.
.
.
.
.

.
.

WA ocean movies, 1993-2000

David Griffin, John Wilkin, Alan Pearce and Chris Chubb
Version 3.6.1 11 April 2002
[updated (www) version] [obtaining a CD, revision history]
[Cataloguing-in-Publication entry of the CD-ROM version]

[installing a movie player]

38°S-21°S, 106°E-117°E (Exmouth Gulf to Albany), 1993-2000 with frames at 5-day intervals. Sea surface height, current and temperature are averages over 10 days, on 18km, 36km and 5km grids, respectively. Trajectories of satellite-tracked, free-drifting buoys are also included.
As above, but for southern WA (39°S-31°S, 110°E-125°E).
Slightly smaller (36°S-21°S, 108°E-117°E) region than movie type 1, for 1995-2000, several frames per day showing more detail of surface temperature.
Validation of surface currents using 'model drifters'.
1993-1998 shown side-by-side, with a simulation of how larvae of western rock lobster are carried about by the winds and currents.
Future work: using Chlorophyll-a estimates
Appendix: Technical details
References
Authors
Acknowledgements

The text is written for a general audience, and an attempt is made to use the movies to demonstrate some basic oceanographic principles. There is also a short discussion of how these data have been applied to the question of where larvae of western rock lobster spend their year in the open ocean. For a fuller discussion of the project methods and results, the reader is referred to Griffin et al. (2001a) and Griffin et al. (2001b).

1. West of WA, frames every 5 days

[1993] [1994] [1995] [1996] [1997] [1998] [1999] [2000] [2001] [2002] [2003] [2004] [Contents]
(www version only, AVI format movies:) [1993] [1994] [1995] [1996] [1997] [1998] [1999] [2000] [2001] [Contents]
Click on a year to play a movie, but first read [Installing and running a movie player]

What is shown

The left panel is a colour-coded map of tidal-residual sea level. The scale shows that bright red, for example, means the sea level is 0.4m higher than the tidal prediction. Black arrow heads show the ocean current velocities. Closely-spaced purple arrow heads show the positions and velocities of Argos satellite-tracked free-drifting buoys, at 12-hour intervals over a 5-day period.

In the centre and right panels, the current velocities and drifter trajectories are repeated from the left panel, the sea level map is shown just as contours (and includes the time-mean), while the colour fill shows the sea surface temperature composited over a 10-day period. In the centre panel, the temperature is shown as the difference from the normal pattern for the time of year. The scales show that in the centre panel, bright red means the water was about 3° warmer than usual, while in the right panel it indicates a temperature of 27°C (at this time of the year - the scale shifts to follow the average seasonal cycle).

Some explanation

Think of the left panel as a coloured-in weather map of the ocean. In the same way that meteorologists can determine the winds from maps of atmospheric pressure, oceanographers can determine the surface current from a map of sea level. In September 1992, a satellite was launched with a radar altimeter that measured sea level accurately enough for this to be done. The arrow heads show the geostrophic current going clockwise around a sea level depression, and anticlockwise around a high. The current is fastest where the sea level changes abruptly, for the same reason that winds are strongest where the isobars on a weather map are close together. The accuracy of the satellite 'altimetric' current can be gauged by comparing its estimates with the trajectories of the drifting buoys. The drifter going around an eddy at 33° 30'S, 111°E, for example, agrees with the satellite estimate. A quantitative comparison of the drifter velocities with altimetric estimates shows that the typical difference is 0.25m/s, or half a knot. Most of that is because an altimeter can only measure sea level directly under itself. The Topex/Poseidon altimeter takes 10 days to complete a global grid of lines 250km apart, by which time the ocean has changed somewhat. The fairly wide spacing of the satellite tracks makes it impossible to see all the detail of the sea surface, so the maps are smoother than reality, and currents often weaker. The relative error (error divided by the speed) is least where the current is strong: drifter speeds of 1.5m/s (3 knots) are quite accurately estimated by altimetry.

The right panel will be more familiar, since satellites measuring sea surface temperature have been flying since the 1980s. The AVHRR is a scanning instrument, so each overpass of the satellite returns a 2000km-wide swath of data. However, it cannot see through cloud, so again, it ends up taking typically 10 days to piece together a complete ('composited' - see Appendix) map. If cloud is very persistent, of course, the resulting composite can be incorrect in places, as can sometimes be seen by looking carefully at the drifters which are colour-coded (a circular back-drop to the arrow head) by the temperature that they recorded - although that, too, can sometimes be wrong.

In the centre panel, we've shown the 'seasonal anomaly' of temperature, or difference from the average seasonal cycle, so that its obvious straight away (without comparing with other years) whether its warmer or colder than usual. Showing the anomaly also makes the eddies and fronts stand out, since the colour bar only has to span 8 ° (-4° to +4°), instead of 16° (12° to 28°). Our estimate of the average seasonal cycle is provided by the CSIRO Atlas of Regional Seas, which draws on decades of sampling from ships to build up a description of the average seasonal cycle of temperature, salinity, etc, around Australia. The spatial resolution of the atlas is 0.5°. In each grid box, the mean, annual and semi-annual cycles are estimated.

The surface temperature maps, in conjunction with the sea level maps, provide the information needed to initialise hydro-dynamic models of the ocean circulation, in much the same way that numerical weather forecasting models need atmospheric pressure and temperature data in order to produce accurate forecasts.

The temperature data show the Leeuwin current bringing warm water southwards down the WA continental slope. The Current meanders and sheds eddies along the way, dissipating some of its flow, but much continues around Cape Leeuwin into the Great Australian Bight. Comparing left and centre panels, one sees that sea level is higher where the water is warmer, especially when an eddy of warm water is formed, such as the one centered near 32°S 113°E. Sea level lows, in contrast, do not always have a strong sea surface temperature signature, although we know from sampling the depths that subsurface cold water is typically 50m closer to the surface in the centre of a deep low. Watch for cyclonic eddies (low sea level and temperature, clockwise rotation) that form off the Abrolhos islands (for example, on 30 June 1995, in association with the Leeuwin current branching offshore and forming a much larger, anti-cyclonic eddy. For more discussion of the Leeuwin current, see the CSIRO web site.

Inter-annual differences and the influence of El Nino

An El Nino is when sea level and surface temperature are lower than usual in the western equatorial Pacific, and high in the eastern equatorial Pacific. The strongest influence, within Australia, of low sea levels along the equator is off WA. This is because sea level variations travel anti-clockwise around land masses in the southern hemisphere, weakening as they go. Hence, off Western Australia, the El Nino signal appears first in the north and has almost vanished at the border with South Australia.

El Nino conditions occurred over the summers of 1992-93 and 1997-98. The influence of El Nino can be seen in the altimeter data as low (about 0.3m) sea level over the continental shelf and slope. The sea level anomaly is greatest at the coast, and to the north, so the currents are more northward (actually less southward - same thing) over the continental slope, and more westward (actually less eastward) in the deep ocean. The influence on currents in the deep ocean is hard to see amongst all the strong eddies, but the influence over the continental slope is clear. Compare, for example, the currents during the El Nino in February 1998 , with the situation in 1996 (above) which was during the opposite phase of El Nino, called La Nina. The difference is even greater earlier in the summers (see the animations). For example, while there was an 'unseasonal' Leeuwin current running in November 1995 (early La Nina), there was certainly none in November 1997 (early El Nino). An important consequence of the Leeuwin current ceasing to flow is that the summer southerly winds are then able to drive coastal upwelling , which is usually suppressed by the opposing flow. Upwelling occurred between Ningaloo and NW Cape in November 1997 and January 1998, and between Cape Leeuwin and Cape Naturaliste in February 1998.

The strongest La Nina conditions during 1993-2000 occured during the summer of 1999-2000 (see 5 January 2000). The extra strength of the Leeuwin during La Nina is clear, another striking image being in 8 June 2000, by which point a huge (about 400km by 200km) region of high sea level existed off the continental slope from north of the Abrolhos to Perth. Within that region is the small cyclonic eddy in close to the Abrolhos, which seems, as discussed above, to often be associated with the formation of large anticyclonic eddies.

2. South of WA, frames every 5 days

[1993] [1994] [1995] [1996] [1997] [1998] [1999] [2000] [Contents]
These movies zoom in on the southern regions of WA and document the rounding of Cape Leeuwin by the Leeuwin Current. The lesser latitudinal extent of the plotted area allows a tighter temperature scale to be used. Watching the movies, you will see many examples of when the Leeuwin separated from the coast to create pairs of cyclonic (clockwise rotating, originating upstream of the offshoot) and anti-cyclonic (anti-clockwise, downstream of the offshoot) eddies that then proceed to drift slowly westwards.

3. Several-per-day maps west of WA

[1995] [1996] [1997] [1998] [1999] [2000][Contents]

What is shown

The arrow heads again show the satellite altimetric current. Drifter positions are shown every 6 hours from 5 days before the image time. The sea surface temperature anomaly, however, is processed differently, to retain more detail. When you run the movie, you can see small features moving along with the current, where there are long enough gaps in the cloud cover. To retain the detail of the raw data, we've not averaged or composited them over time. That is, we've shown an image for every individual satellite pass. That raises the problem of what to do about clouds, and other atmospheric degradations of the data. Whenever the data are flagged as clearly 'bad' (see Appendix), the image simply freezes at the most recent 'good' estimate of the sea surface temperature.

Some explanation

In movie type 1, the gaps in sea surface temperature images due to clouds were interpolated (estimated between observations) by seeking only to estimate the average temperature over a 10 day period. But that process smears out a lot of the detail, so the movie showed how large bodies of water changed their position, but not the movement of water within each body. To see movement of small features, we have to be content to see a very incomplete picture, which also has errors. One reason for the errors is that not all clouds (especially at night) are easy to detect automatically, since their apparent 'temperature' can be close to that of the ocean. A human observer can usually tell cloud from ocean better than today's automated processing. For these maps, which are intended solely for visual inspection (rather than ingestion by the numerical ocean model), we've only blanked out the really obvious areas of cloud. Nor have we blanked out areas where a 'warm skin' has formed on the sea, as happens during light winds. We've relaxed these quality controls so that less good data is erroneously discarded, so each image is as complete as possible.

When you watch the movies, compare the movement of the drifters with what the two remote-sensing techniques (temperature and sea level) tell us about the ocean. We went through this exercise ourselves to assess the strengths and weaknesses of the two techniques, and to provide us with a yardstick of performance so we'd know whether changes to the way we process the data were helping or not.

4. A short segment with 'model drifters'

[Animate (gif format - no special player required)][Contents]
This short movie loop demonstrates the use we are making of the altimeter estimates of the surface current. The colour background of the left-hand panel shows the sea level measured by the altimeter and tide gauges. The black dots move (when you run the animation) at the velocity that we compute, as described above and shown elsewhere as arrow heads, from the sea level. We call these black dots 'model drifters' because they would go at exactly the same speed as real drifters (with sea-anchors) if our estimates of the current were perfect. One such real drifter can be seen at bottom left in purple. In the right-hand panel, the movement of the model drifters is super-imposed on satellite images of the sea surface temperature. The motion of the model drifters corresponds remarkably well with the movement of the water as revealed by the thermal imagery, suggesting we can use the altimeter-derived velocities to track the movement of things in the water.

5. Application to western rock lobster

[Animate (fli format)] [Animate (avi format) - www version only] [Contents]
The motivation to develop and test the 'model drifters', shown above, is to learn more about the movement of larvae of commercially important fisheries, primarily those of Western Rock Lobster. We do this by modifying the basic model-drifter to make it more like a larval lobster. That is, we assume it comes right to the surface at night and be more subject to the wind (like a real drifter with no sea anchor), and let it descend during the day to depth where the currents are a little different. Details of how we do this are given by Griffin et al. (2001a) who explain the meaning of the text along the top of the plot.

What is shown

Each of the 6 panels of this movie shows the model-simulated fate of a particular year-class of lobster larvae. The model uses our best estimates of the winds and ocean currents for each year, and all that is presently known about how lobster larvae of various ages move up and down in the water column by day and night.

The background colour field shows the maps of sea level from the altimeter, while the purple arrows on a 1.9° (latitude and longitude) grid are 10-day averages of the wind stress (approximately proportional to the square of the wind velocity).

The movie starts in early November as the eggs carried by berried females start to hatch and the larvae commence their life in the plankton. The model is of course an idealisation of reality, and since we do not know how the pattern of hatching varies through the season or from one year to the next, we model the process as simply as possible. Every 5 days until the end of February, 25 more modelled larvae are hatched from a constant set of locations (distributed evenly throughout the fishery), joining their older class mates as they are swept about by the winds and currents. The colour of each individual larva indicates where it will eventually go (see key). The ones destined to return to the coast are shown in red, so we can see if there is anything special about where they are in relation to their class mates, at any particular time of the year. According to this model, there is not. The survivors are mixed fairly randomly through the population at any instant. What is most fascinating to see is the degree of mixing that is constantly occurring within the population.

Comparing the six year-classes, the interesting result of this model is that there isn't a huge difference in the direct influence of winds and currents from one year to the next. The biggest difference is between the 1993 (top left) and 1996 (bottom left) year classes. These were years of weak and strong Leeuwin current, respectively, as discussed above. In early 1993, many model larvae are carried a long way north because the Leeuwin current was too weak to counter the influence of the southerly winds. In contrast, the strong Leeuwin current of 1996 swept many model larvae around Cape Leeuwin towards the Great Australian Bight, commencing as early as April. The net effect, however, is that the number remaining in the vicinity of the fishery is fairly constant across years, and is actually quite a large proportion of the number simulated (considering that millions of larvae are produced by each female in reality, the vast majority of which must die).

The red arrows that appear occasionally on the land show the observed rates of settlement of lobster puerulus, the final planktonic larval stage. These rates of settlement form the basis of the catch predictions that are made by Fisheries WA. A goal of the project was to test the widely-believed hypothesis that the direct influences of the winds and current strength accounted for the very large (a factor of five) differences in settlement between years. The fact that the simulation does not explain the observed variation of settlement suggests that that hypothesis is wrong, and that the controlling influence of the environment is not included in the model. Furthermore, the fact that many model larvae do return to the coast at about the time of settlement suggests that the model correctly explains the basics of how the larvae, through their choice of how close to stay to the wind-driven surface layer, avoid being carried either too far north by the wind, or too far south by the Leeuwin current.

Future work: using ocean colour imagery

[1997] [1998] [1999] [2000] [Contents]
We know that at least one of the missing factors in the model must involve an environmental variable that varies along with sea level, from studies that have shown that good years of puerulus settlement occur when sea level is high, as tends to happen during La Nina conditions. (Conversely, sea level is low during El Nino, and settlement is poor). Temperature is one such variable and it is not included in our model as a factor influencing vertical migration, growth or mortality since the sensitivities of these to temperature are not known. Nor is food availability included, for the additional reason that estimates of this variable are not available. We are seeking funding to address these issues using several recent technological developments that make this feasible to do now.

The left and right panels are similar to those of movie type 1, while the central panel shows SeaWiFS estimates (available since September 1997) of the amount of chlorophyll-a in the water. It is clear from this image that there is more chlorophyll in the anticlockwise rotating eddy off Perth than other ocean waters. Chlorophyll is a measure of the abundance of phytoplankton, upon which all marine life, including lobster larvae, ultimately depend. One might suppose that the larvae in that eddy would grow faster and be more likely to survive to return to the fishery than larvae elsewhere.

Appendix: Data sources and methods - technical details

Here we provide a very brief outline of the data used. The dates in parentheses refer to the span of the data used for the plots shown here. Follow links to the glossary and elsewhere for further information.

Sea level maps (Jan 1993-Dec 2000)

Data from up to three altimeters (CNES/NASA Topex/Poseidon and European Space Agency's ERS-1 and ERS-2) at a time, and 12 coastal tide gauges were interpolated on an 18km grid every 5 days by a three-stage optimal interpolation. It is important to include the coastal tide gauges in the analysis because altimeter data are not useful there for a number of reasons including orbit errors near land, incorrect tidal corrections and the increased intensity of rapid (ie daily) variations of sea level. Altimeter data were obtained from NASA Physical Oceanography Distributed Active Archive Center at the Jet Propulsion Laboratory / California Institute of Technology and AVISO at Collecte Localisation Satellites. Tide gauge data (Jan 1993 - Dec 2000) were obtained from the National Tidal Facility and the WA Department of Transport. Atmospheric pressure (Jan 1992 - Dec 2000, for making the isostatic correction to the tide gauge data) was obtained from the NCEP-NCAR Reanalysis (see below), as was wind stress for estimating the sea level at the shelf break (see Griffin et al., 2001)

Currents

These are determined from the sea level maps by assuming a cyclostrophic momentum balance, which is close to a geostrophic balance but includes a small correction for the centrifugal acceleration, which is responsible for a warm core (anti-cyclonic) eddy having a smaller sea level anomaly than a cold core (cyclonic) eddy of equal angular velocity. To reduce crowding on the plots, velocities are only shown at every 2nd or 3rd grid point (ie 36km or 54km spacing).

Sea surface temperature (Jan 1993 - Dec 2000)

These data were received by the WASTAC facility in Perth and processed by CSIRO in Hobart. For the five-daily composite maps, each pixel is the 95th percentile value of all unflagged data within a 10 day, 6km data window. Data are flagged as atmospherically-affected if there is too much brightness in the visible band or too much absorbtion in the thermal band (detected as the difference between channels 4 and 5). For the single-pass (rather than 5-day composite) images, data are also flagged as 'bad' if the temperature anomaly from the seasonal climatology is outside of -2.8 to +3.1°. The 1993-1994 data comprise only afternoon passes (instead of round-the-clock) and were received by ACRES and processed by CSIRO in Hobart.

Chlorophyll-a (Sept 1997 - Dec 2000)

These data are from the SeaWiFS project at the NASA Goddard Space Flight Center. The processing version is SeaDAS3.

Wind (Jan 1993 - Dec 2000)

Daily averages of the wind stress were provided by the NCEP/NCAR 40 Year Reanalysis Project . These data are used to compute the Ekman velocity which we add to the altimetric geostrophic current estimate, eg for making comparisons with drifter velocities, or modelling the movement of sub-surface larvae.

Drifters (Jan 1993 - Dec 2000)

Drifter data were obtained from the World Ocean Circulation Experiment Surface Velocity Program Data Assembly Center via MEDS. The drifters have sea-anchors centred at 15m depth, and hence are less subject to the effects of the wind than objects right at the surface. However, the drifters often lose their drogues and a future version of these movies will include an indication of whether this has happened. The current source of Global Lagrangian Drifter data is AOML.

References

Griffin, D A, J L Wilkin, C F Chubb, A F Pearce and N Caputi (2001a). Ocean currents and the larval phase of Australian western rock lobster, Panulirus cygnus. Marine and Freshwater Research 52, 1187-99.

Griffin, D A, J L Wilkin, C F Chubb, A F Pearce and N Caputi (2001b). Mesoscale oceanographic data assimilative modelling with application to Western Australian fisheries. Fisheries Research and Development Corporation final report 97/139.

The authors

David Griffin is a Physical Oceanographer at CSIRO Marine Research in Hobart. He can be reached on (03) 6232 5244 or by email at David.Griffin@marine.csiro.au.

John Wilkin is a Physical Oceanographer who moved from CSIRO to NIWA in New Zealand, where he continued to work on this project. John's work with altimetry was what sparked this project, and enabled it to succeed. John has since moved to Rutgers University in the US.

Alan Pearce is a Physical Oceanographer at CSIRO Marine Research in Perth. He can be reached on (08) 9422 8215 or by email at Alan.Pearce@marine.csiro.au.

Chris Chubb, a Principal Fisheries Scientist, is the senior rock lobster biologist at Fisheries WA . He can be reached on (08) 9246 8404 or by email at cchubb@fish.wa.gov.au .

Acknowledgements

The work leading up to, and the production of, this CD was supported by the Fisheries Research and Development Corporation (project 97/139). We are also very grateful to the institutions discussed above for data provided, and to many CSIRO and Fisheries WA colleagues for help with numerous tasks.

[Contents]