Estilarme, Coastal and Shelf Science (2000) 50,27-32 Article No. ecss.1999.0528, available online at http://www.idealibrary.com on IDEM @ Sticky Waters in the Great Barrier Reef E. Wolanski" and S. Spagnol Australian Institute of Marine Science, PM B No. 3, Townsville M C, Qld. 4810, A ustralia Received 1 September 1998 and accepted in revised form 1 March 1999 The Great Barrier Reef is a mosaic of regions of high and low reef density. Current meter observations upstream from a region of high reef density revealed that the tidal and low-frequency currents were steered away from the region during spring tides but not during neap tides. A mathematical model suggests that this effect was due to both tidal friction and to the dissipation of energy by eddies behind reefs at spring tides. For a high reef density region, this results in a longer residence time at spring tides than at neap tides. Conversely, this effect also diminishes connectivity between regions of high and low reef density at spring tides. This process may affect the recruitment and dispersion of fish and other larvae in the Great Barrier Reef. It may also invalidate the use of satellite altimetry and tidal harmonic analysis for currents in the Great Barrier Reef. (C) 2000 Academic Press Keywords: water circulation; tidal friction, eddy; coral reefs; Great Barrier Reef; Australia Introduction from neap tides to near-spring tides (Ullman & Wilson, 1998). In the upper reaches of the Fly River, The bottom friction stress is a non-linear function of the same effect results in the mean sea level being velocity (Nihoul et al., 1989). This results in a non­ higher, by typically 10 cm at spring tides (3-5 m tides linear interaction between wave-induced, tidal and peak-to-trough) than at neap tides (1 m tides; low-frequency currents. Oceanographers call this Wolanski et al., 1997). effect tidal friction (Pedlosky, 1982) and engineers a The Great Barrier Reef (Figure 1) is characterized radiation stress (Massei, 1989). Radiation stresses can by a juxtaposition of regions of low reef density control water circulation over a barred beach (Slinn (where the reef block only 10% of the length along et al., 1998). There are many reported instances the shelf) and high reef density (where the reefs block where the net currents are significantly reduced by the about 90% of the length; Pickard et al., 1977). Each presence of strong reversing currents driven by waves of these regions is a few hundred km in length. or tides. Among the first reported cases was a shallow Previous studies of reef oceanography have largely gulf in South Australia (Provis & Lennon, 1983). neglected to consider this large-scale variability. A Waves enhance the bottom friction for wind-driven large spring-neap tide cycle exists on the Great Barrier currents, reducing the flushing of shelf waters and Reef. Therefore, one expects on physical grounds shallow bays such as the Irish Sea and Boston Harbor that regions of high reef density may be less per­ (Signell et al., 1990; Davies & Lawrence, 1995; meable to low-frequency currents at spring tides Signell & List, 1997). Tides also enhance the bottom than at neap tides. Wolanski (1994) coined the friction for wind-driven currents. As a result the term ‘ sticky water ’ to explain this likely effect. His low-frequency currents in Torres Strait and the Gulf supporting field data were however sparse, consisting of Carpentaria are measurably reduced, by a factor of only of drogue trajectories over five days and of two to three, by the strong tidal currents (Wolanski, occasional satellite pictures of chlorophyll distribution 1993; Wolanski et al., 1988). suggesting trapping at spring tides in regions of high This effect introduces low-frequency variability in reef density. regions with large differences between spring and In this study, current meter data are used to neap tides. For instance in the Hudson River estuary, unambiguously demonstrate the ‘ sticky water ’ the bottom drag coefficient increases by about 30% phenomenon in the Great Barrier Reef. A mathemati­ Full sized figures, tables and animations are stored on the CD- cal model is used to show that this effect is due to not ROM accompanying this article. Use a Web browser to access only bottom friction but also to energy dissipation in the start page ‘ default.htm ’ and follow the links. The help file ‘ help.htm ’ provides answers for some common problems. secondary flows behind reefs. It is shown that the ‘'E-mail: [email protected] combined effects of tidally-modulated friction and 0272-7714/00/010027 + 06 $35.00/0 (C) 2000 Academic Press 28 E. Wolanski and S. Spagnol T a b le 1 . Current meter mooring sites, January-March 1994 this region. The trajectories of water-born tracers were predicted from these data using the Lagrangian Water depth Elevation (m) advection-diffusion model described by Oliver et al. Site (m) of current meters (1992) for which the eddy-diffusion coefficient was set to 3-0 m 2 s~ \ A 37 10 and 18 Tidally-predicted currents were calculated from B 55 10 and 30 field data using tidal harmonic analysis. The tidally- C 65 20 D 114 38 predicted currents include the mean current over the E 7 5 whole period of observations. The residual currents were calculated as the difference between the ob­ served and tidally-predicted currents. The wind- driven currents were calculated as the linear fit energy dissipation in secondary flows result in a higher between wind and residual currents. residence time of water in a region of high reef density The results from the field and the model were at spring tides than at neap tides. This effect also visualised using IBM’s Data Explorer (Gallowayet al., diminishes connectivity between regions of high and 1995). low reef density at spring tides. Possible implications for the dispersion and recruitment of coral fish larvae are discussed. Results The waters were vertically well-mixed in temperature M ethods and salinity. The tides were semi-diurnal with a strong The field study was carried out along a cross-shelf spring-neap diurnal cycle (c. 3 m spring tides peak-to- transect on the outer shelf of the central Great Barrier trough, c. 1 m neap tides). In calm weather, a strong Reef of Australia (Figure 1). The transect passes longshore southward current, about 0-2 m s-1, was between Bowden Reef and Darnley Reef. North of observed on the outer shelf (site D) and this net Bowden Reef, reef density is low, i.e. the reefs block longshore current was also apparent on the shelf at site about 10% of the distance along the shelf. South of A (Animations 1 and 2). The tidal currents were Bowden Reef the reef density is high, i.e. the reefs steered very differently by the topography both at block about 90% of the length along the shelf. neap tides (Animation 1) and at spring tides (Ani­ Offshore, in the adjoining Coral Sea, the net mation 2). At site A the currents were longshore, with is southwards with the East Australian current the water flowing southward at rising tide. At sites B (Wolanski, 1994), and in calm weather this southward and C they rotated anti-clockwise with the main axis current also prevails through the Great Barrier Reef. oriented cross-shelf. At site D, on the shelf slope, the Hence the transect line is located just upstream of a currents also rotated with the tides, the main axis region of high reef density. being cross-shelf. These tidal currents modulated a Vector-averaging Aanderaa and InterOcean S4 longshore southward net current. At site E on the reef current meters were deployed along a cross-shelf crest the currents reversed 180° with the tides with no transect at sites A to D (Figure 1) from January to rotation because of the topography of the channel, and March 1994. The distance between B and D is 46 km. they were the largest at spring tides for all sites, A fifth mooring was located at site E in the only gully peaking at 1-3 m s - \ 9 m deep and 54 m wide through the reef crest at During the two days shown in Animations 1 and 2 Darnley Reef. A tide gage was also bottom-mounted for respectively neap and spring tides, there was a net in shallow waters at Old Reef. Wind data from nearby southward current of about 0-15-0-2 m s-1 at both Davies Reef were also obtained. CTD data were inshore and offshore ends of the region of high reef obtained at each mooring site at moorings’ deploy­ density (sites A and D). During that time calm ment and recovery. Table 1 summarizes the water weather prevailed and the wind-driven currents were depth and immersion depths of the meters. All current negligible. These two animations illustrate what meters and the tide gage recorded 30 min averaged happens when in calm weather a net current meets a currents. The water depth on the shelf varies between region of high reef density. At neap tides (Animation 40 and 100 m. In this region only the crest of the reefs 1) the currents at site B pointed for several hours come out of water at low spring tides. towards the passage between Old and Darnley Reef. The depth-averaged 2-D model of King and Hence, the current was able to filter through the reef Wolanski (1996) was used to calculate the currents in matrix. However, at spring tides (Animation 2) the Sticky waters in the Great Barrier Reef 29 currents were deflected offshore or inshore and largely the matrix of reefs (Animation 6).