 AIMS home
About AIMS
Research
Facilities
News
Search
Site map
Site index
Topics index
|
North
west shelf modelling workshop
PRESENTATION
SUMMARY
Large
and mesoscale oceanography: Recent or current studies and/or
future plans
Chair:
Stan Massel
13:30 Stuart Godfrey (CSIRO,Tas) – Seasonal sea level
variations and circulation
Abstract: Wind forcing, tidal
mixing and the seasonal heat budget of Australia’s Northwest
waters (J. Stuart Godfrey and James Mansbridge)
Fifteen years of expendable bathythermograph (XBT) data
from three lines that cross Australia’s North West Shelf and
the shallow part of the Timor Sea – a region of broad
continental shelf referred to below as "The Shelf"
– have been used, with climatological salinity data, to
estimate onshore geostrophic flow onto the Shelf. A wind
stress climatology is used for onshore Ekman transport.
Onshore geostrophic flow in the top 200m onto the whole Shelf
has an annual mean of 3.1 Sv; it is weakest (0.8 Sv) in
May-June, and strongest (5.1 Sv) in January. The onshore Ekman
transport has an annual mean of only –0.5 Sv, but its annual
cycle almost completely cancels that of the geostrophic flow,
so that the net onshore flow varies little through the year
from its annual mean of 2.6 Sv. Most of this probably exits
through southern end of the Shelf, as the annual mean Leeuwin
Current. The small annual cycle of the net onshore flow
suggests that longshore flow at the northern end of the
section must have the same seasonal cycle as at the southern
end.
Data limitations and inter-annual variability in longshore
current data mean that we have not been able to obtain
adequate mass budget closure. There is about 1 Sv of onshore
geostrophic flow below 200m in winter, onto the northern
Shelf. We have used climatological heat fluxes and
temperatures on the Shelf to estimate mean seasonal net heat
export from the Shelf’ export is positive throughout the
year on the northern Shelf, and is negative for only three
months of less on the southern Shelf. Despite the lack of mass
closure, we can infer some mechanisms by which this export is
carried out. The Gentilli (1972) mechanism of alternating
longshelf seasonal flow cannot explain all the observations.
Summer wind-driven upwelling can export substantial amounts of
heat from both the northern and southern Shelf. The deep
winter inflow to the northern Shelf probably upwells into the
Arafura Sea, where nutrient-rich inflow has been observed
(Rochford 1966). This flow has the right magnitude to supply
westward Ekman transports in winter, between Irian Jaya and
Australia. Indicators of tidal mixing (Field and Gordon 1996)
are widespread on the Shelf, and tidal mixing may explain why
summer SST over the Shelf shows no coastal minimum, despite
moderately large offshore Ekman transports. Winter wind-driven
upwelling in the Arafura Sea, and summer upwelling over the
Shelf – both modified by tidal mixing – may play a large
role in controlling the seasonal cycle of SST over
Australia’s northwest waters. Biases in climatological
surface heat fluxes in the summer cyclone season make it
difficult to infer the causes of the summer SST maximum.
|
Presentation and Discussion Notes :
Stuart Godfrey presented the results of his
research on the seasonal cycles of ocean temperatures,
and circulation including geostrophic and Ekman flows
into and out of the NW shelf region.
He commenced by discussing the results of a Global
ocean model with 2 deg E-W and 0.5 deg N-S grid
resolution that is coupled to a Bureau of Meteorology
model of SST variations. He pointed out that the
models show cold and warm features along the West
American Coast and asked why similar anomalies do not
appear off the West Australian coast.
The dynamics of the WA region are indeed quite
different from the rest of the world, yet SST controls
Australian climate. In summer, he estimates an
offshelf Ekman transport of 3 Sv (1 Sv = 106
m3/s), yet he observes no significant drop
in coastal temperatures. He then asked why is there no
upwelling signature in the Coastal SSTs? One reason
for this may be the effects of tidal mixing. The
temperature evolution in the Gulf of Carpentaria is
strongly influenced by tidal variations. Where
internal tidal effects are present, one would expect
to see temperature variations on the scale of the
spring/neap tidal cycle. Such effects of tidal mixing,
are evident in the Indonesian seas. By estimating
steric heights offshore from NW Australia using ship
of opportunity XBT data and comparing these with
observed sea levels it can be demonstrated that sea
level variations agree with fluctuations in the
Leeuwin current, with set up of sea levels occurring
in winter.
Sea level differences along shore have also been
used to estimate onshore geostrophic flows. There is a
steady onshore flow of 2.5 Sv onto the NW Shelf. In
summer, on the NW shelf offshore Ekman flows (upwelling)
oppose the onshelf geostrophic flows. In winter,
outflows occur in very narrow jets corresponding with
the Leeuwin current and an associated undercurrent.
How can you have upwelling and not have SST cooling
at the coast?
- One hypothesis is that tidally generated
internal waves lead to vertical mixing of water as
it upwells.
- Features are observed to be much stronger near
the coast, with internal mixing layers of order
100 m deep inshore (ie, shallower than the 200 m
isobath).
Conclusions regarding the N Australian SST:
- In Dec-Jan SST follows heat flux variations
- Cooling occurs in Jan-April, possibly related to
tidal mixing.
- Rapid cooling related to upwelling occurs as
cold dry winds are established in winter.
- Cooling occurs over the Arafura Sea in winter
due to upwelling, but the effects are disguised by
tidal mixing.
In discussion Miles Furnas pointed out that
subsurface temperatures in the vicinity of Scott Reef
peak during May-June, compared with surface
temperatures which peak during Dec-Jan.
|
13:50 Stephen Walker (CSIRO,Tas) – Shelf-scale modelling
studies
We have been developing and applying numerical
hydrodynamical models of the North West Shelf, primarily to
simulate response to tropical cyclones. The models have been
used by WNI in their determination of design criteria for
offshore structures and operations. To date, the model has
been applied over the western region of the NW shelf with 10-
and 20-km grids, and near the Gorgon and Laminaria gas fields
with 5-km grids. In anticipation of upcoming work on the NW
shelf, preliminary runs with a domain extending from Shark Bay
to Darwin have been conducted.
The model we have been using was developed at CSIRO and is
a three-dimensional, baroclinic, hydrostatic, z-coordinate
model with one-way nesting. Vertical mixing can be prescribed
or calculated using a Mellor-Yamada level 2 scheme, and bottom
drag can be prescribed or calculated using a Grant-Madsen
wave-current boundary layer formulation. The model has been
compared with oceanographic data collected by Woodside, WNI,
ADFA, and CSIRO and performs reasonably well. The greatest
errors can be attributed to errors in wind forcing and
specification of open-ocean boundary conditions. In addition,
the model generally underestimates the amplitude of internal
waves. Experiments with an essentially two-dimensional
(cross-shelf and vertical) "slice" model indicate
that very high resolution can address some, but not all, of
the internal wave problem. It is likely that accurate
reproduction of internal-wave activity will require 1) high
resolution in three dimensions and 2) detailed
boundary-condition information, and possibly 3) a
non-hydrostatic model.
|
Presentation and Discussion Notes :
Stephen focussed on work they are doing using
numerical models to predict the response of the
circulation off NW Australia to tropical cyclones.
Model 1 has a domain ranging from Coral Bay, South
of Ningaloo Reef to Ashmore Reef on the Sahul shelf
including the area near the NW shelf. Using this
model, the response to a numerical representation of
Tropical Cyclone Bobby was simulated. Winds used to
force the model were obtained from the North Rankin
oil well platform. The simulated cyclone was
parameterised to represent TC Bobby by specifying the
wind field. However the approach resulted in wind
speed being overpredicted by a factor of 2 to 3 and
this resulted in some modelling errors.
The model response was compared with thermistor
chain data obtained from moorings. The results showed
that the model reproduced the effects of mixing due to
the TC that resulted in vertical excursions of the
isotherms of up to half of the water depth. The mixed
layer recovered to a stratified state within 5-10
days.
Model 2 has a domain ranging from Darwin to
Carnarvon. Stephen is planning to force this model
with tropical cyclones Jan (1992), Bobby (1995), Frank
(1995) and Oliver (1996). The intention is to couple
the model to models of bottom sediment transport,
nutrient fluxes, and productivity.
Model 3 is a vertical slice model with 600 layers
with a vertical cell size of 0.5m in the top 300 m.
The resolution is 500 m in the horizontal. The model
showed fluctuations in the height of the internal
interfaces in response to the barotropic and
baroclinic tides.
|
14:10 Peter Holloway (ADFA) – Wind-driven circulation,
cyclones
Synopsis:
Time series of currents measured over the continental shelf
and slope regions from the North Rankin region, Exmouth
Plateau and the Timor Sea show:
- weak but persistent sub-inertial currents, typically 20
cm/s
- predominantly poleward flow, particularly from March to
July
- flows frequently into prevailing winds
- currents consistent with Leeuwin Current flow
- ADCP surveys show poleward transports of ~4 Sv
- some examples of equatorward flows responding to winds
in winter/autumn
Tropical cyclones frequently occur on the NWS and produce
strong currents over the shelf and slope regions:
- cyclone currents often observed to produce flows of ~1
m/s
- often only limited vertical mixing of the water column
- inertial ringing often follows the maximum currents
- 3D barotropic modelling gives reasonable agreement with
observed currents during TC Ian (1982)
- 2D model with bottom friction dependant on wind-wave
enhanced motion predicts coastally trapped; wave response
of TC Jane (1983)
Problems:
- accurate specification of cyclone track and wind field
- correct specification of surface drag coefficient
Future work:
- use of 3D stratified models
- correct parameterisation of turbulence and friction will
be critical
- use of satellite (scatterometer) winds to drive ocean
models
|
Presentation and Discussion Notes :
Peter Holloway discussed wind-driven circulation
and tropical currents
The typical circulation pattern of the NW shelf is
characterised by weak but persistent sub-inertial
currents, flowing predominantly poleward typically at
about 20 cm/s. These occur particularly in the period
March to July. The flow is frequently directed into
the prevailing wind, consistent with the seasonal
appearance of the Leeuwin Current offshore. Acoustic
Doppler Current Meter surveys show poleward transports
of about 4 Sv and there are some examples of
equatorward flows in response to the wind during
winter and autumn.
During Tropical Cyclone passages currents may
attain flow rates of order 1 m/s, often attended by
limited water column mixing and inertial
‘ringing’.
A 2D model with bottom friction dependent upon
wind-driven surface waves has been used to predict
coastally trapped waves.
The major problem in attaining accurate simulations
is to correctly specify the drag coefficient.
In the future use of 3d stratified models with
correct parameterisation of turbulence and friction
will be critical. Use of satellite data (eg,
scatterometry) to determine model wind fields for
driving ocean models is also projected.
Historical current meter data shows mostly poleward
flows while in August the flow tends to be
equatorward. At North Rankin winds are predominantly
poleward during March – May and variable at other
times, giving rise to the seasonal variations in the
Leeuwin Current.
Off the southern NW Shelf, data from the Exmouth
Plateau obtained by ESSO in 1978 showed persistent but
weak wind-driven flows.
Further north at Jabiru, on the Sahul Shelf flows
were predominantly poleward with some reversals
associated with winds.
ADCP section transports (after detiding) were also
shown (eg, sections off Rowley Shoals). These show a
narrowing of the flows as you move south.
Although tropical cyclone Jan passed close to the
North Rankin platform, the response was relatively
weak.
Derek Burrage asked if the ‘wave’ and
‘mixing’ response to TCs can be separated?
TC Jan, as observed at North Rankin showed quite
strong flows yet the model produced no significant
response there, which suggests the damping in the
model is excessive.
Tang, Holloway and Grimshaw used a 2D
depth-averaged model of the NW Shelf with quadratic
bottom friction and a positionally-dependent drag
coefficient (Tang and Grimshaw) JGR, 1996. This
resulted in good agreement with the observed response
for propagating coastal trapped waves. (JPO, 1997).
However, some difficulties were experienced in getting
the parameterisation right under extreme TC conditions .
|
14:30 Craig Steinberg (AIMS) – Long-term sea level and
current observations
|
Presentation and Discussion Notes :
Craig Steinberg surveyed the available sea level
and current meter observations acquired by AIMS since
commencing operations in 1993. Physical oceanographic
observations have been concentrated on a transect
between Scott Reef and Adele Island. Tide gauges have
been deployed at Rowley Shoals, Scott Reef and Adele
Island in conjunction with an AIMS Woodside supported
baseline environmental study. A current meter mooring
has been deployed at the shelfbreak shoreward of Scott
Reef, for the purpose of monitoring low frequency
currents over the long term. This mooring was
augmented during 1985 with additional current meters
and a tide gauge set to sample tidal and super-tidal
internal wave energy. Plans for field deployments in
the vicinity of NW Cape were discussed on Day 2.
Stephen Walker asked whether telemetry could be
used for real-time operational modelling in the NW
Shelf context.
|
14:45 Lu Mason (JCU)– Tidal modelling of the Indonesian
seas: a parameterised approach
Researchers: Lance Bode, Alex Harijanto, Luciano Mason
(Dept Civil & Environmental Engineering, James Cook
University, Townsville)
Project context
- Collaboration between AIMS and JCU
- Work will form part of a student project
- Will provide a basis for corrections to satellite
altimetry measurements
Geographic domain
South East Asian, Indonesian and North West Australian
seas.
Main activities
- Collect bathymetric data of sufficient resolution and
accuracy for input into our model
- Collect sufficient tidal data for model calibration
- To perform accurate tidal modelling for use with satellite
altimetry.
Results
- The determination of whether a parameterisation approach
in the representation of islands and fine scale bathymetry
can help improve modelling outcomes.
- Improved knowledge of how complex bathymetry influence the
tide in this region.
Future plans
- To standardise and then categorise available tidal in term
of reliability and accuracy
- Add to bathymetry database
- Improve accuracy of open boundary input
- Calibrate the model for the major tidal constituents.
|
Presentation and Discussion Notes :
Lu presented the work he and Lance Bode are doing
on tidal modelling in the Indonesian seas, including
the Timor Sea and NW shelf. The motivation for the
work is the regional interest in tidal modelling per
se as well as in the low frequency motion, which
requires accurate knowledge of tidal processes. An
important application is the development of tidal
corrections for the processing of satellite altimetry
data. How do tides respond in the Indonesian region?
Do the methods developed for parameterising sub
grid-scale reef topography in the GBR work in the
Indonesian archipelago?
Altimetry tracks are interrupted spatially in this
complex region, while tides change rapidly over small
space scales. The global tidal models developed for
analysing the TOPEX satellite altimetry data including
FES 95 (Le Provost et al., Grenoble) and CSR 3.0
(Eanes, University of Texas) show lots of amphidromes
and other structural complexities in this area.
Differences between these two models in parts of the
region exceed 2m amplitude. There are large errors
even in relatively deep areas. The signal obtained by
differencing the principal lunar tide (M2) results for
the two models has an amplitude of 2 to 4 m in places
within the Indonesian archipelago.
There is plenty of in situ data available from this
region for intercomparison with models, but much of it
is of questionable reliability. Obtaining correct
bathymetry is also difficult, yet important for its
effects on phase speeds and bottom friction. Phase
speeds computed for Digitised map versus 5’ of arc
Etopo5 bathymetry data agree quite well off the NW
shelf, but as frictional differences are inversely
proportional to depth, differences in shallow water
are more critical. The biggest difference, however,
comes from errors in the phase speed.
What improvement results from using the sub-grid
scale parameterisation? This was illustrated with
reference to a 2D ¼ deg resolution model using Etopo
5 bathymetry.
Summary
- There is a need for improved tidal modelling of
this region.
- Bathymetry and in situ tidal elevation data are
inadequate (Quantity versus Quality).
- The geometric complexity of the region lends
itself to parameterisation.
|
|
