Impact of the MJO on the Gulf of Carpentaria during the monsoon

<strong>Impact</strong> <strong>of</strong> <strong>the</strong> <strong>MJO</strong> on
<strong>the</strong> Gulf <strong>of</strong> Carpentaria
during <strong>the</strong> monsoon
CAWCR Annual Workshop
13 November 2012
Melbourne, VIC, Australia
Eric C. J. Oliver 1,2 , Keith R. Thompson 2
1 Institute for Marine and Antarctic Studies
University <strong>of</strong> Tasmania, Hobart, TAS, Australia
2 Physical Oceanography, Department <strong>of</strong> Oceanography
Dalhousie University, Halifax, NS, Canada

The Madden-Julian Oscillation
The Madden-Julian Oscillation (<strong>MJO</strong>) is an intraseasonal (30-90
day period) phenomenon that originates over <strong>the</strong> equatorial
Indian Ocean and propagates eastward
o S
Associated with deep convection and zonal wind anomalies at
both low and high levels in <strong>the</strong> tropics
Characterized by bivariate index [Wheeler and Hendon, 2004]

The <strong>MJO</strong> and Global Sea Level
Oliver and Thompson [JGR, 2010] calculated statistical connections
between <strong>the</strong> <strong>MJO</strong> and global sea level using a coherence-based metric
Nor<strong>the</strong>astern
Indian Ocean
Gulf <strong>of</strong>
Carpentaria
Equatorial
Pacific
Coastal
Americas
0.7
50 o N
0.6
25 o N
0.5
0 o
0.4
25 o S
0.3
50 o S
0.2
60 o E 120 o E 180 o W 120 o W 60 o W 0 o
0.1

The Gulf <strong>of</strong> Carpentaria
The Gulf <strong>of</strong> Carpentaria (GOC) is a shallow (~50 m avg.) sea north
<strong>of</strong> Australia. It neighbours shallow Arafura and Timor Seas and
<strong>the</strong> deep waters <strong>of</strong> <strong>the</strong> Indonesian Archipelago to <strong>the</strong> west as well
as <strong>the</strong> deep waters <strong>of</strong> <strong>the</strong> Western Pacific to <strong>the</strong> East
"GOC" region
for wind
model
domain
tide gauge
locations

The <strong>MJO</strong> and Coastal Sea Level
Sea level from tide gauges also coherent with <strong>MJO</strong> over
intraseasonal time scales:
0.9
0.8
0.7
Groote Ey.
Karumba
Weipa
<strong>MJO</strong> energy
0.6
coherence
0.5
0.4
0.3
0.2
0.1
0
0 0.05 0.1 0.15
frequency [cpd]

Observed Wind
Surface wind (10m) over <strong>the</strong> GOC is predominantly northwesterly
during monsoon season (peak in January) and <strong>the</strong> predominantly
sou<strong>the</strong>asterly trade winds peak six months later, in July
8
1600
6
1400
v [m/s]
4
2
0
−2
actual occurrence
1200
1000
800
600
400
monsoon
northwesterlies
sou<strong>the</strong>asterly
trades
−4
200
−6
−8 −6 −4 −2 0 2 4 6
u [m/s]
0
E NE N NW W SW S SE E
wind direction

Observed Wind
0.8
0.7
sea level
response
<strong>MJO</strong>/wind
coherence
0.08
0.07
Wind over <strong>the</strong> Gulf is
predominantly northwesterly or
sou<strong>the</strong>asterly ( histogram bars)
coherence
0.6
0.5
0.4
0.3
0.06
0.05
0.04
0.03
maximum sea level [m]
The coherence between <strong>the</strong> <strong>MJO</strong>
and surface wind is highest for
ESE and WNW winds ( line)
Sea level in <strong>the</strong> Gulf responds
preferentially to SSE and NNW
winds ( line)
0.2
0.02
0.1
0.01
0
0
E NE N NW W SW S SE E
wind direction
These factors combine to make
sea level in <strong>the</strong> Gulf particularly
responsive to <strong>the</strong> <strong>MJO</strong>

Seasonality <strong>of</strong> ISV
Strong seasonality <strong>of</strong> wind, sea level, and <strong>MJO</strong> relationship over
intraseasonal time scales.
D
N
O
S
A
J
J
M
A
M
F
Power <strong>of</strong> northwesterly surface wind
D
N
O
S
A
J
J
M
A
M
F
Coherence between wind and <strong>MJO</strong>
D
N
O
S
A
J
J
M
A
M
F
Power <strong>of</strong> Karumba sea level
J
0.015 0.02 0.025 0.03 0.035 0.04 0.045
frequency [cpd]
J
J
0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.015 0.02 0.025 0.03 0.035 0.04 0.045
frequency [cpd]
frequency [cpd]
600 800 1000 1200 1400 1600 1800
power
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 6 8 10 12 14 16
coherence
power * 100
Northwesterly surface wind is strongly coherent with <strong>the</strong> <strong>MJO</strong>
during <strong>the</strong> monsoon (Austral Summer) and this corresponds to
season <strong>of</strong> maximum intraseasonal sea level variability .

Numerical Model
Princeton Ocean Model (POM) [Blumberg and Mellor, 1987]
Non-linear, two-dimensional
barotropic
3 o S
depth [m]
4000
Ten-minute spatial resolution
(157 x 103 grid points)
6 o S
9 o S
3500
3000
2500
12 s time step for CFL cond.
12 o S
Groote
Eylandt
Weipa
2000
1500
Radiation conditions at
open boundaries
15 o S
18 o S
Karumba
1000
500
Sea level and both zonal and
meridional currents output daily
125 o E 130 o E 135 o E 140 o E 145 o E
0
Bathymetry from CSIRO, using Geoscience Australia (2009) data
Forced by NCEP/NCAR Reanalysis 2 winds (6 hourly) and we are
considering <strong>the</strong> POM results to be <strong>the</strong> wind-forced, barotropic
component <strong>of</strong> sea level variability in <strong>the</strong> Gulf <strong>of</strong> Carpentaria

Predicted Sea Level
Modeled sea level matches well with tide gauge records
RMS error [cm]:
Correlation:
Gain
Groote Eylandt
6.56
0.76
0.89
Karumba
10.70
0.76
0.89
Weipa
6.91
0.84
0.82
Coherence is high (0.80-0.95) on intraseasonal frequencies (20-100
days) but drops for longer periods (>100 days) - model does not
capture low frequency variability
Karumba
sea level [m]
0.6
0.4
0.2
0
−0.2
−0.4
obs. (high passed)
model prediction
−0.6
Jan
1997
Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
1998
Feb

Predicted Seasonality <strong>of</strong> ISV
Numerical model reproduces <strong>the</strong> seasonal cycle <strong>of</strong> intraseasonal
variability, although with reduced amplitude
D
N
O
S
A
J
J
M
A
M
F
J
D
N
O
S
A
J
J
M
A
M
F
J
observations Groote Eylandt
model
D
N
O
S
A
J
J
M
A
M
F
J
0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.015 0.02 0.025 0.03 0.035 0.04 0.045
observations
Karumba
model
D
N
O
S
A
J
J
M
A
M
F
J
0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.015 0.02 0.025 0.03 0.035 0.04 0.045
0 5 10
0 10 20
600 800 1000 1200 1400 1600
power

Canonical Response to <strong>MJO</strong>
Composites <strong>of</strong> obs. wind, modeled sea level and circulation with <strong>the</strong>
<strong>MJO</strong>: canonical response <strong>of</strong> Gulf <strong>of</strong> Carpentaria to <strong>MJO</strong>
Phase 1
3 cm/s
Phase 2
Phase 3
Phase 4
10
10 o S
8
12 o S
14 o S
6
4
Setdown
during
phases 1-3
16 o S
2
1.01 m/s
Phase 5
1.06 m/s
Phase 6
0.59 m/s
Phase 7
0.18 m/s
Phase 8
0
sea level [cm]
10 o S
12 o S
14 o S
−2
−4
−6
Setup
during
phases 5-7
16 o S
0.45 m/s
1.13 m/s
1.67 m/s
0.68 m/s
−8
136 o E 138 o E 140 o E 142 o E136 o E 138 o E 140 o E 142 o E136 o E 138 o E 140 o E 142 o E136 o E 138 o E 140 o E 142 o E
−10

Predictability: Role <strong>of</strong> <strong>the</strong> <strong>MJO</strong>
The <strong>MJO</strong> index (projected on to phase 7/3) shows remarkable
correlation with intraseasonal sea level in <strong>the</strong> Gulf <strong>of</strong> Carpentaria.
sea level [m]
Karumba
0.6
obs. (high passed)
0.4
model prediction
0.2
0
−0.2
−0.4
<strong>MJO</strong> phase 7 (+) / phase 3 (-)
−0.6
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb
1997
1998
The <strong>MJO</strong> index can be used as an indicator for set-up or set-down
favourable conditions . . . <strong>the</strong> <strong>MJO</strong> can give predictability to <strong>the</strong>
system

Predictability
Potential predictability <strong>of</strong> <strong>MJO</strong>-related sea level variability
quantified using a simple statistical prediction model
Statistical model is a lagged linear regression model <strong>of</strong> sea level
onto <strong>the</strong> <strong>MJO</strong> index at lags from 0 to D days.
sea level at future time k
<strong>MJO</strong> index
regression coefficients

Predictability
Model only includes lags 0 and 8 days. Trained over March-
November for 1979-1994 and validated for March-November
1995-2009:
0.5
<strong>MJO</strong> amplitude > 1
<strong>MJO</strong> amplitude increasing
<strong>MJO</strong> amplitude decreasing
0.4
correlation
0.3
0.2
0.1
0
0 5 10 15 20 25 30
future prediction time, k [days]
Statistical model can account for at 10-25% <strong>of</strong> <strong>the</strong> variance with
leads times up to 20 days
Can be supplemented with forecasts <strong>of</strong> <strong>the</strong> <strong>MJO</strong> index...

The Gulf <strong>of</strong> Thailand
Similar study in Gulf <strong>of</strong> Thailand shows a very similar phenomenon:
<strong>MJO</strong>-related wind-driven sea level and circulation variability during <strong>the</strong>
monsoon season (July-January)
Phase 1
wind
0.80 m/s
Phase 2
wind
0.97 m/s
Phase 3
wind
0.76 m/s
Phase 4
wind
0.36 m/s
5
12 o N
9 o N
6 o N
4
3
2
Setup
during
phases 8, 1-3
3 o N
0 o
12 o N
5 cm/s
Phase 5
wind
0.99 m/s
5 cm/s
Phase 6
wind
1.00 m/s
5 cm/s
Phase 7
wind
0.50 m/s
5 cm/s
Phase 8
wind
0.48 m/s
1
0
−1
sea level [cm]
9 o N
6 o N
−2
−3
Setdown
during
phases 4-7
3 o N
−4
0 o
5 cm/s
99 o E 102 o E 105 o E 108 o E
5 cm/s
5 cm/s
5 cm/s
99 o E 102 o E 105 o E 108 o E 99 o E 102 o E 105 o E 108 o E 99 o E 102 o E 105 o E 108 o E
−5
Oliver, E.C.J., Intraseasonal variability <strong>of</strong> sea level and circulation in <strong>the</strong> Gulf <strong>of</strong> Thailand:
The role <strong>of</strong> <strong>the</strong> Madden-Julian Oscillation, accepted for publication in Climate Dynamics

Conclusions
Madden-Julian Oscillation, an intraseasonal tropical phenomenon, is
connected to global patterns <strong>of</strong> variability in sea level.
Surface wind over <strong>the</strong> Gulf <strong>of</strong> Carpentaria is highly correlated to <strong>the</strong>
<strong>MJO</strong> and is also well suited for setting up sea level.
Numerical model confirms that observations are mainly wind-driven
sea level set up during <strong>the</strong> Australian-Indonesian Monsoon
Winds that lead to sea level set-up are part <strong>of</strong> a global system related
to <strong>the</strong> <strong>MJO</strong>: <strong>the</strong>re is potential for predictability and this is demonstrated
using a simple real-time prediction model
Related publications:
i) Oliver and Thomspon, 2010, J. Geophys. Res.
ii) Oliver and Thompson, 2011, J. Geophys. Res
iii) Oliver, accepted for publication in Clim. Dyn.
Thank you!
NUNATSIAVUT PSSSP