3. Monsoon Modeling

1
Celebrating the Monsoon Bangalore, India 7-25
2007 Climate Modeling of Monsoon Bin Wang
 Department of Meteorology and IPRC, SOEST
University of Hawaii, Honolulu, HI 96822, USA
2

• Dynamic model is ultimate tool for climate
  prediction (Probabilistic forecast)
• MME depends on good models. Improvement of models
  is of central importance for improvement of
  climate forecast

3
Coordinated evaluation of climate models

• MIPs AMIP, CMIP, S-MIP, PMIP,..
• IPCC models 20C,
• Climate prediction models DEMETER, APCC/CliPAS
• Use of forecast type experiments to evaluate
  models and study climate sensitivities should be
  encouraged.
• Identify common errors
• Set the priority for future field studies

4
Gaps First Pan-WCRP workshop (2005)

• Global Phenomena
• diurnal cycle
• annual cycle,
• intraseasonal oscillation,
• atmospheric moisture distribution and transport
• aerosol-monsoon-cloud interaction
• Model processes surface fluxes, planetary
  boundary layer and cloud.
• Land surface need better observations of land
  surface conditions roles of atmosphere-land
  coupling in developing monsoon precipitation
• Ocean improve (and sustain) observations
  importance of air-sea interaction and ocean
  processes in modelling of ISO and ENSO-monsoon
  relationship
• Regional foci processes over the Maritime
  Continent, Pacific cold tongue and western
  boundary currents, and Indonesian through flow.

5
What is best strategy for improving climate
modeling?

• Seamless evaluation with a set of good metrics
• Diurnal cycle,
• Annual cycle,
• Intraseasonal variation,
• Interannual variability,.
• Global constraints Energy and water balance in
  coupled climate models

6
DC and AC Issues

• What determines the structure and dynamics of the
  annual cycle (AC) and diurnal cycle (DC) of the
  coupled atmosphere-ocean-land system?
• What are the major weaknesses of the climate
  models in simulation of the AC and DC?
• Do models getting DC and AC right will improve
  the modeling of low-frequency variability
  (intraseasonal to interannual)?

7
Why Care about the diurnal cycle

• Initial model errors are often indicative of the
  systematic climate biases
• Diurnal cycle provides a effective varification
  of the model physical parameterization surface
  fluxes, planetary boundary layer and cloud.
• Diurnal cycle is most relevant to monsoon
  modeling

8
? Diurnal Range (Pmax-Pmin) map ?3B42 1998-2006
Large diurnal cycle is associated with monsoon
JJA
DJF
9
? Annual mean Diurnal Precipitation 3B42 ?
Most complex diurnal cycle is located in monsoon
regions
PCs
EOF1 (62 )
EOF2 (27 )

• Oceanic
• Continental
• Coastal

Kikuchi and Wang (2007)
10
? Phase propagation in the coastal regime (1)
?EEOF analysis (1)
(a) S. Asia (46, 36)
(c) W. Africa (50, 36)
(b) America (44, 37)
09 (12) LST
12 (15) LST
15 (18) LST
18 (21) LST
21 (00) LST
11
? Phase propagation in the coastal regime (2)
?EEOF analysis
(d) MC (49, 39)
(f) MDGSCR (51, 36)
(e) S. America (45, 37)
09 (12) LST
12 (15) LST
15 (18) LST
18 (21) LST
21 (00) LST
12
Global distribution of diurnal rainfall peak
Model produces diurnal rainfall peak 2-3 hour too
early
Lau and Kim (2007 JMSJ Special Issue)
13
Standard Deviation of Climatological pentad mean
precipitation (May-Sept.)
Longitudinal distribution of SD averaged for
10-20N
All models simulate larger-than-observed
amplitudes of climatological seasonal variations
of the Indian summer monsoon but underestimate
the amplitudes in the western Pacific.
Kang et al. 2002
14
AGCMs simulate climatology poorly over the WNP
heat source region
Kang et al. 2004, Cli Dyn
Wang et al. 2004, Cli Dyn
15
Global precipitation climatology Metrics
16
JJAS minus DJFM
MVEOF1
AM minus ON
MVEOF2
17
Upper panel Shading Annual range (AR) in unit
mm/day ARlocal summer minus local winter
Local summer MJJAS in NH and NDJFM in SH Black
contour global monsoon domain two criteria AR
gt 250 mm Local summer/annual total rainfall gt
50 green line ITCZ (JJA)/blue line ITCZ
(DJF) the maximum rainfall at each longitude
between 30S-30N Lower panel annual mean
rainfall in unit mm/day
18
Modeling/prediction of Global Monsoon Domain
Number of Model
The monsoon precipitation index (shaded) and
monsoon domain (contoured) captured by (a) CMAP
and (b) the one-month lead MME prediction. (c)
The number of model which simulates MPI over than
0.5 at each grid point.
19
Performance on Annual Cycle and its Linkage with
Seasonal Prediction skill
Annual Mean Precipitation
The models performance in simulating and
forecasting seasonal mean states is closely
related to the models capability in predicting
seasonal anomalies.
20
MJO Simulation AMIP intercomparison Slingo et
al. (1996) Lau et al. (1996). Individual model
studies Lau and Lau, 1986 Park et al., 1990
Slingo and Madden, 1991 Wang and Schlesinger,
1999 Hendon, 2000 Inness et al., 2001 Maloney
and Hartmann, 2001 Maloney, 2002 Wu et al.,
2002 Inness and Slingo, 2003 Inness et al.,
2003 Lee et al., 2003 Liess and Bengtsson,
2003 Liess et al., 2003 Waliser et al., 2003a
ECMWF, 2004. Prominent shortcomings A tendency
for weak variability propagate too fast
sensitivity to mean state conditions
particularly in the Indian and western Pacific
Ocean a less than ideal representation of the
modulation by the annual cycle improper phase
relationships between convection and surface heat
flux components.
21

• ISV Variance is too small
• MJO variance does not come from pronounced
  spectral peak but from over reddened spectrum
  too strong persistence of equatorial
  precipitation (13/14)

22
Satellite Observed Boreal Summer ISO (1998-2005)
Numbers four phases, phase interval 8 days

• Northward propagation in Bay of Bengal (Yasunari
  1979, 1980, Sikka and Gadgel 1980) and
  northwestward propagation in WNP (Nitta 1987)
• Formation of NW-SE tilted anomaly rain band
  (Maloney and Hartmann 1998, Annamalai and Slingo
  2001, Kemball-Cook and Wang 2001, Lawrence and
  Webster 2002, Waliser et al. 2003)
• Initiation in the western EIO (60-70E) (Wang,
  Webster and Teng 05)
• Seesaw between BOB and ENP and between EEIO and
  WNP.

23

• Problems that tend to be unique to the boreal
  summer ISO simulations
• Sperber et al. (2001) 7 models.
• The models usually fail to project the
  subseasonal modes onto the seasonal mean
  anomalies
• Waliser et al. (2003) 10 AGCMs
• Most problematic feature is the overall lack of
  variability in the equatorial Indian Ocean.
• Most of the model ISO patterns did exhibit some
  form of northward propagation. But they often
  show a southwest-northeast tilt rather than the
  observed northwest-southwest tilt.
• The fidelity of a model to represent N.H. summer
  versus winter ISV appears to be strongly linked.
• Double ITCZ inadequate global teleconnectionLack
  of ocean coupling (Fu et al. 2003, Zheng et al.
  2004)

24
ISV prediction skill in DEMETER and CliPAS models

• Predictability of unfiltered daily precipitation
  (signal/noise)
• Indian Ocean (60E100E, 10S20N) and
• (b) WNP (120E-140E, EQ-20N).

Nearly all models show drop of forecast skill
after about a week in the summer monsoon regions.
Confirmed by anomaly correlation.
25
ISO Potential Predictability
Air-Sea Coupling Extends the Predictability
of Monsoon Intraseasonal Oscillation
ATM Forecast Error
CPL Forecast Error
Signal
ATM 17 days, CPL 24 days
Fu et al. 2006
26
Interannual variations of the boreal summer
intraseasonal variability predicted by ten
atmosphere-ocean coupled models Hye-Mi Kim and
In-Sik Kang School of Earth and Environmental
Science, Seoul National University, Seoul,
Korea Bin Wang and June-Yi Lee Department of
Meteorology and International Pacific Research
Center, University of Hawaii at Manoa, Honolulu,
Hawaii, USA
27
Climatological ISV activity
The observed ISV activity exhibits its largest
values over the WNP and Indian monsoon regions.
Model tend to underestimate the variability in
these regions and over estimate the variability
in equatorial WP.
28
Interannual standard deviations of the boreal
summer ISV activity
Notable interannual variation of ISV activity is
over the WNP in observation. Most models have the
largest variability over the central tropical
Pacific and exhibit a wide range of variability
in spatial patterns that differ from observation.
29
EOF modes of the observed and model composite IAV
of ISV activity
Although models have large systematic biases in
spatial pattern of dominant variability, the
leading EOF modes of the ISV activity in the
models are closely linked to the models ENSO,
which is a feature that resembles the observed
ISV and ENSO relationship.
30
Scatter plot of predictability on summer mean
precipitation (horizontal axis) and ISV activity
(vertical axis) over the Asian monsoon region
(40-180E and 20S-30N).
31
Findings

• Predictability of the IAV of ISO activity is
  positively correlated with models predictability
  of IAV of the seasonal mean rainfall.
• The ENSO-induced easterly vertical shear
  anomalies in the western and central tropical
  Pacific, where the summer mean vertical wind
  shear is weak, result in ENSO-related changes of
  ISV activity in both observation and models.
• The predictability of ISV activity can be
  improved since the model errors are systematic
  and related to the external SST condition.

32
Need to understand Monsoon ISO Multi-Scale
Interrelation
Slingo 2006 THORPEX/WCRP Workshop report
33
ISV Dynamics

• What are major modes of the Monsoon ISV (Boreal
  summer ISO)?
• What is the typical multi-scale structure of ISO?
  What is the 3-D structure of thermodynamic fields
  of ISO?
• How are organized convections linked to large
  scale forcing? What are triggers for organized
  convective events in general?
• How do we get a complete theoretical framework
  for describing characteristics of MJO? Are the
  multi-scale structure/interaction important for
  ISO and how?
• What is the role of mesoscale systems in
  determining the heating profile
  (convective/stratiform) and how does this impact
  the evolution of ISO? How to get them right?

34
Cloud and rainfall profile during genesis process
Height (km)
(a) Phase 1
(b) Phase 2
(c) Phase 3
mm/hour
mm/hour
mm/hour
Composite vertical profiles of precipitation rate
anomalies measured by the TRMM precipitation
radar (TRMM/2A25) over the equatorial Indian
ocean (5ºS-5ºN, 60º-70ºE) where ISO initiates.
The composite was made for 28 ISO events. Each
phase spends about four days. Orange and green
color represent convective and stratiform rain
rate, respectively. During the initiation of the
convective anomalies (phase 2 and 3), the
stratiform and convective rains have comparable
rates the prevailing cloud type experiences a
trimodal evolution from shallow (phase 1) to deep
convection (phase 2), and finally to anvil and
extended stratiform clouds (phase 3). These
information is critical for validation of
numerical models. From Wang et al. (2006).
35
MJO (MISO)

• How should we evaluate our models and measure ISO
  predictability and prediction skill? US-CLIVAR
  MJO working group Metrics
• What is the current level of performances and
  common problems in the models? How to correct
  these systematic errors?
• How does the errors in simulating ISO impacts
  simulations of the interannual variability?
• To what extent the MISV is predictable?
• What roles does atmosphere-ocean-land interaction
  play in MISV?

36
Interannual Variability Sperber and Palmer
(1996) 32 models from AMIP results. Asian
Monsoon variability is not well simulated. The
Webster and Yang index (wind shear) is better
simulated than the all-India rainfall. IAV is
better simulated in models which are able to
generate a better climatology. After model
revision, simulation of the interannual
variability was significantly improved (Sperber
et al., 1999). Wang et al. (2004) 10 AGCMs
Wang et al. (2005) 5 two-tier model 21-year
hindcast Wang et al. (2007) 10 CGCM (one-tier
system 21-year hindcast DEMETER and APCC/CliPAS
models)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
A fundamental Challenge in Climate Prediction
Wang et al. 2005
State-of-the-art AGCMs, when forced by
observed SST, are unable to simulate
Asian-Pacific summer monsoon rainfall (Fig. a).
The models tend to yield positive SST-rainfall
correlations in the summer monsoon region (Fig.
c) that are at odds with observation (Fig.b).
Treating monsoon as a slave to prescribed SST
results in the models failure, which suggests
inadequacy of the tier-2 climate prediction
system.
a. 5-AGCM ensemble hindcast skill
b. OBS SST-rainfall correlation
c. Model SST-rainfall correlation
41
Area averaged correlation coefficients (skills)
42
Simultatious
Rainfall Leads SST by 1-month
SST leads Rainfall by 1-month
Observed Rainfall-SST correlation (1979-2002)
Wang et al. 2005
43
Interannual variation

• How accurate do coupled climate models predict
  major modes of interannual variability of A-AM?
• What roles does atmosphere-land interaction play?
  
• How predictable is the continental monsoon
  interannual variability (IAV)?
• How to improve seasonal prediction in continental
  monsoon regions?

44
Comparison of Spatial Patterns
45
MME capture ENSO-MNS relation
PC Spectra MMEs underestimate QB Peak and total
variances
PC time series MMEs are highly correlated with
CMAP
S-EOF1 concurs with ENSO
SEOF2 leads ENSO by 1 year
46
Comparison of MME forecast with Reanalyses
The MME beats two re-analyses in capturing both
the spatial patterns and temporal evolutions of
the two leading modes
47
Discussion

• The difficulty in current numerical simulation of
  the annual cycle in the East Asian summer monsoon
  is rooted in the relatively weak external
  forcing.
• The difficulty in current numerical simulation of
  the interannual variability in the Indian summer
  monsoon is rooted in the relatively weak internal
  forcing within the coupled climate system and
  monsoon chaos.
• The difficulty in modeling of the MJO and ISV is
  primarily due to deficiencies in atmospheric
  internal dynamics handing convective interaction
  with large scale dynamics (parameterization
  problem) and in handling multi-scale interaction.

48

• Improving model physical parameterization
• Monsoon processes are sensitive to
• Cumulous parameterization
• Cloud-radiation interaction
• Aerosol impacts
• Coupled model bias
• Waliser (2005) When a model does exhibit a
  relatively good MJO/ISO, we can at best only give
  vague or plausible explanations for its relative
  success. This inhibits the extension of
  individual model successes to other more
  ISV-challenged models.
• Ccollaboration among various modeling groups
  (large-scale modelers, meso-scale modelers, and
  cloud modelers
• Strategy for validation of the models
  representation of the physical processes.

49
Atmosphere-land interaction Effects of the
soil moisture and snow cover over the Eurasian
continent on Asian monsoon variability. Yasunari
(1991), Dirmeyer(1999) the land surface
condition in Spring has an impact on the
following summer monsoon. Shen et al. (1998)
have investigated the impact of the Eurasian
snowfall and concluded that it plays a part but
does not overwhelm the SST-impact.
50
Effects of model resolution Orography is better
represented as the resolution is increased and
the rainfall associated with orography is better
represented in higher resolution models. High
resolution models may be able resolve better
Mei-Yu front. Kawatani and Takahashi (2003), Sumi
et al. (2004) demonstrated that the Baiu front
can be well simulated by increasing the
horizontal and vertical resolutions.
51
Modeling Need

• Design monsoon metrics for assessing model
  performance and identify key modeling issues.
  Provide one-stop data source for cross-panel use?
  
• Develop effective strategy for improving models.
• Use large-domain CR or CSR simulation to provide
  surrogate data for studying convective
  organization, and multi-scale interaction in MJO.
  
• Determine directions for developing next
  generation climate models.