Eddy generation in the Agulhas and Madagascar ... - CiteSeerX

29 downloads 0 Views 2MB Size Report
data from 3 very different spaceborne sensing systems — ATSR, TMI and SeaWiFS. ATSR is an infra-red radiometer providing sea surface temperature (SST) at ...
Eddy generation in the Agulhas and Madagascar Retroflections: Feature tracking with thermal and colour imagery Graham D. Quartly and Meric A. Srokosz Southampton Oceanography Centre, Empress Dock, Southampton, Hants, SO14 3ZH [email protected]

Abstract The oceans around South Africa contain two important retroflecting current systems — the Agulhas Retroflection to the south of South Africa and a smaller system to the south of Madagascar. Both are rapidly-evolving systems, with considerable change in the position of the currents, and the shedding of eddies. To understand the evolution of these systems we examine data from 3 very different spaceborne sensing systems — ATSR, TMI and SeaWiFS. ATSR is an infra-red radiometer providing sea surface temperature (SST) at high spatial resolution; the problem of cloud cover mean that short-period composites are often far from complete. TMI determines SST from microwave emissions; this affords a view through most clouds, but the spatial resolution is much poorer. SeaWiFS uses visible wavelengths to provide images of chlorophyll content; it suffers from similar cloud problems to ATSR. None of these sensors provide a direct measurement of currents, but give an indication of coherent flow features. Ultimately, the task is to utilise the best aspects of each dataset to provide an integrated view of these complex current systems.

1.

Introduction

The principal exchange of surface waters between the Indian and Atlantic Oceans takes place via the Agulhas Retroflection, a sharp bend in the Agulhas Current to the south of South Africa (Fig 1). This feature is not fixed in location, but elongates westward over a period of a few weeks before the 'pinching off' of an Agulhas ring, which then moves northwestwards. MOZAMBIQUE CURRENT NG

23˚S

CU LH AS

Port Elizabeth

EAST MADAGASCAR RETROFLECTION

AG U

G BEN

33˚S

RR

Natal Bight

U

28˚S

EN T

UPW ELA

ELLI

Delagoa Bight

38˚S AGULHAS RETURN CURRENT Agulhas Plateau

AGULHAS RETROFLECTION

43˚S

5˚E

20˚E

35˚E

50˚E

Fig. 1. Schematic of the main currents around South Africa. A similar but smaller region lies to the south of Madagascar. Although much less well surveyed, there are drifter records that imply that it also occludes rings that progress westwards

(Lutjeharms, 1988). De Ruijter et al. (1999) suggest that rings shed from the Madagascar Retroflection could be responsible for most of the Natal Pulses, large meanders of the Agulhas Current that propagate polewards along the South African coast (Gründlingh, 1979). These in turn are believed to be responsible for initiating ring shedding by the Agulhas Current (van Leeuwen et al., 2000). Thus events in the Madagascar Retroflection may have an important rôle in the total flux of warm salty Indian surface water into the South Atlantic. Many researchers have used altimetry to study the currents in this area. However it is difficult to monitor the currents accurately, given the uncertainties in the geoid, the wide spacing of TOPEX/Poseidon tracks (Fig. 2) and the marked changes between an altimeter's revisits. [ The ERS altimeters offer much finer spatial sampling at the expense of poorer temporal sampling. ] Gründlingh (1995) tracked various eddies using TOPEX/Poseidon data. At the southern end of the Mozambique Channel he noted that most anticyclonic ('warm core') eddies moved westwards, whereas the cyclonic ('cold core') ones sometimes headed southwards.

-23

m 0.3 -28 0.25 0.2

-33

0.15 -38

0.1 0.05

-43 5

20

35

50

Fig.. 2: Rms variability in sea surface height from TOPEX/Poseidon, with the altimeter ground-tracks overlaid. Here we examine the wide swath information available from three different sensors, although none of these make a direct observation of currents.

2.

Infra-red observations of SST

The ATSR (Along-Track Scanning Radiometer) is a multi-channel infra-red radiometer on-board the ERS-1 & ERS-2 satellites, which observes the sea surface at a resolution of 1 km. Because of its dual-viewing geometry and good on-board black body calibration it is able to recover sea surface temperature (SST) to an accuracy of 0.3 K (Mutlow et al., 1994). Together the ERS satellites have monitored the Earth for ~10 years, although the ATSR dataset is not quite continuous because of instrumental problems. Figure 3 shows ATSR-1 data for a period of about a month during July-August 1994. Each image is a composite of 3 days' data (since, in the absence of clouds, such a period is sufficient for nearly global coverage by ATSR). Although the series portrays a relatively good selection from the ATSR archive, clouds mar the images. Composites calculated over a larger period

would clearly have fewer gaps, but would suffer from smearing of features that move within the compositing interval. In the sequence shown there is initially a well-defined retroflecting current at 45˚E 28˚S (Fig. 3a), which during the ensuing month heads further south (Figs. 3b,c,d) before appearing to break off as a separate feature. Clouds make the tracking of such features very difficult. The increased evaporation over warm features makes them more prone to obscuring by cloud (Fig. 3f), and simple edge detection algorithms have to be wary of spurious cold SST values due to undetected cloud (e.g. the thin blue strip on the edge of the detected cloud at 45˚-46˚E 28˚-30˚S in Fig. 3e). These composite images indicate the narrowness (~50km) of the features being studied, and also the first few show the relatively cold water (~21˚C) upwelled in the region immediately to the south of Madagascar where the shelf slope gets shallower (Lutjeharms and Machu., 2000, DiMarco et al., 2000).

Fig. 3: SST composites to the south of Madagascar as observed by ATSR during alternate 3-day periods in 1994. a) 25-27th Jul. b) 31st Jul-2nd Aug. c) 6-8th Aug. d) 12-14th Aug. e) 18-20th Aug. f) 24-26th Aug.

3.

Passive microwave observations of SST

A very different technique is to infer SST from the emissions at microwave frequencies. The SST effect is most pronounced at low frequencies (6-11 GHz). Most microwave radiometers flown in space have operated at higher frequencies; one of the few exceptions is the TRMM Microwave Imager (TMI) which has been operating since December 1997. Its lowest frequency is 10.7 GHz, which is adequate for SST retrievals above 10˚C. As the diffraction-limited footprint is proportional to frequency, the effective resolution of TMI is ~60 km (Kummerow et al., 1998) despite TRMM's low altitude (350 km). The accuracy of SST retrievals is ~0.6K (Wentz and Meissner, 1999; Chelton et al., 2000),

with some dependence upon wind strength and direction. Although nearly-complete coverage is achieved in 1 day (on account of its wide swath and ability to observe through non-raining clouds), we have again produced 3-day composites to help reduce the measurement errors. The sequence shown in Fig. 4 comes from January-February 2000 (summer) and thus has much warmer values than in Fig. 3. Figures 4a-d appear to show a pulse of warm water along the path of the East Madagascar Retroflection reaching 44˚E 28˚S (Fig. 4d). However other large patches show significant change in the six days between composites; this is unlikely to be due to advection but may be due to a change in the wind field, either through different biases in the SST recovery, or through wind's control of the "skin effect" (the difference between the top millimetre of the surface sensed from space and the bulk temperature of the top few metres of water). This highlights a problem for both infra-red and passive microwave SST retrievals: the "observation" is very much of a surface parameter which may be only weakly related to the subsurface signal. For example, rings shed by the Agulhas Retroflection are known to lose their surface manifestation of temperature within a few months (Walker and Mey, 1988), although they may be detected by altimetry up to two years after shedding, on account of their sea surface height signature.

Fig. 4: SST composites to the south of Madagascar as observed by TMI during alternate 3-day periods in 2000. a) 22-24th Jan. b) 28-30th Jan. c) 3rd-5th Feb. d) 9-11th Feb. e) 15-17th Feb. f) 21st-23rd Feb.

4.

Ocean colour observations of chlorophyll

A method of overcoming the "surface barrier" is to monitor features using radiation that penetrates well below the surface and is thus not so responsive to atmosphere's effects on the surface skin. Ocean colour sensors such as SeaWiFS (launched August 1997) detect the reflection of visible light within the top ten metres or more. With careful processing, including

accurate modelling of atmospheric radiances, chlorophyll concentrations can be retrieved with an absolute accuracy of ±35% (Hooker and McClain, 2000). However the quality of the recovered data may allow the detection of spatial or temporal changes much smaller than that. Figure 5 shows a sequence of 5-day composites of SeaWiFS data. Again there are missing pixels due to persistent cloud, but the effect is less marked than in Fig. 3 as SeaWiFS has a wider swath than ATSR and a larger compositing period is used. There is a region of higher chlorophyll content just to the south of Madagascar, where there is year-round upwelling of cold nutrient-rich water (cf. the infra-red images for a different period in Fig. 3). A similar feature occurs along the South African coast between Richard's Bay and Durban (29˚-30˚S), where again there is a shallow shelf.

Fig. 5: Composites of chlorophyll concentration for the region to the south of Madagascar as observed by SeaWiFS during alternate 5-day periods in 2000. a) 25-29th May, b) 4-8th Jun., c) 14-18th Jun., d) 24-28th Jun., e) 4-8th Jul., f) 1418th Jul. The green arrow indicates probable migration of an eddy. The sequence of images shows a patch of heightened chlorophyll moving southwest then dividing, with the main component heading west. It is conjectured that this coherent "blob" is an eddy containing nutrient-rich waters upwelled off the coast of Madagascar. Although chlorophyll is not a conservative tracer in that the value for a parcel of water can vary over time due to biological productivity, ocean colour appears to do a reasonable job of delineating features for many months of the year. However, during the period of the spring bloom (roughly June to August) there is a lot of temporal variation throughout the area of interest as phytoplankton growth takes off in response to increased light levels and shallowing of the mixed layer, and this can make it harder to discern the presence of eddies.

5.

Future work

For reasons of limited space this paper has concentrated on the Madagascar Retroflection. Although bigger in size than it's neighbour, the Agulhas Retroflection is harder to observe. Cloud cover in that region is typically 90% obscuring observations of the sea surface by ATSR and SeaWiFS, whilst the chosen orbit for the TRMM satellite precludes the recovery of passive

microwave observations of SST south of 38˚S (see Fig. 1). Further passive microwave datasets are becoming available: the Multichannel Scanning Microwave Radiometer (MSMR) on-board the Indian satellite Oceansat-1 (launched in 1999) and the Advanced Microwave Scanning Radiometer (AMSR, due for launch in 2001/2) have channels operating at frequencies as low as 6.6 GHz, which is essential for accurate microwave recovery of SST from colder waters. These lower frequencies imply a larger instrument footprint; it is yet to be seen whether the coarser resolution will still allow useful monitoring of sharp thermal features, or is more appropriate for climatological studies.

6.

Acknowledgements

We are grateful to the Rutherford Appleton Laboratory for provision of the ATSR data, NASDA/EORC for the TMI data and NASA/GSFC for the SeaWiFS data.

7.

References

CHELTON, D. B., F. J. WENTZ, C. L. GENTEMANN, R. A. DE SZOEKE, and M. G. SCHLAX, 2000: Satellite microwave SST observations of transequatorial tropical instability waves. Geophysical Research Letters, 27, 1239-1242. DE RUIJTER, W. P. M., P. J. VAN LEEUWEN, and J. R. E. LUTJEHARMS, 1999: Generation and evolution of Natal Pulses: Solitary meanders in the Agulhas Current. Journal of Physical Oceanography, 29, 3043-3055. DIMARCO, S. F., P. CHAPMAN, and W. D. NOWLIN, 2000: Satellite observations of upwelling on the continental shelf south of Madagascar. Geophysical Research Letters, 27, 3965-3968. GRÜNDLINGH, M. L., 1979: Observation of a large meander in the Agulhas Current. Journal of Geophysical Research, 84, 3776-3778. GRÜNDLINGH, M. L., 1995: Tracking eddies in the southeast Atlantic and southwest Indian oceans with TOPEX/POSEIDON. Journal of Geophysical Research, 100, 24977-24986. HOOKER, S. B. and C. R. MCCLAIN., 2000: The calibration and validation of SeaWiFS data. Progress in Oceanography, 45, 427-465. KUMMEROW, C., W. BARNES, T. KOZU, J. SHIUE, and J. SIMPSON, 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. Journal of Atmospheric and Oceanic Technology, 15, 809-817. LUTJEHARMS, J. R. E., 1988: On the role of the East Madagascar Current as a source of the Agulhas Current. South African Journal of Science, 84, 236-238. LUTJEHARMS, J. R. E. and E. MACHU, 2000: An upwelling cell inshore of the East Madagascar Current. Deep-Sea Research, 47, 2405-2411. MUTLOW, C. T., A. M. ZÁVODY, I. J. BARTON, and D. T. LLEWELLYN-JONES, 1994: Sea surface temperature measurements by the along-track scanning radiometer on the ERS-1 satellite: Early results. Journal of Geophysical Research, 99, 22575-22588. VAN LEEUWEN, P. J., W. P. M. DE RUIJTER, and J. R. E. LUTJEHARMS, 2000: Natal pulses and the formation of Agulhas rings. Journal of Geophysical Research, 105, 64256436. WALKER, N. D. and R. D. MEY, 1988: Ocean/atmosphere heat fluxes within the Agulhas Retroflection region. Journal of Geophysical Research, 93, 15475-15483. WENTZ, F. J., and T. MEISSNER, 1999: AMSR Ocean Algorithm. Algorithm Theoretical Basis Document version 2 (RSS Tech. proposal 121599A), 58pp.