Persistent phytoplankton bloom events, surface cooling, and associated fronts in the Strait of Ombai were seen by Sea-viewing Wide Field-of-view Sensor (SeaWiFS), Advanced Very High Resolution Radiometer (AVHRR), and Synthetic Aperture Radar (SAR) imagery. It is suggested that the cooling events were caused by variable flow through the strait and its interaction with complex bathymetry. During the strongest cooling events a sharp, midstrait frontal feature was seen with a cross-front gradient in sea surface temperature (SST) of the order of 4°C and a gradient in chlorophyll concentration of over 1 mg/m3. The frontal feature, associated with increased ocean color signal, extended, during major events, 100s of kilometers downstream of the strait and at times influenced nearly all of the Savu Sea. We show that variability in the bloom events can be tied to seasonal monsoons, ENSO, and their effect on thermocline depth.
 In oligotrophic regions, including typical tropical seas, biological production is often limited by macronutrients. Waters below the nutricline are rich in nutrients but can only support phytoplankton growth should a mechanism for vertical exchange present itself. Three major causes (equatorial divergence, thermocline tilting/uplift, and wind stress-induced Ekman suction) have been shown to be associated with tropical blooms over basin scales [Longhurst, 1993]. Yet even over these large scales, there can be found significant differences in bloom dynamics between the major ocean basins.
 On smaller scales, it has been shown worldwide that phytoplankton blooms are often related to mesoscale features such as eddies and fronts in open ocean, shelf, and coastal waters [Wolanski and Hamner, 1988; Marra et al., 1990; Franks, 1992; Longhurst, 1993; Yoder et al., 1994; Franks and Chen, 1996]. Two of the physical driving forces for phytoplankton concentrations at fronts are strong tidal currents acting over steep or complex bathymetry and forcing through topography. In the latter case, a strong mean flow interacts with headlands or islands, producing gradients in their lee [Wolanski and Hamner, 1988; Franks, 1992]. In these situations, fronts are associated with vertical mixing and increased nutrient input into the euphotic zone.
 The Strait of Ombai is a relatively narrow and deep passage with complex bathymetry. It lies between the Indonesian island of Alor and the northern coast of East Timor and runs roughly SW-NE (Figure 1). It connects the waters of the Banda Sea in the north to those of the Savu Sea in the south. There is evidence for both strong mean flow and tidal energy in the strait [Chong et al., 2000; Molcard et al., 2001]. Given the physical structure of the strait of Ombai, it is likely that some combination of tidal currents or topographic effects will lead to fronts and vertical flows that could drive increased phytoplankton concentrations.
 In terms of seasonality the Indonesian archipelago has forcing dominated by the monsoons; the “dry” southeast monsoon (SEM), from June through September, and the “wet” northwest monsoon (NWM), from December through March. The months in between are described as the “transition” months and are typically characterised by weak, variable winds [Koninklijk Nederlands Meteorologisch Institute, 1949]. Physically forced blooms in the region are going to be modulated seasonally and in an interannual sense by variability in the depth of available nutrients. Elsewhere in the Indonesian archipelago, ENSO has been shown to modulate both the depth of the thermocline [Ffield et al., 2000] and wind-driven upwelling processes [Susanto et al., 2001], and it is well known that seasonal monsoon winds strongly influence the depth of the thermocline regionally [Gordon, 1986; Wyrtki, 1987]. Here we present evidence of a persistent upwelling event that enhances the local productivity in the strait of Ombai and that has variability on both seasonal and interannual timescales.
 We analyzed a suite of remotely sensed data sets for the region including ocean color, sea surface temperature (SST), and Synthetic Aperture Radar (SAR) imagery. Estimations of chlorophyll concentration, as seen by SeaWiFS were provided as 9 km, level 3 grided data products from the Goddard Earth Science (GES) Distributed Active Archive Center (DAAC). Further investigation of daily, 4 km, level 2 products was performed on data provided by the Remote Sensing Facility at the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Marine Labs in Hobart, Tasmania. An optimally interpolated advanced very high resolution radiometer (AVHRR) product, calculated according to Walker and Wilkin  and also provided by the CSIRO Marine Labs, was analyzed from 18 March 1991 to 15 February 1999 covering a spatial domain from 45°S to 4°S and 107°E to 160°E. The AVHRR data product had a resolution of 10 days and 5 km and was derived from daily passes. The interpolation is of great benefit in tropical regions that have an atmosphere of high cloud cover and high water vapor content.
 Individual SAR images were found through the EOS Data Gateway for the strait of Ombai, and full resolution RADARSAT-1 scansar images were provided by the Alaska SAR Facility. Postprocessing of these images was performed at the CSIRO Remote Sensing Facility. To complement the SAR image used, a very fortunate AVHRR sun-glint image was found in data provided by the facility (courtesy of P. Tildesley).
 Wind fields analyzed are a National Centers for Envrionmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) atmospheric re-analysis hindcast of atmospheric conditions from 1982 through 1998 with a grid spacing of roughly 200 km [Kalnay and Atlas, 1986]. The altimetry product used is a grided data set of sea level anomalies from the Archiving, Validation, and Interpretation of Satellites Oceanographic Maps of Sea Level Anomalies (AVISO MSLA) Topex/Poseidon-ERS product and the Darwin tide gauge data (from the Australian National Tidal Facility), covers 1990–1999.
Figure 2 shows an image of monthly averages of both SST and estimated chlorophyll for November 1997 in the strait of Ombai. What is clearly seen is a well-defined plume of cold water with high chlorophyll that appears to begin near the northern headland at the entrance to the strait and continues down the northern side of the strait in the direction of flow. A smaller feature is evident in SST off the southern side of the entrance. While the 9 km ocean color image is too coarse to show clearly the chlorophyll feature in the southern part of the strait, it is sometimes seen in the limited number of images available at the 1 km scale (remapped L2 data received from Australian Institute of Marine Science not shown). It appears that a “core” of warm water with lower chlorophyll proceeds down the strait, slightly off the strait's axis, and is bounded by the fronts associated with the colder mixing features. Figure 3 shows photos taken in the strait of Ombai showing a massive front in March of 1997. The front extended to the horizon in either direction. The brown “slick” seen at the convergence between the two water masses contained large pieces of flotsam, including logs. In and around the time these photos were taken there was significant anecdotal evidence of large pods of dolphin, schools of fish, and aggregations of feeding sea birds in the strait (T. S. Moore, personal notes and photos).
 In November 1997, at the height of the 1997/1998 El Nino, when winds were anomalously from the east (Figure 4), sea level was the lowest in at least 5 years (Figure 5) and presumably the nutricline was shallowest, there occurred one of the most “extreme” events of those seen. Across front, the gradient in surface temperature was as large as 4 degrees and gradients in chlorophyll concentration were as large as 1 mg/m3. An analysis of SST and ocean color data at higher temporal resolution (not shown) shows a succession of cold-water events (indicative of upwelling) over the entire month.
Figures 6a–6d shows a SAR image of the Strait of Ombai from a RADARSAT pass on 6 December 1997. Figures 6b–6d are close-ups of different portions of the region. In 6c and 6d some dark (“low”) backscatter regions appear adjacent to the coasts of Alor and East Timor along with edge features that suggest strong fronts. Cloudless daily SeaWiFS data available 2 days later on 8 December 1997 (Figure 7) has similar spatial pattern to the SAR image and supports the interpretation of biological activity.
Figures 6a and 6b show a feature that we had not expected. The strong signal of wave packets or wave-like features can be seen radiating from the northeastern end of the strait, near its entrance. We were fortunate to find a NOAA AVHRR image, from October 1995, affected by a strong “sun-glint”(Figure 8). A remarkably identical pattern, likely tidally forced internal waves [Osborne and Burch, 1980], can be seen radiating northward from the eastern end of the strait.
 From our visual analysis of a 3 year time series of monthly and near-weekly composite images of both SST and concurrent SeaWiFS data (including Figures 2 and 10), we have found that low temperatures in the strait of Ombai, particularly when juxtaposed with significantly warmer waters, are a good proxy for high chlorophyll concentrations (often greater than 1 mg/m3). Looking back at historical records we can then get some idea of the persistence and variability of SST and thus chlorophyll in the strait. Figure 9a shows the temporal mean SST averaged over all 10-day records from 1992 through 1998. Evidence of surfaced cold water, both in the north and south of the strait, are clearly seen in the overall average. Figures 9b and 9c show similar data but only for the months of the “dry” SEM (shallow thermocline) and “wet” NWM (deeper thermocline), respectively. When compared with the NWM the SEM shows strong temperature fronts in the strait. Figure 10, from November 1998, which was a period of a strong La Nina, offers a contrast of La Nina to El Nino conditions when compared with Figure 2. November 1998 lacks the strong bloom/cold surface water conditions.
 A time series of SeaWiFS derived ocean color for two regions around the Strait of Ombai (Figure 11) shows clearly the seasonality of the blooms, the influence of the late 1997 ENSO event, as well as the significance of the strait in the production of the blooms. The regions chosen were the northern side of the strait along the Alor coast, which we called the “bloom region,” and an area just to the east of the strait which we called the “prestrait” region. Chlorophyll concentration values from SeaWiFS derived 8 day composites were binned in these regions from late September 1997 through the end of 2000. The 8 day composite time series is the highest temporal resolution ocean color dataset for the Strait of Ombai that offers reasonable coverage given cloudiness.
 The time series shows a significant difference in the magnitude and temporal pattern of chlorophyll between the two regions, themselves physically only 20 or so kilometers apart. A clear correlation between the “bloom region” signal level and the monsoon seasonality is seen, with magnitudes peaking in the SEM, but with some bloom activity starting as early as late February/early March. The blooms cease with the onset of the NWM except in the case of late 1997, a period of strong El Nino, where chlorophyll levels are consistently high (near 1 mg/cubic meter). Disregarding the ENSO influenced late 1997 NWM, the “wet” phase of the monsoon seems to be the only period where the chlorophyll levels in the “prestrait” zone are near or greater than the “bloom” region. Missing data corresponds to periods of extended cloud and give some indication of precipitation. Of particular note is the lack of cloudiness in the 1997–1998 NWM period. The extreme anomalous dryness of this period is well documented [Gutman et al., 2000].
 The Indonesian archipelago is unique in that it is the only interocean connection that occurs at low latitude on the planet. The strait of Ombai, at the northeastern end of the Savu Sea between the islands of Alor and Timor (Figure 1), is one of two deep water passages in the interocean connection that link the waters of the Pacific and Indian Oceans. The exchange of water between the Pacific and the Indian Oceans is known as the “Indonesian Through Flow”(ITF).
 Regional thermocline depth and the strength of the ITF both have been shown to have strong seasonal variability [Wyrtki, 1987]. During the dry monsoon, from about June through September, winds in the eastern archipelago prevail from the southeast. Wind-driven currents are in the direction of the Indian Ocean, and the ITF peaks in strength. In the eastern archipelago at this time the sea level drops and the thermocline shallows. Across the archipelago, significant links have also been found between ENSO events and both the strength of the ITF and the depth of the thermocline [Gordon and Fine, 1996; Meyers, 1996; Ffield et al., 2000]. In general, during El Nino (La Nina) events, the thermocline shallows (deepens), and the ITF weakens (strengthens).
 Vertical mixing is strong in the region and it is likely that this mixing is primarily due to dissipation of tidal energy over complex bathymetry [Ffield and Gordon, 1992, 1996; Hatayama et al., 1996]. Additionally, evidence of significant internal wave generation in the archipelago points to strong tidal flows and stratification. One of the earliest reports of what was perhaps the surface expression of an internal wave packet in the Indonesian seas was in the Halmahera Sea in October 1860 by Wallace [1989, p. 542] :
We heard a dull roaring sound, like a heavy surf, behind us. In a short time the roar increased, and we saw a white line of foam coming on, which rapidly passed us without doing any harm, as our boat rose easily over the wave. At short intervals, ten or a dozen others overtook us with great rapidity, and then the sea became perfectly smooth, as it was before.
 Wallace felt that these must be “earthquake waves” (tsunamis) because he later learned that there had been an earthquake on Halmahera that day. They are, however, reminiscent of the observations of internal waves in the Andaman Sea, just north of Sumatra, by Osborne and Burch  and, whether triggered by tides or earthquake, they may have been caused by excitations to the subsurface density structure.
 We observed in a sun-glint image the surface expressions of three packets of internal waves that had propagated from the Ombai strait into the Banda Sea (Figure 8). The remarkable similarity to the SAR image (Figure 6b) suggests that this may be a common feature. In the sun-glint image the leading wave packet almost reaches SE Sulawesi and is relatively indistinct. The spacing between the leading waves of the two other ranges from 105 km at the western end of the curved wave fronts to 140 km at the eastern end. Assuming that the wave packets are generated with each semidiurnal tide, then these spacings would result from the wave packets propagating at 2.3–3.1 m/s.
 The surface expression of internal waves in the SAR and sun-glint imagery gives further evidence of the strength of the tidal energy in the strait of Ombai. Such waves require a well-stratified water column for tidal forcing to produce and propagate them. These structures are very similar to those described by Osborne and Burch .
Nilsson and Tildesley  found that for SAR imagery, apart from winds, the largest differences in backscatter seemed to be associated with different water masses, and these differences were often due to natural surfactants, common in higher productivity regions. Analyzing SAR images, they found that under appropriate conditions they could discriminate water masses that have different biological activity via these “darker” regions of lower backscatter. While only a single appropriate SAR image was available for the Strait of Ombai, we did observe some similar relationships between “low backscatter” and “frontal lines” and the corresponding SeaWiFS imagery.
 We have observed chlorophyll concentrations, indicative of phytoplankton blooms in the Ombai Strait. There are three likely mechanisms for such ocean color signatures. The first is flow separation at the entrance to the strait. Wolanski and Hamner  describe such a case and highlight the resulting topographically controlled fronts and their influence on the local biology (Figure 12). The second mechanism is tidal mixing. Molcard et al.  describe, in the Strait of Ombai, very strong diurnal plus semidiurnal tidal currents and, from a small set of CTD profiles, “step like features” in the thermocline that point toward significant tidal mixing. Local surface expressions of mixed water would be carried down stream with the mean flow. The final mechanism is cross-strait uplift of the subsurface layers. Molcard et al.  characterize the mean flow as vertically uniform in direction (toward the Savu Sea). However, the current velocity drops considerably beneath the upper 200 m layer. There should be a level of no motion near 1300 or 1400 m as this is the sill depth at the boundary between the Indian and Savu Seas, and we expect no communication between the basins below that depth [Molcard et al., 2001]. Given this, geostrophy would set up a situation on the northern end of the strait where the pycnocline would shoal (Figure 13). Additionally, should the mean flow bend significantly around Alor island as it entered the strait, “centrifugal upwelling” could potentially cause the shoaling of lower layers, even to the point of surfacing (John Hunter, UTAS Antarctic CRC, personal communication, 2001). The combination of these two “uplift” effects acting only in the north of the strait supports the fact we found the vast majority of high chlorophyll concentrations originating off the southern coast of Alor Island. However, without considerable in situ data and higher frequency remotely sensed data, to quantify the relative contributions of the potential flow related driving mechanisms would be purely speculative.
 The availability of cold, nutrient rich water, given a set amount of energy available for mixing or uplift, will depend mostly on how close macronutrients are to the surface layers. As such, we expect that the expressions of cold water and phytoplankton blooms seen in the remotely sensed imagery will be linked to thermocline depth. This linkage leads to a likely mechanism for variability on longer timescales as well. We would expect that bloom events would be stronger and more prevalent during the SEM when it has been shown that both the mean flow through the strait is strongest and the thermocline depth is shallowest (Figure 5). Further, ideas that the bloom features may somehow be tied to runoff are unlikely as the events are mainly present during the driest period of the year.
 During the strong La Nina in November 1998, there is less evidence of surface expressed cold water compared to a year earlier during the strong El Nino. Figures 4 and 5 shed some light on possible mechanisms behind this interannual variability of mixing in the Strait of Ombai. During the El Nino of 1997–1998, winds in November (Figure 4) were anomalously from the east. This pattern, more typical of the SEM, leads to lower sea levels and shallower thermocline depths. The data in Figure 5 bear this out as there are clear correlations between satellite derived sea level near the entrance and downstream of the Strait of Ombai and the Southern Oscillation Index. Tide gauge data from the port of Darwin (∼750 km to the southeast of the strait) are provided as an in situ reference and generally agree with the seasonal and interannual pattern of the remotely sensed sea level data.
 There is some controversy whether or not Kelvin waves, forced by wind bursts in the Indian Ocean, reach the strait of Ombai [Chong et al., 2000; Molcard et al., 2001], but if true, they could play a role as well in controlling these mixing events. Coastally trapped downwelling waves could both slow (or reverse) the mean flow in the strait as well as depressing the thermocline.
 We conclude that there are persistent flow-induced upwelling and frontal features in the Strait of Ombai. There is subjective evidence of extremely strong tides in the strait, and therefore we suggest that at least part of the energy driving upwelling in the strait be derived from this source. It is likely that the mean flow associated with the ITF plays a role as well. The presence and strength of surface expressions of cold, nutrient rich waters are directly tied to periods of shallow thermocline depth. If thermocline depth is important to the development of upwelling features in the Ombai Strait, then monsoons and ENSO will have an effect in that they force seasonal and interannual variability in thermocline depth.
 These upwelling events and corresponding frontal features in the Strait of Ombai coincide with large phytoplankton blooms and significantly effect the concentration and distribution of chlorophyll in the Savu Sea. The Strait is one of the few major conduits for the ITF, and vertical mixing in the throughflow region has been shown to be a critical driving force in the heat and freshwater balance between the Pacific and Indian Oceans [Gordon, 1986; Gordon et al., 1994; Godfrey, 1996; Gordon and Fine, 1996]. As such, these upwelling events in the Strait not only play a part in the local productivity and carbon cycle but may represent a mechanism that impacts regional climate variability as well.
 We would like to thank George Cresswell for his ideas on the oceanography of the Indonesian seas and for bringing the Alfred Wallace observation to our attention. We also wish to thank Paul Tildesley for SAR post-processing and providing the AVHRR “sun-glint” image and John Hunter, Chris Rathbone, Richard Matear, David Griffin, Jim Mansbridge, and Neil White for valuable support and discussions. We would also like to offer thanks for our NASA agency support from both our NRA 97-MTPE-01 and our SAR RADARSAT-99-OES-10 projects. The first author wishes to thank Nan Bray for the opportunity to visit the Strait of Ombai.