale?
Which is the link between precipitation and SSTs?
Climate of the Twentieth Century
 25 state-of-the-art global coupled models participated in the AR4, 7 considered
 Focus is on the retrospective integrations used to simulate the climate of the
twentieth century in coupled models (20c3m)
 The 20c3m simulations attempt to reproduce the overall climate variations during
1850-present with best estimates of natural (e.g., solar radiations, volcanic
aerosols) and anthropogenic (e.g., GHGs, sulfate aerosols, ozone) forcings
 Multiple realizations are available for each participating model from the LLNL
Program for Climate Model Diagnostics and Intercomparison (PCMDI) archives.
 Only the first run (run 1) is analyzed
 Period: 1979-1999 (21 years) at monthly scale
Data Used
Models:
4 from the US:
 NCAR Community Climate System Model version 3 (NCAR CCSM3),
 GFDL Coupled Model version 2.1 (GFDL-CM2.1)
 NCAR Parallel Coupled Model (PCM),
 GISS model (GISS-EH)
2 from EU:
 UKMO Hadley Centre Coupled Atmosphere–Ocean General Circulation
Model version 3 (HadCM3)
 MPI for Meteorology ECHAM5/MPI - Ocean Model (OM)
1 from JAPAN:
 CCSR(UT)/NIES/JAMSTEC Model for Interdisciplinary Research on Climate
version 3.2 (MIROC3.2)
To facilitate intercomparison, datasets regridded to R30 (~2.25°x3.75°)
Validating datasets: GPCP (2.5°x2.5°), ERA40 (2.5°x2.5°), HadSST (1°x1°), AIR
Climate Models Analyzed
AGCM res.
OGCM res.
Flux correction
T85L26(1.4°x1.4°)
384x320L40
N
GFDL
144x90L24
(2.0°x2.5°)
1°x0.33°L50
N
PCM
T42L26 (2.8°x2.8°)
384x288L32
N
GISS-EH
72x46L17
(4°x5°)
2°x2°L16
N
HadCM3
T63L18
(2.75°x3.75°)
1.25°x 1.25°L20
N
ECHAM5
T63L31
(2.0°x2.5°)
1.5° x1.5°L40
N
MIROC-H
T106L56
(1.125°x1.125°)
0.1875°x 0.28125°
L47
N
CCSM3
Seasonal mean (Jun-Sep)
precipitation (mm day-1)
and 850 hPa winds (m s-1)
Seasonal mean (Jun-Sep)
standard deviation of
precipitation (mm day-1)
Zonally averaged
precipitation (mm day-1)
between 70°-100°E as a
function of latitude
Annual cycle of monthly precipitation (mm day-1) over India (land points) and the
Indian Ocean (60°-100°E; 5°-30°N)
Root Mean Square Difference (RMSD; mm day-1) and Spatial Correlation of monthly
precipitation over (60°-100°E; 0°-30°N) with respect to GPCP
Annual cycle of Standard Deviation of monthly precipitation (mm day-1) over India
(land points) and the Indian Ocean (60°-100°E; 5°-30°N)
Seasonal mean (Jun-Sep)
1000-300 hPa vertically
integrated stationary moisture
flux (kg m-1 s-1) and its
convergence (mm day-1;
shaded)
Latitude-height crosssection between 75°-85°E
of winds (streamlines) and
vertical velocity (Pa s-1 ;
multiplied by 100)
Trend of Jun-Sep precipitation
(mm day-1year-1)
Seasonal lead-lag correlations between JJA precipitation over India and SST over the
Indian Ocean based on monthly data (Observations: All-India Rainfall and HadSST)
Lag -6 = DJF SST leading JJA PCP; Lag -3 = MAM SST leading JJA PCP
Lag 0 = JJA SST and JJA PCP; Lag +3 = SON SST lagging JJA PCP
Amplitude and phase (arrows) of the climatological mean annual cycle of Precipitation
Amplitude and phase (arrow) of the climatological mean annual cycle of SST
Lead-lag correlations between JJAS precipitation and local SST based on monthly
data (Observations: GPCP and HadSST) - Lag is referred to SST (± 1 month)
R = 0.22 is 95%
conf. level
Conclusions
 A realistic simulation of precipitation is a very challenging task, considering it
is the result of many dynamical and physical processes and feedbacks
 Current coupled models, although improved considerably with respect to the
versions used only few years ago, still show large deficiencies and biases in
the simulation of regional patterns and variability of precipitation
 A wide spread of responses among the models exists and many features of
the mean monsoon are not correctly simulated
 Regional-scale hydroclimate simulation is still a very challenging task
 Caution has to be used in interpreting modeling results
 A significant improvement is desirable (e.g., orographical precipitation, airsea interaction), both for temporal and spatial characteristics of precipitation