Introduction Monsoon Systems Monsoon Variability Surface Layer Response Phytoplankton Response Thermohaline Circulation and Water Masses Fisheries of the Arabian Sea
This review will begin with some basic aspects of the monsoons and their climatic variability. The basic physics of the surface layers of the Arabian Sea and their relation to primary production are then treated. This is followed by a discussion of the intermediate-depth low-oxygen layer and its relationship to the circulation and biogeochemical balances.
The Asian monsoon, which can be broken into an eastern and Indian branch, consists of a build-up of high pressure over the Asian highlands in the northern-hemisphere winter. This produces the northeast Indian monsoon (NE monsoon) with near-surface winds which flow out of the northeast over the Indian subcontinent and out across the Arabian Sea. The other branch brings flow off the Chinese mainland out over the island archipelagos of the Philippines and Indonesia. Both branches cross the equator and push the intertropical convergence into the southern hemisphere. The situation reverses with the build-up of high pressure south of the equator in the Indian Ocean in April and lowering of surface pressure over the Asian highlands as summer approaches. This leads to a reversal in the winds proceeding first along East Africa and then across the center of the Arabian Sea and onto the Indian Subcontinent. The flow across the Arabian sea within this southwest monsoon takes the form of an intense low-level jet, known as the Findlater jet after Findlater (1969). As depicted in Findlater's (1969) work the jet typically bifurcates before it reaches the Indian coast into two branches which are denoted as the Somali and Split jet. The placement and time dependence of this split is important in setting the bounds of the wind stress curl induced open ocean upwelling which is a dominant feature of the Arabian Sea monsoon response. The NE and SW monsoons are separated by transition periods or monsoon lulls during which the wind speeds drop and the sea surface heats.
A major factor controlling the air-sea interaction during the monsoon and the structure of the near-surface ocean responses to this cycle is the alternate temperature rise to nearly 30 deg. C during the lulls and the cooling which occurs through air-sea heat exchange during the high-wind periods. The phenomenon known as Arabian Sea cooling (Duing and Leetmaa, 1980, McCreary and Kundu, 1989) is most intense in the SW monsoon when the northern and western Arabian Sea are dominated by upwelling. This involves coastal upwelling along both the Arabian and Somali coasts (Currie et al., 1973; Schott, 1983; Currie, 1992), and open-ocean upwelling in the mid-basin associated with the wind-stress maximum under the Findlater jet (Smith and Bottero, 1977; Swallow, 1984; Bauer et al., 1991). The upwelling and surface heat exchange creates thermal anomalies in the western Arabian Sea which vary from year to year (Brown and Evans, 1981). While it has been suggested that these in turn modify the monsoon rainfall over India (Shulka, 1975; Raman et al., 1992), this effect is not seen in all models (Washington et al., 1977). During the SW monsoon, large-scale upwelling also occurs off southwest India (Banse, 1968; Shetye, 1984).
The onset of the SW monsoon is variable but usually takes place between mid-May and mid-June. Southern regions have a larger range of onset dates than more northerly ones (cf. Krishnamurti, 1987). Onset in Bombay between 1977-1980 occurred within 7 days of June 15th (Krishnamurti, 1987). Onset over the northern Arabian Sea is essentially a tropical storm, the "onset vortex", which varies in intensity, but is strong enough to be a danger to shipping. The onset in Bombay is a good indication of the presence of the onset vortex, although Bombay is not necessarily affected in any given year. The monsoon is interrupted by lulls in rain and wind known as monsoon breaks. An example of these cycles between active monsoon periods and breaks in two normal monsoon years (see longer term variability below) is given in Fig. 2. It is not clear to what degree breaks involve wind reduction over the Arabian Sea. The monsoon breaks seem to be related to the passage of individual convective complexes (cloud bands) (Webster, 1987). At a slightly longer period there are 30-50 day oscillations in the monsoon (Krishnamurti et al., 1985a,b) which correspond to a global-scale oscillation in the tropics.
Interannual variations in monsoon strength appear from the scale of the bi-annual oscillation in the tropical atmosphere (Cadet and Diehl, 1984; Dube et al., 1990) to longer-period variations related to cycles ranging from ENSO (El Nino/Southern Oscillation) to glacial-interglacial time scales. The origin of the Southern Oscillation index stems from Walker's (1910, 1924) work on monsoon prediction. The result was a system more in tune to a Pacific basin phenomenon which became a major focus in oceanography in the 1970's (Bjerknes, 1969; Wyrtki; 1975). The ENSO phenomenon has seen a wealth of quantification over the last two decades. The connection between the ENSO and monsoons over the Arabian Sea, however, remains unclear. Barnett (1985) and Webster and Yang (1992) are but two of several attempts to consider the interaction between monsoon and ENSO. The latter authors suggest a selective coupling between the onset phase of the Asian summer monsoons and global-scale summer circulation patterns with weak global trade winds being tied to weak Asian monsoons. Although the exact nature of the latter coupling is uncertain in the sense of whether or not ENSO events start in the Indian sector, it is clear that Indian rainfall is less following ENSO events and heavy during anti-ENSO periods. A good example is the 1987 monsoon which had almost no rain over portions of India (Krishnamurti et al., 1989). It is important to note that much of the change in this monsoon involved low winds south of the equator and along the Somali coast. Winds over the central Arabian Sea were near climatological values (Bauer et al., 1991). There is little discernible trend in monsoon rainfall records over the last hundred years, but there is considerable interdecadal variation. The 1900's, 1920's, and 1960's all had lower rainfall, while the 1910's, 1930's, 1940's, 1950's, and 1970's saw higher rainfall (Fig. 3).
Finally, paleoceanographers have been able to use sediment records from the northern Arabian Sea to consider the longer-term variations tied to ice age cycles with their associated solar radiation forcing, and longer-term events such as the rise of the Tibet-Himalayas massif (Prell, 1984; Kutzbach, 1987; Prell and Kutzbach, 1987). Sediment records and modelling (Luther et al., 1990; Prell et al., 1990) suggest weaker monsoons at the height of the last glaciation (18 Kyr BP) and stronger monsoons at the interglacial maximum (9 Kyr BP).
Extremes of near-surface thermal stratification at the heights of the two monsoons are shown in Fig. 5. Note the overall shift in the slope of the thermocline; the large scale gradients of the thermocline are changed throughout the depth of sampling. Currents in the NE monsoon case are to the west (into the page). The flow reverses in the SW monsoon such that the flow is eastward and slightly stronger than in the northern winter. The SW monsoon section reveals a major mid-ocean upwelling zone separated from the coastal upwelling zone by a reversed flow. Smith and Bottero (1977) first suggested a separation of these two upwelling domains. The reversal can be interpreted as the westward end of a permanent anticyclonic circulation over the Murray Ridge (Quraishee, 1984, Bauer et al., 1991). Connections between these two upwelling systems in three dimensions remain to be fully established. This has ecological implications, particularly in respect to the coupling between coastal upwelling and mid-ocean conditions. This is one of the many cases where the Arabian Sea is a model in which to understand interactions between physics and biology in other flow-coupled pelagic ecosystems.
Mixed layers are deepest in the NE monsoon in the northwestern area (average about 125 m, Banse 1984; see also Fig. 5). During this season, they shoal to less than 20 m around 5 deg. N (Bauer et al., 1991). Conditions during the SW monsoon in 1987 show the deepest mixed layers (~60 m) around 15 deg. N with a minimum in mixed layer depth under the Findlater jet. The shallow mixed layers under the maximum winds are coincident with the upwelled dome of fluid in the SW monsoon (Fig. 5). Deepening of the mixed-layer by wind stirring is curtailed by the upwelling of the thermocline at rates of order 10-5 m s-1 (Bauer et al., 1991). The deep mixed layers to the south of the wind maximum can only be explained through advection of dense mixed layers southward in the Ekman layer and the subsequent downward motion of the thermocline associated with Ekman pumping, with maximum vertical velocities of -2.5 X 10-5 m s-1 (Bauer et al., 1991). The simple two-dimensional mixed-layer model of Bauer et al. (1991) suggests that the upwelled structure reaches an equilibrium in mixed-layer depth where the upward motion balances the turbulent entrainment rate. The deep mixed layers to the south continue to deepen in the simple model since turbulent entrainment and downwelling both act to increase mixed-layer depth.
The description based on single sections through the region in single years cannot give a reasonable picture of the fields in light of the evidence discussed above for interannual variations. A powerful tool for quantifying the representative nature of limited ship based measurements is remote sensing. A series of papers (Banse and McClain, 1986; Brock and McClain, 1992; Brock et al., 1991) describe phytoplankton blooms in the Arabian Sea from CZCS data. There are blooms in both monsoons in the northern extreme of the basin (Banse, 1987), presumably caused by mixing related to cooling over the Gulf of Oman and Murray Ridge. Brock and McClain (1992) suggest interannual variability, which might be driven by the biannual oscillations in the forcing as discussed above. The reliable record, however, is very short. This should be improved with the SeaWiFS deployment in the time frame of the proposed Arabian Sea efforts.
The physical and biological dynamics responsible for the oxygen minimum involves a complicated balance between the advection or mixing of oxygen into the layer and the removal of oxygen in the water column by respiration. Oxygen is added to the Arabian Sea by inflow of cold oxygen rich waters from the south up either the Somali or Indian coasts or by horizontal mixing between the Arabian Basin and the equatorial zone. Some oxygen is added through the introduction of warm, low oxygen waters from the marginal basins, the Red Sea and Arabian or Persian Gulf (Banse, unpubl.; Olson et al., 1993). Alternatively, oxygen can mix into the intermediate layers from the shallower or deeper adjacent waters (Swallow, 1984; Naqvi et al., 1992). The biological uptake of oxygen to balance these inputs depends ultimately on the primary production and how it is utilized. The available data suggests that the total phytoplankton production on an annual basis is not extraordinary (Olson et al., 1993) although the intense bloom and decay cycles may enhance the ratio of new to old production. A key issue in understanding the bloom and decay cycle involves better quantification of the grazing down of the phytoplankton stocks during the monsoons and the details of the trophic dynamics during the oligotrophic monsoon lulls.
There are many indications that the low-oxygen zone may be very variable and that it has varied considerably in the past. There are at least two studies that suggest the time scales for respiration versus oxygen input are very short compared to intermediate and deep water circulation times. Using anthropogenic chloroflurocarbon F-11 as an inert tracer, high-quality oxygen measurements and the productivity work quoted above, Olson et al. (1993) estimate that the time scale of oxygen renewal in the Arabian Sea is 23 years. A similar calculation based on the nitrogen cycle rather than that of oxygen gives time scales on the order of a year for the near surface portions of the low oxygen layer (Naqvi et al., 1992). Either estimate is very short relative to the time scales estimated for chlorofluorocarbons for the geographically analogous oxygen minima in the eastern North and South Atlantic Oceans (Doney and Bullister, 1992; Warner and Weiss, 1992). These time scales suggest that the oxygen minimum layer might be sensitive to decadal variations in the monsoonal forcing of surface productivity and circulation and to changes in the input from the marginal seas.