A Near-Annual Pacific Ocean Basin Mode

2478 JOURNAL OF CLIMATE A Near-Annual Pacific Ocean Basin Mode IN-SIK KANG AND JONG-SEONG KUG School of Earth and Environmental Sciences, Seoul National University, Seoul, South Korea SOON-IL AN International Pacific Research Center, University of Hawaii at Manoa, Honolulu, Hawaii FEI-FEI JIN Department of Meteorology, University of Hawaii at Manoa, Honolulu, Hawaii (Manuscript received 5 September 2003, in final form 5 January 2004) ABSTRACT Some fairly regular and nearly annual variability in the equatorial Pacific after the major 1997/98 El Niño event is studied. Sea level, sea surface temperature (SST), and surface wind anomalies of this variability are tied together in a way similar to the slow cycles of El Niño Southern Oscillations (ENSO). Despite a slightly longer-than-annual time scale, similar variability was also found in the datasets prior to 1997/98. This fastcoupled mode is superimposed on the slow 3 5-yr ENSO cycles in the tropical Pacific. It contributed to the occurrences of some minor El Niño and La Niña events. Zonal currents associated with the equatorial waves play a dominating role in generating SST anomalies of this fast mode through anomalous zonal advection. It is suggested that this fast mode may be best understood as a coupled Pacific Ocean basin (POB) mode. The existence of this fast mode, which appears independent of the slow ENSO mode, has important implications in understanding and predicting the tropical Pacific SST anomaly. 1. Introduction In addition to the known typical 3 5-yr ENSO cycles, there is a wide variety of other time scales in the tropical climate system. The dominant ENSO phenomenon only account for a fraction of the tropical Pacific variability. The frequency spectrum of the Niño-3 index (5 S 5 N, 150 90 W), which is indicative of tropical climate variability, shows variability also in the relatively short time scales of 8 18 months (Jin et al. 2003), and on quasi-biennial (Meehl 1987; Ropelewski et al. 1992) and decadal time scales (Tourrre et al. 1999; Zhang et al. 1999). Recently, Jin et al. (2003) show that there is significant variability of near-annual time scales in the tropical region. The present study is an extended work of Jin et al. (2003), so we will further show the observed characteristics of the near-annual mode and discuss its dynamical mechanism. As early as two decades ago, Cane and Moore (1981) found the so-called gravest ocean basin mode as a free solution of the linear reduced-gravity model under equa- Corresponding author address: In-Sik Kang, School of Earth and Environmental Sciences, Seoul National University, Seoul 151-742, South Korea. E-mail: kang@climate.snu.ac.kr torial -plane approximations of the tropical upperocean dynamics. The gravest ocean basin mode consists of an equatorial Kelvin wave and a combination of all symmetric Rossby waves such that no normal-flow boundary conditions are satisfied through wave reflections at the east and west boundaries. The period of this mode is equal to the time taken by Kelvin waves crossing the basin plus the time for the first-symmetry Rossby wave crossing the basin. With the gravity wave speed of about 2.5 m s 1 in the reduced gravity model, this yields a period of about 8 months for the Pacific Ocean. As a free-mode solution of the linear tropical ocean dynamic equations, the equatorial thermocline variation associated with this mode is characterized by a see-saw pattern in the western and eastern Pacific, whereas the upper-ocean zonal current has a maximum at the central equatorial Pacific and vanishes at the eastern and western boundary. This is the least damped mode in the dense spectrum of the tropical ocean dynamics (Jin 2001). Cane et al. (1990) showed analytically that by considering thermocline feedback (vertical advection of the anomalous subsurface temperature by the mean upwelling), this mode could be destabilized and transformed into a coupled mode with a greatly reduced frequency; they suggested that this coupled mode is responsible for ENSO. However, Jin and Neelin (1993) 2004 American Meteorological Society

15 JUNE 2004 KANG ET AL. 2479 FIG. 1. Time series of SST anomalies ( C) averaged over 2 S 2 N, 170 120 W. Solid and dashed lines indicate unfiltered and 24-month low-pass-filtered data, respectively. and Jin (1997a,b) suggested that the ENSO mode has its root in the low-frequency spectrum of tropical ocean dynamics compared to that of the ocean basin mode. The analytical solution of Neelin and Jin (1993) showed that the zonal advection by anomalous currents is effective for destabilizing the Pacific Ocean basin (POB) mode. In our recent study using the observed datasets (Jin et al. 2003), we found that there is a significant variability with a period of around 12 18 months. This mode appears to be related to the tropical ocean dynamics mode known as the gravest ocean basin mode (Cane and Moore 1981; Jin 2001). Its characteristics are modified through ocean atmosphere dynamical coupling. In this paper, we will describe the observed characteristics of the fast POB-like mode and discuss its mechanism and its significance in understanding and predicting the tropical Pacific SST anomaly. Section 2 describes the utilized data. The observed characteristics of the near-annual mode are described in section 3. The related dynamics of the near-annual mode is investigated using a simple model in section 4. The summary and discussion are given in section 5. 2. Data The data utilized are monthly means of sea surface temperature (SST), wind stress, oceanic current, and sea level over the tropical Pacific. The SST data are obtained from the National Centers for Environmental Prediction (NCEP), and were constructed on the basis of the EOF of observed SST (Reynolds 1988) and reconstructed after January 1981 using the optimum interpolation technique (Reynolds and Smith 1994). NCEP ocean assimilation data for 1980 2001 are used (Ji et al. 1995; Behringer et al. 1998; Vossepoel and Behringer 2000). Observed surface and subsurface ocean temperatures as well as satellite altimetry sea level data from Ocean Topography Experiment Satellite Mission (TOPEX/Poseidon) were assimilated into a Pacific Ocean basin general circulation model. The model was forced with weekly mean NCEP operational atmospheric analyses of surface winds and heat fluxes. Recently, Vossepoel and Behringer (2000) showed that the assimilation of TOPEX/Poseidon observations improves the dynamic height simulation. They demonstrated that the application of altimetry improves the mean salinity and leads to a more accurate calculation of the ocean density structure. 3. Observed characteristics of a near-annual Pacific Ocean basin mode The time series of the SST anomaly shown in Fig. 1 was averaged over the region of 2 S 2 N, 170 120 W, which is similar to the domain for the Niño-3.4 index (5 S 5 N, 170 120 W) except in meridional extent. This time series shows clearly some short-term fluctuations on top of the longer-term variations dominated by slow cycles of ENSO. In particular, there are some distinct fluctuations with nearly annual periodicity after the 1997/98 event of El Niño. Fluctuations with similar time scales occurred between all the major El Niño events, which are shown clearly in the local variance spectrum of the wavelet analysis of the time series (Fig. 1 from Jin et al. 2003). There is significant variance in the 1970s and earlier 2000s with the periodicity from 1.0 to 1.5 yr. The warm phases of the relative fast fluctuations become minor ENSO events. Overall, episodic fluctuations with a time periodicity of about 12 18 months are evident and are superimposed on the slow cycles dominated by major El Niño events. In order to focus on the relatively fast mode of variability, we first take a close look at the group following

2480 JOURNAL OF CLIMATE FIG. 2. Longitude time cross section of detrended (a) SST, (b) zonal wind, (c) surface zonal current (surface to 50 m), and (d) thermocline height anomalies along the equator for Oct 1998 Dec 2001. The linear trend for the period of Oct 1998 Dec 2001 is removed in each variable. the 1997/98 El Niño events as Jin et al. (2003) tried. The anomalous SST, zonal wind, zonal current, and sea level height are defined by removing the climatological annual cycle and the local linear trend of the anomalies during the period of late 1998 to the end of 2001. This allows us to isolate the fast fluctuations more clearly. As shown in Fig. 2, there are almost three cycles of fairly regular oscillations with nearly annual periodicity. The anomalies of all of these variables are clearly related to each other. These fast vacillations are quite different from the features of climatological annual cycles. The overall features of these fluctuations and relations among different fields are instead somewhat similar to those of ENSO, although the time scale here is much shorter. Moreover, there is some tendency for a westward propagation of anomalies in the SST and zonal wind from the eastern Pacific to western Pacific. Note that the westward propagation is also featured in the seasonal cycle but only over the eastern Pacific. The anomalies in surface zonal currents are very strong and are not dominated by anomalous Ekman currents that should be locally related to the zonal wind forcing (Figs. 2c and 3a). The large zonal current anomalies, thus, have to be largely associated with the equatorial waves. The modest equatorial sea level anomalies, which can be used to infer the thermocline depth or subsurface temperature anomalies, are correlated with the SST anomalies in the central and eastern Pacific. The equatorial SST anomalies are also produced by anomalous zonal advection due to anomalous zonal currents. The anomalous surface zonal current anomaly along the equator is dominated by the geostrophic current as illustrated in Fig. 3. Thus, the surface current anomalies along the equator are associated with equatorial oceanic Kelvin and Rossby waves. The latter could be driven by anomalous equatorial wind forced by the equatorial SST anomalies. Thus, the fast mode in this study is not likely to be the SST mode (Jin and Neelin 1993; Neelin and Jin 1993) but to be an ocean basin mode. It is known that tropical ocean atmosphere interaction can produce coupled modes through the zonal advective feedback. It was suggested first by Gill (1985) that zonal advective feedback destabilizes the equatorial oceanic Rossby waves. It was further shown by Neelin and Jin (1993) that the POB mode can be destabilized effectively by the zonal advective feedback. This kind of zonal advective feedback has been even suggested as responsible for the origin of ENSO (Picaut et al. 1997), although it was argued that the slow ENSO cycle is more likely dominated by the so-called thermocline

15 JUNE 2004 KANG ET AL. 2481 FIG. 3. Same as Fig. 2, except for (a) upper-ocean current (surface to 150 m) and (b) geostrophic current. feedbacks through anomalous vertical advection of subsurface temperature anomalies (Jin and An 1999; An and Jin 2001). In fact, some of the features shown in Fig. 2 resemble those of the weakly coupled POB mode of Neelin and Jin (1993). In particular, the analytical solution has features including near-annual periodicity, relatively large zonal current anomalies in the central equatorial Pacific, and the westward propagation of SST anomalies in the central to western Pacific. Therefore, we propose that this relative fast mode of variability associated with wave-induced strong zonal current anomalies can be viewed as a coupled POB mode independent from the slow ENSO mode. The fact that this mode occurs between major El Niño events also supports the hypothesis because the zonal advective feedback generating this fast mode becomes more effective when the background zonal temperature gradient is large. The spatial patterns associated with the fast mode are further examined by using the composite maps of the extreme and transition phases for SST and wind anomalies, and sea level and ocean current anomalies, as shown in Fig. 4. The three episodes of the fast nearannual mode are chosen for the period of 1998 2001. The three warm mature phases are chosen in January of 1999, 2000, and 2001. Also, three cold mature phases are chosen in July of 1999, 2000, and 2001. The composites show conditions 2 months before the mature phases, at the mature phase, and 2 months after, as the difference between warm and cold conditions. During the peak of warm phase, the SST pattern has its center located more toward the central to eastern equatorial Pacific compared to that of ENSO. The wind stress tends to have westerly wind anomalies in the western Pacific and easterly wind anomalies in the eastern Pacific. At the peak warm phase, the anomalous zonal currents are strong and positively reinforce the SST anomalies, clearly indicating the strong positive feedback through the zonal advection. The positive but modest equatorial sea level anomalies, indicating warm subsurface temperature anomalies, also favor positive SST anomalies. At the transition phase, anomalous zonal currents are strongly negative, whereas the equatorial sea level anomalies are only slightly negative. The anomalous cold advection by the large negative zonal currents enhances and expands the cold SST anomalies from the east equatorial Pacific to the central to western Pacific and leads the equatorial SST anomalies into the cold phase. Therefore, the main positive feedback and phase transition mechanisms of the fast mode are from the zonal advective feedback. For the slow ENSO mode, however, the thermocline feedback is primary for both the growth and phase transition of the slow cycle of ENSO, and the zonal advective feedback is in concert with but plays a secondary role to the thermocline feedback (Jin and An 1999; An and Jin 2001). To illustrate that this relatively fast mode of variability is not just limited to the past few years, we apply a 24-month high-pass filter for the period of 1980 2001 when we have NCEP National Center for Atmospheric Research (NCAR) reanalysis data. Figure 5a shows the time longitude cross section of the filtered SST along the equator for the period of 1991 2001. Relative fast variations with time scales of 12 18 months are clearly seen in the figure. We now examine the SST budget using the following equation to find out the dominant term responsible to the fast POB equation: T Tm Tm T u w wm Q. (1) t x z z The first, second, and third terms of the right-hand side of Eq. (1) are the anomalous zonal advection term, the anomalous upwelling term, and the thermocline displacement term associated with climatological mean upwelling, respectively. The meridional advection could be included in Q, because it plays a role of damping near the equator (Kang et al. 2001). Zonal and vertical advection terms are displayed in Figs. 5b 5d. The figures clearly show that the fast SST variation is dominated by the anomalous zonal advection term, particularly in the western and central Pacific, while vertical advection is still significant in the eastern Pacific. It is noted that the zonal advection term is in a quadrature

2482 JOURNAL OF CLIMATE FIG. 4. (a) (c) Composites of SST and wind stress anomalies, and (d) (f) sea level and zonal current anomalies at (a), (d) 2 months before the mature phase of warm SST condition, (b), (e) the mature phase, and (c), (f) 2 months after the mature phase. relationship with the SST variation, indicating that the SST tendency is controlled by the zonal advection. The vertical terms of Figs. 5c and 5d are confined in the eastern Pacific and appear to play minor roles over most of the Pacific basin except in the eastern Pacific. This is in contrast to the slow ENSO variation, which is mainly controlled by the thermocline displacement term (An et al. 1999; Jin and An 1999; Kang et al. 2001). 4. Unstable POB modes The previous section shows that the fast POB mode is separable from the slow ENSO mode and the major mechanism appears to be different from that of the ENSO, indicating that the two modes are coexisting in the tropical Pacific basin. In the present section, we confirm the coexistence of the two modes in observed basic states, using the simple two-strip model of An and Jin (2001). The shallow-water model for ocean dynamics under the longwave approximation on the equatorial plane is reduced to a single equation for the thermocline depth anomaly, h (Jin 1997b): 2 y ( t m)h [(2/y) y yy]( t m)h ( h y ) 0. (2) x x y x Equation (2) has been nondimensionalized by the oceanic Rossby deformation radius L o c o/ for y, the ocean basin width L ( 15 000 km) for x, the Kelvin wave crossing time L/c o (2 month) for t, the reference 2 depth H( 150 m) for h, and H c o /L for the wind stress x, respectively. Here, c o g H is an internal gravity wave speed, g is the reduced gravity parameter, and is the variation of the Coriolis parameter with latitude. The meridional component of the wind stress is neglected because the wind perturbations are predominantly zonal. The damping rate m is taken as (2.5 yr) 1. No normal motion at the eastern boundary and zerointegrated mass flux at the western boundary are assumed as the boundary conditions (e.g., Cane and Sarachik 1981; Jin 1997b) For simplicity of the dynamics, hemispheric symmetry of the system is assumed though some hemispheric asymmetry are observed (Kug et al. 2003). And, for the two-strip approximation, an equatorial strip and an off-equatorial strip are taken into consideration. As

15 JUNE 2004 KANG ET AL. 2483 FIG. 5. (a) Longitude time cross section of 24-month high-pass-filtered SST anomalies for a period of Jan 1991 Dec 2001. (b), (c), and (d) The tendency terms of the high-pass-filtered SST associated with anomalous zonal advection, anomalous thermocline displacement, and upwelling, respectively. The values plotted are the meridional averages for 5 S 5 N. shown in Jin (1997b), the ocean dynamics equations for h e in the equatorial strip (y 0) and h n for the offequator strip centered at y n can be approximately written as ( t m)(he h n) h x e xe 2 ( t m)hn h x n/yn ( y x/y) y y n 2 /yn xe, (3) where is about 1.0 1.25, depending on y n and the ratio of the oceanic and atmospheric Rossby radii of deformation; and xe is the wind stress anomaly evaluated in the equatorial strip. The first part of Eq. (3) describes the Kelvin wave signal along the equator with an inclusion of the effect of Rossby waves because h n is the thermocline depth associated with Rossby wave signals in the off-equatorial strip. The boundary conditions for the two-strip model are written as h (x, t) rh(x, t), h (x, t) r h (x, t), n E E e E e W W n W where r W and r E are reflection parameters depending on the boundary conditions, and x E and x W indicate eastern and western boundary, respectively. A change of SST may be described by the thermodynamics of a constant depth mixed layer embedded in the upper-layer ocean. Adopting the equatorial strip approximation to the SST equation and considering only the dominant processes, the equatorial SST anomaly equation is linearized about an upwelling climate state and the zonal gradient of climatological mean SST (An and Jin 2001): where T c(x)t (x)h a(x)u, (4) t e e e m

2484 JOURNAL OF CLIMATE FIG. 6. Longitudinal distributions of the coefficient values in the SST equation of Eq. (4). (a) The zonal advection and (b) upwelling terms, respectively. The solid line indicates the values obtained from the time means for 1980 2001, and the dashed line for the 1999 2001 means. L w H L w c(x) T, (x) o(x) c H T c H 1 1 o 1.5 o 1.5 L T(x) a(x), T x where a(x), (x), and c(x) are the coefficients associated with the zonal advection, thermocline displacement, and a Newtonian cooling term, respectively. The coefficients, as functions of longitude, are computed using the NCEP ocean assimilation data. Figure 6a shows longitudinal distributions of the coefficient a(x) for the 21- yr mean state of 1981 2001 (solid line) and for the 3-yr mean state of 1999 2001 (dashed line). As seen in Fig. 6a, the zonal advection coefficients for the two periods are not much different from each other along most of the equatorial Pacific, except the western Pacific. On the other hand, the upwelling coefficients (x) along the central and eastern Pacific are quite different for the two periods. The values for the recent 3-yr means are about half of those of the 21-yr means, indicating that the upwelling was not active for the recent period. The weak upwelling is related to the relatively weak zonal wind stress prevailing for the 3-yr period (not shown), particularly in the central and eastern Pacific. Considering the two terms shown in Fig. 6, one can say that for the recent 3-yr period, the zonal advection plays a more important role for the SST variation than that of the 21- yr mean, because of the relatively weak mean upwelling associated with weaker easterly wind stress. The wind stress forcing in Eq. (3) is expressed in terms of SST as follows: [ ] n j j xe xe Te e j x (x) V (x) V (x)t (x), (5) j j where V and VT indicate idealized wind stress and xe e SST patterns having sinusoidal function, respectively. It is assumed the wavelength of the sinusoidal wind stress and SST pattern is same as the ocean basin scale. In this formulation, the wind stress pattern was obtained by shifting the SST by 0.4. We used two idealized patterns (n 2), and they are in quadrature relationship between each other. Note that the formula of Eq. (5) is similar to that of the statistical atmosphere model based on SVD analysis between SST and wind stress (e.g., Kang and Kug 2000). In Eq. (3), the finite difference method is applied to the x dependence of the variables. The spatial scale of the ocean basin is chosen to be 16 200 km and the ocean basin is discretized by equally spaced 50 grids. The eigenmodes of Eqs. (2) and (3) are obtained in the grid system, and the unstable modes are discussed in the following. For the basic state of the 22-yr means of 1980 2001, two unstable modes are obtained. The first unstable mode has an e-folding growth rate of 0.33 yr 1 and an oscillation period of 5.2 yr. The longitude time section of SST for the first unstable mode is shown in Fig. 7a. The variation pattern is very close to that of ENSO, which has large SST variations in the eastern Pacific and small variations in the western Pacific. On the other hand, the second unstable mode shown in Fig. 7b has basinwide SST variations, although westward propagation is also seen in the figure. The second unstable mode has an e-folding growth rate of 0.18 yr 1 and an oscillation period of 0.8 yr. The spatial pattern and oscillation period of the second unstable mode are similar to those of the observed counterparts shown in the previous section, indicating that the observed near-annual mode is in fact reproducible by using the present simple model. It is important to note that the present model results demonstrate the coexistence of the slow ENSO and the fast near-annual modes in the tropical Pacific as unstable modes. For the basic state of the recent 3-yr means of 1999 2001, the first unstable mode becomes the fast mode, which has a time scale of about 1 yr (Fig. 8). The variation pattern is very close to that shown in Fig. 7a, except that the westward propagation is obvious. The e-folding growth rate is 0.25 yr 1. The slow ENSO mode now becomes the second unstable mode. The spatial pattern and oscillation period of the second unstable mode are very similar to those of the first unstable mode obtained using the 22-yr mean basic state. The e-folding rate of the slow mode is now 0.12 yr 1, which is much weaker than that of the fast mode. Therefore, for the recent 3-yr period, the fast unstable mode is dominant and the ENSO mode is not well developed compared to the fast mode. The results indicate that the fast nearannual mode and the slow ENSO mode are coexisting as coupled unstable modes in the tropical Pacific, and that the fast mode can pop up for some period of the years, when the slowly varying basic state is favorable

15 JUNE 2004 KANG ET AL. 2485 FIG. 7. Longitude time cross section of SST for (a) first and (b) second most unstable eigenmodes along the equator. The eigenmodes are obtained based on the basic state of the 21-yr mean of 1980 2001. for the fast mode. For the recent 3 yr, the mean state is favorable to the fast mode mainly because of the weakening of the mean upwelling in the equatorial Pacific. Figure 9 shows the eigenmode of the fast mode for the recent basic state. The eigenmode consists of its real and imaginary parts, which have a quadrature relationship with each other. The spatial pattern of the real part leads that of the imaginary part. The real part may represent the mature phase of the fast mode. There is equatorial SST warming. In response to the warm SST, zonal wind stress is westerly over and to the west of the warm SST and easterly to the far east of the warm SST. The wind stress and related wind stress curl induce off-equatorial upwelling Rossby waves to the west of the warm SST, and downwelling Rossby waves to the east of the warm SST. The distribution of the zonal current, as FIG. 8. As in Fig. 7, except for the basic state of the 1999 2001 mean.

2486 JOURNAL OF CLIMATE FIG. 10. Trajectory plots for the zonal current and SST averaged over central Pacific from observation (dashed line) during 1998 2001 and two-strip model (solid line). FIG. 9. Spatial patterns of the eigenmode shown in Fig. 8a: SST (solid line), off-equatorial thermocline depth (dashed line), equatorial thermocline depth (dotted line), and zonal current (dash-dot line). shown in Fig. 9a, is contributed by these Rossby waves. Note that the equatorial Kelvin wave plays a role in reducing the zonal current induced by the Rossby waves. The imaginary part may represent a spatial pattern of onset phase. The equatorial SST and thermocline depth anomaly is relatively weak. The off-equatorial thermocline depth anomalies, however, are significantly stronger in the central Pacific. This is a result of propagating the Rossby waves in the real part. Note that the Rossby waves result from easterly wind stress and anticyclonic wind stress curl. The Rossby waves induce a strong easterly zonal current, and develop cold SSTs. The phase transition is largely due to the wave dynamics of the ocean basin mode. During the warm phase, the upwelling westward propagating off-equatorial Rossby waves are reflected to the upwelling Kelvin wave at the western boundary. The upwelling Kelvin wave arrives at the central and eastern Pacific in a relatively short time and induces negative zonal current anomalies and thermocline shallowing and, thus, produces cold SST anomalies. In Fig. 10, we plot the phase diagram for the SST anomalies and the zonal current anomalies in the central Pacific from the fast mode and from the observation. It is clear that when SST anomalies are near 0, there is anomalous zonal advection, which brings it into the next phase. We compared the relative contributions of the thermocline term and anomalous zonal advection term in the SST [Eq. (4)] for the fast and slow modes. We calculate this ratio for the fast mode in the central Pacific region close to the location of maximum SST anomalies. We found that the ratio is about 0.35, clearly indicating that the zonal advection plays the dominating role in generating the SST anomalies of this coupled mode. In contrast, we also calculate this ratio for the slow mode in the eastern equatorial Pacific where the maximum SST anomaly appears for this mode. The ratio is about 10.0, indicating that for the slow mode, it is dominated by the so-called thermocline feedback. The mechanism of the slow mode can be explained by the recharge oscillator mechanism as proposed by Jin (1996, 1997a,b). The local wind stress in the eastern Pacific tends to contribute to reinforce this mode. During the warm phase, in addition to the westerly wind anomalies over the SST anomaly and to its west, there are also easterly wind anomalies to the far east of the SST anomaly. The easterly wind anomalies and the associated wind stress curl anomalies off the equator generate downwelling Rossby waves in the eastern Pacific. These Rossby waves produce negative equatorial zonal current anomalies that propagate westward. These initiate a cold SST anomaly by cold advection in the east. These cold SST anomalies in the east further enhance the easterly wind stress anomalies and, thus, amplify the cold advection anomalies. This cold advection adds to the cold advection in the central to western Pacific that resulted from

15 JUNE 2004 KANG ET AL. 2487 FIG. 11. Longitude time cross sections of SST for near annual mode of the two-strip model when eastern Pacific wind is removed. the upwelling Kelvin waves from the western boundary reflection. If we remove the wind stress pattern in the eastern equatorial Pacific, the SST anomalies are confined to the central Pacific and it becomes a standing oscillation (Fig. 11). The growth rate is much reduced and becomes negative although the period remains nearly unchanged. Thus, the structure of atmospheric wind stress response plays an important role in the instability of the coupled POB mode. 5. Summary We have revealed the existence of a fast near-annual mode of coupled variability in the tropical Pacific ocean atmosphere system. This mode of variability appears to be related to the theoretical coupled POB mode controlled by the zonal advective feedback (Cane and Moore 1981; Neelin and Jin 1993). It appears independent of the slow ENSO mode of variability. The observational evidence presented in this paper indicates the existence of a fast coupled mode in nature. The zonal advective feedback seems to be indeed responsible for this fast coupled mode because the SST anomalies are always strongly associated with the zonal current anomalies related to the equatorial waves. It is known that the POB mode is associated with strong zonal current anomalies in the central equatorial Pacific. Moreover, it is in the central equatorial Pacific where the zonal SST gradient is largest. Thus, the zonal advective feedback can be very effective in destabilizing and transforming the POB mode into this kind of fast coupled mode. Using a two-strip model, it is demonstrated that there are indeed two independent leading coupled modes, an interannual mode, which is also known as the recharge oscillator and the fast mode, which is a coupled POB mode. The phase transition of the fast coupled mode is naturally from the wave dynamics of the ocean basin mode. Its stability is sensitive to the basic state. Its SST tendency is dominated by the anomalous zonal advection due to anomalous currents induced by the oceanic Kelvin and Rossby waves. The local wind stress pattern in the eastern Pacific contributes to instability of this mode. If we remove the wind stress pattern in the eastern equatorial Pacific, the SST anomalies are confined to the central Pacific and it becomes a standing oscillation (Fig. 11). The growth rate is much reduced and becomes negative although the period remains nearly unchanged. Thus, the structure of the atmospheric wind stress response plays an important role in the instability of the coupled POB mode. The existence of the fast coupled mode has consequences in understanding the coupled dynamics of ENSO. This result implies that the coupled slow ENSO mode is not related to the gravest ocean basin mode as indicated by the earlier simple coupled wave oscillator model for ENSO (Cane et al. 1990). The existence of an independent fast coupled mode indirectly confirms that the slow ENSO mode is related to the low-frequency spectrum of ocean dynamics as suggested in Jin et al. (1993), and Jin (1997b, 2001). Moreover, the coexistence of this coupled fast mode and the coupled slow ENSO mode are at least partly responsible for the richness of the coupled variability in the Tropics. The existence of the coupled POB mode in nature also poses a new challenge to the ENSO prediction. When the background in the equatorial central to eastern Pacific is warm, such as in the early 1990s, this fast mode variability may surface as minor in nearly-annual El Niño events, whereas during cold conditions in the background, the fast mode leads to nearly-annual La Niña events such as in the past few years. As perhaps a weakly coupled mode, it may be stochastically excited and its predictability limit is likely to be less than its half-lifetime cycle, or up to two seasons. Further studies are needed for a better understanding this fast mode variability and its interaction with the slow ENSO mode, which will improve our the skill in predicting major and minor El Niño and La Niña events. Acknowledgments. I.-S. Kang and J.-S. Kug are supported by the SRC program of Korea Science and Engineering Foundation and the Brain Korea 21 project. S.-I. An has been supported by Frontier Research Sys-

2488 JOURNAL OF CLIMATE tem for Global Change through its sponsorship of the International Pacific Research Center, and F.-F. Jin was supported by NSF Grant ATM-0226141 and NOAA Grants GC01-229 and GC01246, and SOEST 6362 and IPRC 260. REFERENCES An, S.-I., and F.-F. Jin, 2001: Collective role of zonal advective and thermocline feedbacks in ENSO mode. J. Climate, 14, 3421 3432.,, and I.-S. Kang, 1999: The role of zonal advection feedback in phase transition and growth of ENSO in the Cane Zebiak model. J. Meteor. Soc. Japan, 77, 1151 1160. Behringer, D. W., M. Ji, and A. Leetmaa, 1998: An improved coupled model for ENSO prediction and implication for ocean initialization. Part I: The ocean data assimilation system. Mon. Wea. Rev., 126, 1013 1021. Cane, M. A., and D. W. Moore, 1981: A note on low-frequency equatorial basin modes. J. Phys. Oceanogr., 11, 1578 1584., and E. S. Sarachik, 1981: The periodic response of a linear, baroclinic equatorial ocean. J. Mar. Res., 39, 651 693., M. Münnich, and S. E. Zebiak, 1990: A study of self-excited oscillations of the tropical ocean atmosphere system. Part I: Linear analysis. J. Atmos. Sci., 47, 1562 1577. Gill, A. E., 1985: Elements of coupled ocean atmosphere models for the tropics. Coupled Ocean Atmosphere Models, Elsevier Oceanography Series, Vol. 40, Elsevier, 303 328. Ji, M., A. Leetmaa, and J. Derber, 1995: An ocean analysis system for seasonal to interannual climate studies. Mon. Wea. Rev., 123, 460 481. Jin, F.-F., 1996: Tropical ocean atmosphere interaction, the Pacific cold tongue, and the El Niño Southern Oscillation. Science, 274, 76 78., 1997a: An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. J. Atmos. Sci., 54, 811 829., 1997b: An equatorial ocean recharge paradigm for ENSO. Part II: A stripped-down coupled model. J. Atmos. Sci., 54, 830 847., 2001: Low-frequency modes of the tropical ocean dynamics. J. Climate, 14, 3874 3881., and J. D. Neelin 1993: Modes of interannual tropical ocean atmosphere interaction A unified view. Part III: Numerical results. J. Atmos. Sci., 50, 3523 3540., and S.-I. An, 1999: Thermocline and zonal advective feedbacks within the equatorial ocean recharge oscillator model for ENSO. Geophys. Res. Lett., 26, 2989 2992., J.-S. Kug, S.-I. An, and I.-S. Kang, 2003: A near-annual coupled ocean atmosphere mode in the equatorial Pacific Ocean. Geophys. Res. Lett., 30, 1080, doi:10.1029/2002gl015983. Kang, I.-S., and J.-S. Kug, 2000: An El-Niño prediction system with an intermediate ocean and a statistical atmosphere model. Geophys. Res. Lett., 27, 1167 1170., S.-I. An, and F.-F. Jin, 2001: A systematic approximation of the SST anomaly equation for ENSO. J. Meteor. Soc. Japan, 79, 1 10. Kug, J.-S., I.-S. Kang, and S.-I. An, 2003: Symmetric and antisymmetric mass exchanges between the equatorial and off-equatorial Pacific associated with ENSO. J. Geophys. Res., 108, 3284, doi:10.1029/2002jc001671. Meehl, G. A., 1987: The annual cycle and interannual variability in the tropical Indian and Pacific Ocean regions. Mon. Wea. Rev., 115, 27 50. Neelin, J. D., and F.-F. Jin, 1993: Modes of interannual tropical ocean atmosphere interaction A unified view. Part II: Analytical results in the weak-coupling limit. J. Atmos. Sci., 50, 3504 3522. Picaut, J., F. Masia, and Y. du Penhoat, 1997: An advective reflective conceptual model for the oscillatory nature of the ENSO. Science, 277, 663 666. Reynolds, R. W., 1988: A real-time global sea surface temperature analysis. J. Climate, 1, 75 86., and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate, 7, 929 948. Ropelewski, C. F., M. S. Halpert, and X. Wang, 1992: Observed tropospheric biennial variability and its relationship to the Southern Oscillation. J. Climate, 5, 594 614. Tourre, Y. M., Y. Kushnir, and W. B. White, 1999: Evolution of interdecadal variability in sea level pressure, sea surface temperature, and upper ocean temperature over the Pacific Ocean. J. Phys. Oceanogr., 29, 1528 1541. Vossepoel, F. C., and D. W. Behringer, 2000: Impact of sea level assimilation on salinity variability in the western equatorial Pacific. J. Phys. Oceanogr., 30, 1706 1721. Zhang, R.-H., L. M. Rothstein, and A. J. Busalacchi, 1999: Interannual and decadal variability of the subsurface thermal structure in the Pacific Ocean. Climate Dyn., 15, 703 717.