1. Introduction
The prediction of the intensity of the atmospheric monsoon circulation in the Indo–Pacific domain and the roles played by the underlying ocean is a challenging problem. This is mainly due to the infancy of real-time observing networks in the ocean and the atmosphere in this region and the subsequent shortness of the records. As a result, the computation of heat storage in the upper-ocean layers, an important parameter for climate studies, has been a difficult task. This is why the various efforts to study variability of the Indian monsoon on different timescales have concentrated primarily on surface parameters (Godfrey et al. 1995). Linkages in winds and sea level pressure between the Indian monsoon and global climate signals such as El Niño–Southern Oscillation (ENSO) have been established (Barnett 1983; 1984a,b; Meehl 1987; Kiladis and Diaz 1989; Chen and Yen 1994; Terray 1995; among others). Correlation between the Indian and Australasian monsoon winds and precipitation have also been presented (Shukla and Paolino 1983; Bhalme and Jadhav 1984; Cadet 1985; Mooley and Shukla 1987). Selective interaction and a 3–6 year link between the Asian monsoon and ENSO were studied by Webster and Yang (1992) and Yanai and Li (1994), respectively. While Yasunari (1990) found evidence of the Indian monsoon modulating ENSO, the modulation of the Indian monsoon by ENSO has also been reported (Joseph et al. 1994; Ju and Slingo 1995).
Recently Villwock and Latif (1994), by investigating the interannual variability of the sea surface temperature (SST), found ENSO signal to be the dominant signal in the Indian Ocean. Tourre and White (1995, TW hereafter) examined the interannual variability of the mean temperature anomalies and the heat storage in the upper 400 m of the Indian Ocean (HS) within the context of the global ocean, finding ENSO signal dominating both the surface and subsurface. Tourre and White found that for each variable the dominant mode of variability in the Indian Ocean is in phase with the peak El Niño phase off the west coast of South America, while the second mode leads the first by at least 6 months. This has been known for some time in the Pacific Ocean (e.g., Gill and Rasmusson 1983; White et al. 1985), but finding these two modes in SST and HS in the Indian Ocean corroborated results obtained by Meehl (1993) for selected regions only. Moreover, this indicates a slow evolution for the ENSO signal and suggests an eastward propagation along the equator similar to its better-known counterpart in the Pacific Ocean as already noted by Meehl (1987), Kiladis and van Loon (1988), and Yasunari (1990).
The relationship between the ENSO signal in upper-ocean temperature and the monsoon winds along the east coast of Africa and along the west coast of Indonesia has been examined (Kiladis and Diaz 1989; Meehl 1994) but is still poorly known. In this study, we follow the work of TW by examining in more detail the evolutionary character of the ENSO signal at the surface and subsurface of the Indo–Pacific domain in relation to the overlying wind stress. Particular emphasis is given to the eastward propagation of ENSO from the Indian Ocean into the maritime continent and on into the western Pacific Ocean. The phase relationships in our results are suggestive of particular physical mechanisms of the ocean–atmosphere system, but the latter remain to be substantiated through extensive use of dynamical models.
2. Observations and methods
In this study the evolution of ENSO in the Indian and western Pacific Oceans is investigated. To that effect monthly gridded anomalies of SST, HS, zonal surface wind stress (ZSWS), and meridional surface wind stress (MSWS) are constructed over the Indo–western Pacific domain only, that is, 40°S–50°N, 40°E–180° and for 13 years (1979–91). The monthly SST and HS anomalies are calculated from the available subsurface temperature profiles collected in those regions. The geographical distribution of these temperature profiles for a representative year (i.e., 1985) together with a time sequence of the number of observations collected per month from 1979 to 1991 are given in Fig. 1. Approximately equivalent numbers of observations (i.e., 900–2000 profiles per month) were collected during the first ten years of the record, often varying considerably from month to month, with about 200–800 profiles per month collected during the last two years of the record. This reduction in sampling density toward the end of the record occurs because of the 2–3 year delay in the submission of temperature profiles to the National Oceanographic Data Center (NODC), from whence these observations come. Spatial sampling is uneven as well, with the northwest Pacific and northwest Indian Oceans displaying higher sampling densities.
The largest portion of the temperature profiles collected for this 13-yr period (i.e., 90%) is composed of expendable bathythermograph profiles. XBT probes are deployed principally from the Ships-of-Opportunity (SOO) program. The SOO are vessels instrumented to measure upper-ocean temperatures along specific track lines and are coordinated by the Integrated Global Ocean Services System (IGOSS 1991).
Temperature profiles resulting from the global observing network have been subjected to rigorous quality control procedures at the Joint Environmental Data Analysis (JEDA) Center of the Scripps Institution of Oceanography. The procedures are defined under the guidelines of the Global Temperature–Salinity Project (GTSP) and are described by White et al. (1988). One of the steps in quality control procedures requires the computation of the anomalous temperatures about the mean annual cycle of SST and HS at each observation location. These anomalous observations are then used to construct 2° latitude by 5° longitude monthly maps using objective analysis techniques (Gandin 1963). The decorrelation scales used in objective mapping are configured to grid for the least-dominant scales of variability, that is, 5° of latitude, 10° of longitude, 3 months, and a noise-to-signal ratio of 1.0 (White 1995). The map of normalized errors (not shown) display the smallest errors (i.e., less than 0.75) along repeated transects through the western Pacific and the eastern Indian Oceans. Larger errors (0.9 of the signal standard deviation) occur where sampling density is small, like around 20°S and between 70° and 100°E in the southern Indian Ocean. The reference period for the climatology is chosen to be ten years from 1979–88 because of the relative uniformity in the distribution of the temperature profiles collected over this period (see lower panel of Fig. 1). However, anomalies are computed for the 13 years from 1979 to 1991 using this base period of 1979–88. As such, anomalies do not sum to zero over this 13-yr period.
The pseudo wind stress was obtained from the objective analysis of ships and buoys data developed at The Florida State University and described by Legler (1991).
To highlight the leading space–time features of the ENSO signal in the interior of the Indian–Pacific regions, already evidenced by TW, extended empirical function (EEOF) analysis is used, following Graham et al. (1987). EEOF analysis technique enables one to isolate the evolutionary character of a dominant signal (Preisendorfer 1988). In essence, the familiar EOF maps for each mode are replaced by a sequence of maps (or windows) lagged by a constant time period. The amplitude function associated with each mode modulates the central window (lag 0) of the time sequence. Additional information about ranges of phase velocities and phasing between variables is obtained from time–longitude diagrams.
The 13-yr period chosen for this study contains the end of a weak El Niño (1979), the beginning of a long lasting El Niño (1991), and extends over two full ENSO cycles. An extreme cold event also occurred in the Pacific Ocean during 1988–89. The record length is limited by the subsurface sampling and the paucity of associated data prior to the 1970s. As a consequence, the authors acknowledge the limitations in the interpretation of the results including the levels of statistical significance. For example when the data are being band passed (2–7-yr recursive filter) the root-mean-squares (rms) in section 3b are reduced. They represent 65% and 50% of the rms for the unfiltered SST and HS data in TW’s paper. This is because the lowest frequency signals are not well resolved. The problem is compounded by the fact that “ENSO cycle-related” variability displays, at least in the Pacific Ocean, spatial signatures similar to “decade-to-century” scales variability (Zhang et al. 1997). Confidence in our results is, however, obtained from the strong in-phase relationship between SST, HS, ZSWS, and MSWS as shown later. Moreover when a similar analysis (EEOF with the same band passing) is performed with a newly developed dataset (1955–92) but for SST and HS only, the principal components are almost identical during the overlapping period (White and Cayan 1997, manuscript submitted to J. Phys. Oceanogr.). This suggests that the ENSO timescale signal in this paper was properly extracted.
3. Climatologies of the Indo–Pacific regions
a. Annual mean
The Indo–Pacific distributions of the annual mean SST, HS, and surface wind stress for the base period 1979–88 are displayed in the upper panels of Figs. 2, 3, and 4. Major features of the mean SST include the “Warm Pool” (Wyrtki 1989) where SST is above 28°C in the central, eastern Indian Ocean and western Pacific Ocean (Fig. 2, top). The 28°C isotherm expands westward along the equator until 50°E with no equatorial upwelling structures farther west. The maximum meridional SST gradient is found around 20°S between 80° and 100°E in the Indian Ocean and in the vicinity of the Kuroshio Current in the northwest Pacific Ocean between 35° and 45°N. High temperature can also be found in the western South Pacific Ocean under the climatological South Pacific convergence zone (SPCZ). Major features of the mean HS (Fig. 3, top) include representations of the shallow tropical gyres in the Indian and Pacific Oceans, as HS is proportional to the relative geopotential anomaly and steric height of the sea surface. In the Indian Ocean regions of maximum HS also correspond to the location of the atmospheric Arabian and Mascarean highs. In the south Indian Ocean near 8°S and between 50° and 75°E, a well-developed trough is to be associated with the South Equatorial Countercurrent (SECC) on its equatorward side (Godfrey 1989). Around 25°S and between 60° and 100°E is the topography that yields to the South Equatorial Current (SEC). The topography associated with the North Equatorial Current (NEC) in the western Pacific Ocean is clearly seen.
Major features of the mean surface pseudo–wind stress (Fig. 4, top) include maximum values around 20°S and between 75° and 95°E (larger than 75 m2 s−2) equatorward of the Mascarean anticyclone. The remaining southwesterly branch along the Horn of Africa and over the Arabian Sea reflects, after yearly averaging, the dominant summer monsoon signal in that region. Cores of trade winds are found in the western Pacific Ocean at 15°N and 20°S. Minimum values lie along the equator in both oceans over the warm pool where there is maximum confluence (and convergence) of the Indo–Pacific trade winds system.
b. Root-mean-square of filtered (ENSO signal) anomalies
It has been shown by Tanimoto et al. (1993) that empirical orthogonal function analysis (EOF) in the Pacific Ocean is incapable of separating ENSO signal from other frequency signals. Therefore, prior to the analyses, the datasets are being band passed using a 2–7-yr recursive filter (Kaylor 1977) to retain 3–6-yr features. The filter is applied by setting the half-power cutoff at frequencies corresponding to 2-yr and 7-yr periods. The length of the filter is altered at the ends of the time series so the record lengths of filtered and unfiltered anomalies are the same.
The distributions of the rms in the Indo–Pacific domain for the ENSO timescales SST, HS, and zonal and meridional surface pseudo wind stresses (ZSWS and MSWS respectively) anomalies are displayed in the lower panels of Figs. 2, 3 (SST and HS) and in the middle and lower panels of Fig. 4 (ZSWS and MSWS). It is worth noticing that the band passing on a relatively short 13-yr period reduces the amplitudes of the rms (when compared with rms from TW) as expected and already explained in section 2. Rms maxima occur where mean gradients are large. Rms maxima for SST and HS, on ENSO timescales, occur where rms maxima for the observed scale are found (see Figs. 2a and 2b, bottom in White 1994). This implies that ENSO timescale variability in SST and HS also occur in response to horizontal displacements in major frontal zones (associated with the mean general circulation) along the east African and Australian coastlines. Major features of the rms for SST (Fig. 2, bottom) include maxima in the western Indian Ocean where Ekman transport, Somali Current, and coastal upwelling dominate. Along the west Australian coastlines, large values are found in the vicinity of the Leeuwin Current. Minima are located in the warm pool region. The largest values in the northwestern Pacific Ocean are found where the largest gradients of net heat balance exist, that is, the northwestern boundaries of the Kuroshio Current. Major features of the rms for HS (Fig. 3, bottom) include the largest values on the eastern side of the warm pool in the western Pacific Ocean. Maxima are found in the northwestern Pacific, zonally distributed where the zonal Kuroshio Extension is found. The highest rms values in the Indian Ocean, a third of the highest values in the west Pacific Ocean, are found along the Somalia coastline and western Arabian Sea where intense upwelling occur during the boreal summer. High values are also found in the southeastern Indian Ocean and the Timor Sea. Major features of the rms for ZSWS and MSWS (Fig. 4, middle and bottom) in the Indian Ocean include maxima in the vicinity of the core of the southeasterly associated with the Mascarean high. High values of MSWS are found along the east African coastline depicting ENSO variability there associated with the monsoon variability. Maxima are also found in the SH western Pacific Ocean in the vicinity of the SPCZ.
4. Space–time evolution of the Indo–Pacific ENSO signals
a. Extended empirical orthogonal function analyses
When rotated analysis (Varimax criterion after Kaiser 1958) is performed for SST and HS ENSO timescale anomalies in the interior of the Indo–Pacific domain, two leading modes for each field, which represent 47% and 41% of the total variances respectively, emerge (not shown). Similar results are obtained when the EOFs are not rotated, which corroborates results from TW. Moreover, for each field, the maxima of the amplitude functions are lagged by approximately 6 to 12 months. Taken together these lagged modes indicate a slow eastward propagation of the ENSO signal. This corroborates TW results when they analyzed the 13-month low-passed data for the Indian Ocean only.
Since EOF and rotated EOF yield similar results, further insight about the space–time evolution of the ENSO signal, for SST, HS, ZSWS, and MSWS anomalies is gained by using the EEOF method from the correlation matrix. For each of the analyzed variables, EEOF yields to the first two modes in quadrature, which represent ENSO timescale variance.
For SST and HS, the first modes explain similar amounts of variance (34% for SST and 38% for HS). The same apply to ZSWS and MSWS (26% for the ZSWS and 19% for the MSWS). The amplitude functions for the first modes are displayed in Fig. 5. They show all four fields varying in phase from 1981 to 1989 (2 years on each end of the original time series having been truncated, which corresponds to the length of the lagged sequence for the full cycle). Peak amplitudes occur during the mature phase of El Niño in the southeast Pacific (i.e., winters–springs of 1982–83 and 1986–87). The timescale of the ENSO cycle is similar to the one obtained for a 37-year record and for the Pacific Ocean (White and Cayan 1997, manuscript submitted to J. Phys. Oceanogr.).
The spatial evolution of the ENSO signal in these four fields is clarified when the lagged sequences of the EEOF loadings are examined. Out of a set of 17 maps every 3-month, time sequences of five regularly lagged maps (12-month interval or lag 3) that represent a full ENSO cycle are displayed in Fig. 6 (SST and HS) and Fig. 7 (ZSWS and MSWS). In these figures and for each field the evolution of the dominant spatial patterns, that is, the ENSO signal, is apparent. Windows are displayed every 12 months (or lag numbers multiplied by 3). The maps at the central location (lag 0) represent approximately the spatial distribution during the phase of peak positive amplitude that occurred almost at the same time as the mature phase of El Niño occurred in the southeast Pacific. In those maps SST and HS display both positive anomalies in the central equatorial Indian Ocean, while negative anomalies of both ZSWS and MSWS are found in the central Indian Ocean north of the equator. This distribution of the pseudo wind stress leads to maximum northeasterly anomalies (against the southwesterly mean flow, see Fig. 4, top). These anomalies persist for at least 6 months and yield to weaker summer monsoons associated with ENSO (summers of 1982 and 1987). Rainfall deficits over India (Shukla and Paolino 1983) and Bangladesh (Ahmed and Karmakar 1993) follow.
From the lagged sequence of the SST loadings of the first EEOF (34% of the total variance) displayed in the left column of Fig. 6, we see in the western Pacific an equatorward propagation of the positive (negative) loadings starting at lag 0 (lag −8 or year −2). This also applies south of Taiwan where there is an equatorward mean flow from the Mindanao eddy. In the Indian Ocean positive loadings first appear in the western tropical regions (lag −8 or year −2) and then propagate eastward in the direction of the maritime continent. Maximum values of the loadings that appear at lag −4 are found in the central Indian Ocean at 10°S (lag 0). Eastward propagation of the SST ENSO signal is well pronounced in the vicinity of the SECC around 10°S and continues in the southeast Indian Ocean along western Sumatra and Java islands to finally enter the Timor Sea (lags 4 and 8 or years +1 and +2, respectively). At lag 8 a merging occurs, at the equator in the western Pacific Ocean, between the Indian Ocean and Pacific Ocean ENSO signals.
The lagged sequence for the HS loadings of the first EEOF (38% of the total variance) is displayed in the right column of Fig. 6. Around 25°N and 25°S, positive loadings in the western Pacific Ocean seem to be associated with incident ENSO Rossby waves identified by White and Cayan (1997, manuscript submitted to J. Phys. Oceanogr.). After reflection along the western boundary the ENSO signal appears to be transmitted equatorward south of Taiwan and penetrates the tropical regions (lags 4 and 8 or years +1 and +2 respectively). The slow equatorward propagation of the waves along the western boundary and their interactions with the western boundary currents are yet to be determined. In the Indian Ocean positive loadings appear first in the Arabian Sea (lag −8 or year −2) and then propagate slowly eastward in the equatorial band with maximum loadings around 10°–15°S. The eastward propagation continues into the Timor Sea and the Banda Sea (lags 4 and 8 or years +1 and +2 respectively). Positive loadings from both the eastern Indian Ocean and the western Pacific Ocean merge during lag 4 (year +1) at the equator in the western Pacific Ocean. This is the time when negative loadings start to appear in the Arabian Sea. At lag 8 (year +2) maximum positive loadings are found around the date line.
Lagged sequence for the ZSWS loadings of the first EEOF (26% of the total variance) is displayed in the left column of Fig. 7. Positive loadings (eastward anomalies for peak values of the amplitude function) are found over the maritime continent along the equator, while negative loadings (westward anomalies) are found along the equator in the western Indian Ocean (lag −8 or year −2). This is the time when the western equatorial Pacific Ocean is warmer and warming starts to occur in the western Indian Ocean. This pattern yields to maximum diffluence (divergence) in the central Indian Ocean. The ZSWS anomalies oriented in a northwest–southeast direction in the 10°N–10°S regions propagate slowly eastward. The origin of the signal, within the domain under study, appears to be where the core of the southeasterly are usually found (lags −4 and −8 or years −1 and −2 respectively). During the fully developed ENSO warm phase (lag 0) northerly anomalies are present in the Arabian Sea and Bay of Bengal.
Lagged sequence for the MSWS loadings of the first EEOF (19% of the total variance) is displayed in the right column of Fig. 7. Negative loadings (northerly anomalies for peak values of the amplitude function) appear first in the southwestern Indian Ocean in the Seychelles Islands region north of Madagascar (lag −8 or year −2). At lag −4 (year −1) the negative loadings have spread northeastward and are found along the Horn of Africa and the Arabian Sea. This is approximately the time when warm SST and HS ENSO anomalies are found in the same regions (Fig. 6). Subsequently the anomalies, oriented in a northwest–southeast direction, propagate slowly eastward in the tropical Indian Ocean. Maximum negative values found to the south (10°S) enter the Timor Sea at lag 4 (year +1).
When the ZSWS and the MSWS sequences of anomalies are taken together, northeasterly anomalies are present over the northern Indian Ocean during the warm phase of the ENSO signal (Fig. 7, lag 0). This anomalous circulation, as already mentioned, will weaken the intensity of the summer monsoon.
In summary, it has been shown in this section, that by using EEOF analyses for SST, HS, ZSWS, and MSWS, a slow eastward propagation in all variables is evidenced from the first EEOF modes. Anomalies that start in the west-southwest Indian Ocean are subsequently found in the maritime continent–warm pool regions (the Savu Sea, Timor Sea, and Banda Sea). Further insight about phasing between the four variables as well as the ranges for the phase velocities of the propagation is gained from time–longitude diagrams presented in the next section.
b. Time–longitude diagrams: The slow eastward propagation of the ENSO signal
Slow westward propagation of ENSO signal in HS (i.e., 10 cm s−1) has been observed throughout the tropical Pacific Ocean (White et al. 1985; Kessler 1990), while slow equatorward propagation of SST and HS anomalies along the western boundary of the tropical Pacific Ocean has been observed by White (1994). Along the western boundary of the Pacific Ocean, the incidence of the ENSO signal has been shown to instigate ENSO boundary waves along the thermocline (White and Tai 1992). Subsequently evidence of slow eastward propagation of SST and HS anomalies along the equator and eastern boundaries of the Indian Ocean have been shown by White (1994). Propagation speeds of the ENSO boundary waves range from 5 to 20 cm s−1. ENSO signals in SST have also been observed propagating in the direction of the background mean circulation. This suggests that SST anomalies associated with the ENSO timescale are advected by the mean geostrophic flow. A similar conclusion is reached by Latif and Barnett (1994) when they discussed how the mean horizontal currents contribute to wave propagation on decadal timescale. When Chelton and Davis (1982) examined the poleward propagation of the ENSO signal along the west coast of Central and North America in sea level height, they found an averaged velocity of 40 cm s−1. This is close to the findings by Kessler (1990) who examined the poleward propagating ENSO signal along the eastern boundary of the North Pacific Ocean. These velocities are an order of magnitude slower than the velocities expected from the Kelvin carrier waves (McCreary 1976; Hurlburt et al. 1976; Pares-Sierra and O’Brien 1989).
In order to assess the phasing and the velocity of the Indo–Pacific ENSO signal, averaged time–longitude diagrams of ZSWS anomalies (10°N, 10°S) and time–longitude diagrams of SST (10°S) are constructed (Fig. 8). The anomalies are band passed as before.
Eastward propagation of ENSO signal in ZSWS and SST is evidenced in the time–longitude diagrams (Fig. 8, top and middle). In the Indian Ocean, during the warm phases of ENSO (1982–83 and 1986–87) easterly anomalies are present with maxima around 90°E. The other maxima are found in the vicinity of the date line. The averaged slopes during the 13-yr period of the eastward propagating ENSO signal in ZSWS and SST anomalies yield to a mean phase velocity of 15–25 cm s−1.
Eastward propagation at 10°S of the ENSO signal in HS anomalies is also evidenced until 140°E (Fig. 8, bottom). The averaged slopes during the 13-yr period of the eastward propagating HS anomalies lead to a mean phase velocity of 20 cm s−1. Similar results were obtained by White (1994) for HS anomalies propagation along the equatorial Indian Ocean and the Indonesian archipelago (from approximately 50° to 120°E).
It is worth noticing that maxima are found in the Timor Sea around 120°E (end of 1982 and middle of 1985, 1987, 1989, and 1991). These are approximately the time when maxima in HS anomalies are found between 150° and 160°E in the western Pacific Ocean (White 1994). The ENSO signal were found to propagate equatorward along the western boundaries from both north and south tropical Pacific Ocean. The results presented in this paper not only bring further evidence of HS anomalies propagating into the Timor Sea but reconfirm the subsequent merging of ENSO signal (from both Indian and Pacific Oceans) into the warm pool and in the western Pacific domain (as evidenced by the EEOF analyses). The ENSO signal then continue to propagate eastward along the equatorial Pacific Ocean as already known.
When the time–longitude diagrams of ZSWS and SST anomalies are compared, it is found that ZSWS anomalies lag the SST anomalies by approximately 10 to 12 months (between 56°E and the date line), the SST and HS anomalies being more or less in phase between 56° and 140°E. Possible implication from the lagged relationship between ZSWS and SST anomalies are discussed in the next section.
5. Discussion and conclusions
In the Indo–Pacific regions, both oceanic and atmospheric variables (SST, HS, ZSWS, and MSWS) display ENSO timescale signals that propagate slowly eastward from the western Indian Ocean to the eastern Indian Ocean into the Timor Sea at speeds ranging from 15 to 25 cm s−1. This is similar to the magnitude of slow eastward phase propagation of ENSO-scale thermocline depth anomalies observed in the equatorial Pacific by White et al. (1985), Graham and White (1988, 1990), White et al. (1989), and Kessler and McPhaden (1995). The Indo–Pacific ENSO wave in the present study is not confined to the equator, as observed in these previous studies; rather, it has significant meridional extent, ranging from 20°S to 20°N.
The slow eastward propagation of the ENSO timescale signal does not resemble any known equatorial or tropical waves of the ocean and atmosphere other than those of the equatorial coupled system (Hirst 1988a,b; Neelin 1991; and others). However, this equatorial coupling is incapable of explaining the slow eastward propagation in the Tropics off the equator. A schematic diagram is given in Fig. 9, which summarizes the eastward propagation associated with ENSO cycle and the interaction between oceanic and atmospheric variables therein. In this figure, ENSO timescale signals begin in the western Indian Ocean where maximum rms variability (somewhat reduced due to band passing) is found in all variables. Zonal winds are directed from regions of cold SST anomalies (
The source for the slow propagation of a coupled ocean–atmosphere ENSO signal in the Indo–Pacific domain appears to be wind-induced upwelling along the African coastline, which generates SST and HS anomalies of the correct sign. The subsequent eastward propagation could be due to a convective adjustment between cold (warm) SST anomalies and high (low) sea level pressure (SLP) anomalies in the overlying atmosphere. The subsequent anomalous SLP patterns yield ageostrophic ZSWS anomalies, which in turn cause the initial SST anomaly to be displaced eastward off the coast of Africa. The role that wind-driven pressure gradient and Ekman flows have in altering the ENSO-scale SST anomaly needs to be better understood before the physics of this coupled eastward propagation is explained.
One of the principal issues is the role ENSO has upon the Indian monsoon. Ogallo (1988), Kiladis and Diaz (1989), Hastenrath et al. (1993), and Meehl (1994) demonstrated that surface SST anomalies in the Indian Ocean modulate the seasonal rainfall over East Africa. Dash (1992) demonstrated this same influence upon monsoon rainfall over India. Consequently, ENSO-scale SST and wind stress anomalies observed in this study will contribute to the modulation of rainfall variability in the neighboring continental domains. An example of this is given during the summers of 1983 and 1987, when northeasterly wind anomalies diminish water vapor advection over the Indian continent, reducing rainfall (Sikka 1980; Shukla 1987). While ENSO signal modulates the Indian monsoon, an important question is whether the land–atmosphere–ocean processes that lead to the monsoon variability on the annual cycle also play a role on the 3–6 yr timescale of ENSO? And do these land–atmosphere–ocean processes contribute to the slow eastward propagation of the ENSO signal in the Indo–Pacific domain?
Acknowledgments
Our thanks extend to Arthur (Ted) Walker, who as programmer conducted the analyses presented in this study. Support is provided by the National Oceanographic Data Center, the National Ocean Service, and the Office of Global Programs of NOAA (NOAA NA 90AA-D-AC416) in concert with the Tropical Ocean Global Atmosphere Program. Support is also provided by the National Science Foundation (OCE-9196889) in concert with the World Ocean Circulation Experiment. Support accrues from the LDEO of Columbia University (Yves Tourre) and the SIO of University of California, San Diego (Warren White and Yves Tourre) as well.
REFERENCES
Ahmed, R., and S. Karmakar, 1993: Arrival and withdrawal dates of the summer monsoon in Bangladesh. Int. J. Climatol.,13, 727–740.
Barnett, T. P., 1983: Interaction of the monsoon and Pacific trade wind system at interannual time scales. Part 1: The equatorial zone. Mon. Wea. Rev.,111, 756–773.
——, 1984a: Interaction of the monsoon and Pacific trade wind system at the interannual time scales. Part II: The tropical band. Mon. Wea. Rev.,112, 2380–2387.
——, 1984b: Interaction of the monsoon and Pacific trade wind system at interannual time scales. Part III: A partial anatomy of the Southern Oscillation. Mon. Wea. Rev.,112, 2388–2400.
Bhalme, H. N., and S. K. Jadhav, 1984: The Southern Oscillation and its relation to the monsoon rainfall. J. Climatol.,4, 509–520.
Cadet, D. L., 1985: The Southern Oscillation over the Indian Ocean. J. Climatol.,5, 189–212.
Chelton, D. B., and R. E. Davis, 1982: Monthly mean sea level variability along the west coast of North America. J. Phys. Oceanogr.,12, 757–784.
Chen, T. C., and M. C. Yen, 1994: Interannual variation of the Indian monsoon simulated with NCAR community model: Effect of tropical Pacific SST. J Climate,7, 1403–1415.
Dash, S. K., 1992: Some results of monsoon numerical experimentation with UGAMP climate model. Rep. on Simulation of Interannual and Intraseasonal Monsoon Variability WCRP-68, WMO/TD-470, 2.7–2.12. [Available from WMO, Publications Sales Unit, Case Postale 2300, CH-1211 Geneva 2, Switzerland.].
Gandin, L., 1963: Objective analysis of meteorological fields. Gidro-meteorologicheskoe, Leningrad, USSR, 242 pp.
Gill, A. E., and E. Rasmusson, 1983: The 1982–83 climate anomaly in the equatorial Pacific. Nature,306, 229–234.
Godfrey, J. S., 1989: A Sverdrup model of the depth-integrated flow for the World Ocean, allowing for island circulation. Geophys. Astrophys. Fluid Dyn.,45, 89–112.
——, A. Alexiou, A. AG. Ilahude, D. M. Legler, M. E. Luther, J. P. Mc Creary Jr., G. A. Meyers, K. Mizumo, R. R. Rao, S. R. Shetye, J. H. Toole and S. Wacongne, 1995: The role of the Indian Ocean in the global climate system. Recommendations Regarding the GOOS and OOSDP, 89 pp. [Available from Oceanography Dept., Texas A&M University, College Station, TX 77843.].
Graham, N. E., and W. B. White, 1988: The El-Niño cycle: A natural oscillator of the Pacific Ocean–atmosphere system. Science,240, 1293–1302.
—— and ——, 1990: The role of the western boundary in the ENSO cycle: Experiments with coupled models. J. Phys. Oceanogr.,20, 1935–48.
——, J. Michaelsen, and T. P. Barnett, 1987: An investigation of the El Nino–Southern Oscillation cycle with statistical models. Predictor field characteristics. J. Geophys. Res.,92, 14 251–14 270.
Hastenrath, S., A. Nicklis, and L. Greischer, 1993: Atmospheric–hydrospheric mechanisms of climate anomalies in the western equatorial Indian Ocean. J. Geophys. Res.,98, 20 219–20 235.
Hirst, A. C., 1988a: Slow instabilities in tropical ocean basin-global atmospheric models. J. Atmos. Sci.,45, 830–852.
——, 1988b: Unstable and damped equatorial modes in simple coupled ocean–atmosphere models. J. Atmos. Sci.,45, 606–630.
Hurlburt, H. J., J. Kindle, and J. J. O’Brien, 1976: A numerical simulation of the onset of El Niño. J. Phys. Oceanogr.,6, 621–631.
IGOSS, 1991: Rep. of the Sixth Session of IGOSS, 86 pp. [Available from WMO, CP 2300, 1211 Geneva, Switzerland.].
Joseph, P. V., J. K. Eischeid, and R. J. Pyle, 1994: Interannual variability of the onset of the Indian summer monsoon and its association with atmospheric features, El Niño, and sea surface temperature anomalies. J. Climate,7, 81–105.
Ju, J., and J. M. Slingo, 1995: The Asia summer monsoon and ENSO. Quart. J. Roy. Meteor. Soc.,121, 1133–1168.
Kaiser, H. F., 1958: The varimax criterion for analytic rotation in factor analysis. Psychometrika,23, 187–200.
Kaylor, R. E., 1977: Filtering and decimation of digital time series. Tech. Note BN 850, 44 pp. [Available from Institute for Physical Science and Technology, University of Maryland at College Park, College Park, MD 20742.].
Kessler, W. S., 1990: Observations of long Rossby waves in the northern tropical Pacific. J. Geophys. Res.,95, 5183–5219.
——, and M. McPhaden, 1995: Oceanic equatorial waves and the 1991–93 El Niño. J. Climate,8, 1757–1774.
Kiladis, G. N., and H. van Loon, 1988: The Southern Oscillation. Part VII: Meteorological anomalies over the Indian and Pacific sectors associated with the extremes of the oscillation. Mon. Wea. Rev.,116, 120–136.
——, and H. F. Diaz, 1989: Global climatic anomalies associated with extremes in the Southern Oscillation. J. Climate,2, 1069–1090.
Latif, M., and T. P. Barnett, 1994: Causes of decadal climate variability over the North Pacific and North America. Science,266, 634–637.
Legler, D. M., 1991: Producing surface wind products for oceanographers. Proc. IGOSS (IOC/WMO) Ocean Products Workshop, Tokyo, Japan, Japan Meteor. Agency, 59–76.
McCreary, J. P., 1976: Eastern tropical ocean response to the changing wind systems. J. Phys. Oceanogr.,6, 632–645.
Meehl, G. A., 1987: The annual cycle and interannual variability in the tropical Indian and Pacific Ocean regions. Mon. Wea. Rev.,115, 27–50.
——, 1993: A coupled air–sea biennial mechanism in the tropical Indian and Pacific regions: Role of the ocean. J. Climate,6, 31–41.
——, 1994: Coupled land–ocean–atmosphere processes and South Asian monsoon variability. Science,266, 263–267.
Mooley, A. D., and J. Shukla, 1987: Variability and forecasting of the summer monsoon rainfall over India. Monsoon Meteorology, C. P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, 26–59.
Neelin, J. D., 1991: The slow sea surface temperature mode and the fast-wave limit: Analytical theory for tropical interannual oscillations and experiments in a hybrid coupled model. J. Atmos. Sci.,48, 584–606.
Ogallo, L. A., 1988: Relationship between season rainfall in East Africa and the Southern Oscillation. J. Climatol.,8, 32–38.
Pares-Sierra, A., and J. J. O’Brien, 1989: The seasonal and interannual variability of the California Current system: A numerical model. J. Geophys. Res.,94, 3159–3180.
Preisendorfer, R. W., 1988: Principal Component Analysis in Meteorology and Oceanography. Elsevier, 425 pp.
Shukla, J., 1987: Interannual variability of monsoons. Monsoons, J. S. Fein and P. L. Stephens, Eds., Wiley, 399–463.
——, and D. Paolino, 1983: The Southern Oscillation and the long-range forecasting of monsoon rainfall over India. Mon. Wea. Rev.,111, 1830–1837.
Sikka, D. R., 1980: Some aspect of the large scale circulations of summer monsoon rainfall over India in relation to fluctuations in the planetary and regional scale circulation parameter. Proc. Ind. Acad. Sci.,89, 179–195.
Tanimoto, Y., N. Iwasaka, K. Hanwa, and Y. Tobe, 1993: Characteristics variations of sea surface temperature with multipled time scales in the North Pacific. J. Climate,6, 1153–1160.
Terray, P., 1995: Space–time structure of monsoon interannual variability. J. Climate,8, 2595–2619.
Tourre, Y. M., and W. B. White, 1995: ENSO signals in global upper-ocean temperature. J. Phys. Oceanogr.,25, 1317–1332.
Villwock, A., and M. Latif, 1994: Indian Ocean response to ENSO. Proc. Int. Conf. on Monsoon Variability and Prediction, Vol. II, Geneva, Switzerland, World Meteor. Org., 530–537.
Webster, P. J., and S. Yang, 1992: Monsoon and ENSO: Selectively interactive systems. Quart. J. Roy. Meteor. Soc.,118, 877–926.
White, W. B., 1994: Slow El Niño–Southern Oscillation boundary waves. J. Geophys. Res,99, 22 737–22 751.
——, 1995: Design of a global ocean observing system for gyre-scale upper ocean temperature variability. Progress in Oceanography, Vol. 36, Pergamon, 169–217.
——, and C.-K. Tai, 1992: Reflections of interannual Rossby waves at the maritime western boundary of the tropical Pacific. J. Geophys. Res.,97, 14 305–14 322.
——, G. Meyers, J.-R. Donguy, and S. E. Pazan, 1985: Short-term climatic variability of the Pacific Ocean during 1979–1982. J. Phys. Oceanogr.,15, 917–935.
——, S. E. Pazan, G. W. Withee, and C. Noe, 1988: Joint Environmental Data Analysis (JEDA) Center for the scientific quality control of upper ocean thermal data in support of TOGA and WOCE. Eos, Trans. Amer. Geophys. Union,69, 122–123.
——, Y.-H. He, and S. E. Pazan, 1989: Off-equatorial westward propagating Rossby waves in the tropical Pacific during the 1982–83 and 1986–87 ENSO events. J. Phys. Oceanogr.,19, 1397–1406.
Wyrtki, K., 1989: Some thoughts about the west Pacific warm pool. Proc. West Pacific Int. Meeting and Workshop on TOGA COARE, Noumea, New Caledonia, Australia, ORSTOM, 99–109.
Yanai, M., and C. Li, 1994: Interannual variability of the Asian summer monsoon and its relationship with ENSO, Eurasian snow cover and heating. Proc. Int. Conf. on Monsoon Variability and Prediction, Vol. I, Geneva, Switzerland, World Meteor. Org., 27–34.
Yasunari, T., 1990: Impact of Indian monsoon on the coupled atmosphere/ocean system in the tropical Pacific. Meteor. Atmos. Phys.,44, 29–41.
Zhang, Y., J. M. Wallace, and D. Battisti, 1997: ENSO-like interdecadal variability: 1900–93. J. Climate,10, 1004–1020.
(Upper panel) Example of distribution in the Indo–Pacific regions (40°S–50°N, 30°E–180°) of available upper-ocean temperature observations collected for 1985. These observations consist of profiles of temperature in the upper 400 m from mostly XBTs, but also include small numbers of digital bathythermographs and hydrographic stations. (Lower panel) Time sequence of the number of temperature profiles collected each month in the Indo–Pacific domain for the 1979–91 period.
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
(Upper panel) Indo–Pacific distribution of the mean sea surface temperature (SST), computed for the 1980–89 period. The contour interval for the mean SST is 1°C, with shading for mean values greater than 25°C. (Lower panel) Rms of SST anomalies (°C) associated with the 3–6 year ENSO signal in the Indo–Pacific domain. Contour interval is 0.05°C, with shading for rms values greater than 0.2°C.
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
(Upper panel) Indo–Pacific distribution of the mean heat stored in the first 400 m of the ocean (HS) computed for the 1980–89 period. The contour interval for the mean HS is 109 J m−2, with shading for mean values greater than 30 × 109 J m−2. (Lower panel) Rms of HS anomalies (109 J m−2) associated with the 3–6 year ENSO signal in the Indo–Pacific domain. Contour interval is 0.1 × 109 J m−2 with shading for rms values greater than 0.2 × 109 J m−2.
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
(Upper panel) Indo–Pacific distribution of the mean surface pseudo wind stress for the 1980–89 period. The contour interval is 5.0 m2 s−2 (Middle panel) Rms of ZSWS anomalies (m2 s−2) associated with the 3–6 year ENSO signal in the Indo–Pacific domain. Contour interval is 1.5 m2 s−2 with shading for rms values greater than 5.5 m2 s−2. (Lower panel) Same as above but for MSWS anomalies. Contour interval is 3.0 m2 s−2, with shading for rms values greater than 2.5 m2 s−2.
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
Amplitude functions of the first EEOF of 2–7 year bandpassed SST, HS, ZSWS, and MSWS anomalies for the 1979–91 period. The length of the time period is shortened by the length of the EEOF lagged sequence (i.e., 2 years on both ends of the record).
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
Lag sequence of loadings distribution (spatial patterns) of the first 2–7 year bandpassed EEOF for SST anomalies (left), HS anomalies (right). The full sequences represent a complete ENSO cycle. The lags (windows) are displayed every 12 months. Shaded areas are associated with negative loadings.
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
Lag sequence of loadings distribution (spatial patterns) of the first 2–7 year bandpassed EEOF for ZSWS anomalies (left) and MSWS anomalies (right). The full sequence represents one complete ENSO cycle. The lags (windows) are displayed every 12 months. Shaded areas are associated with negative loadings.
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
Schematic diagram of the propagation of the ENSO signal (SST and ZSWS anomalies) in the Indo–Pacific domain. The windows are lagged by 12 months. Negative SST anomalies are shaded;
Citation: Journal of Physical Oceanography 27, 5; 10.1175/1520-0485(1997)027<0683:EOTESO>2.0.CO;2
