1. Introduction
The upper-layer circulation of the western equatorial Pacific Ocean (WEPO) is complex in structure and highly variable on a wide range of time scales. It consists of the zonal equatorial currents such as the westward-flowing South Equatorial Current (SEC) in the surface layer and the eastward-flowing Equatorial Undercurrent (EUC) in the thermocline (e.g., Wyrtki and Kendall 1967; Wyrtki 1974; Kessler and Taft 1987; Delcroix et al. 1992; Reverdin et al. 1994; Reid 1997) and western boundary currents such as the seasonally reversing New Guinea Coastal Current (NGCC) and the underlying New Guinea Coastal Undercurrent (NGCUC; Fig. 1; Lindstrom et al. 1987; Ueki 2003; Radenac et al. 2016). The slanted western boundary of the WEPO, particularly the New Guinea island (NGI), is a fundamental constraint regulating the structures of the NGCC and NGCUC and their variabilities (e.g., Ridgway et al. 1993; Butt and Lindstrom 1994; Mackey et al. 2002; Melet et al. 2010a, 2013; Cravatte et al. 2011; Hristova and Kessler 2012). The northwestward-flowing NGCUC carries the thermocline and intermediate water masses of Southern Hemisphere origin and serves as the major source of the EUC (e.g., Tsuchiya et al. 1989; Fine et al. 1994; Qu and Lindstrom 2004; Goodman et al. 2005; Grenier et al. 2014; Radenac et al. 2016; Wang et al. 2019).
Climatological JJA currents (m s−1; white vectors; currents weaker than 0.02 m s−1 are not plotted) at (a) 10 and (b) 250 m in the western tropical Pacific based on the GODAS data of 1981–2010. The green dots denote the locations of the four subsurface moorings: M3N, M2N, M0, and M1S. The North Equatorial Countercurrent (NECC), South Equatorial Current (SEC), New Guinea Coastal Current (NGCC), Equatorial Undercurrent (EUC), and New Guinea Coastal Undercurrent (NGCUC) are highlighted.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Climatological JJA currents (m s−1; white vectors; currents weaker than 0.02 m s−1 are not plotted) at (a) 10 and (b) 250 m in the western tropical Pacific based on the GODAS data of 1981–2010. The green dots denote the locations of the four subsurface moorings: M3N, M2N, M0, and M1S. The North Equatorial Countercurrent (NECC), South Equatorial Current (SEC), New Guinea Coastal Current (NGCC), Equatorial Undercurrent (EUC), and New Guinea Coastal Undercurrent (NGCUC) are highlighted.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Climatological JJA currents (m s−1; white vectors; currents weaker than 0.02 m s−1 are not plotted) at (a) 10 and (b) 250 m in the western tropical Pacific based on the GODAS data of 1981–2010. The green dots denote the locations of the four subsurface moorings: M3N, M2N, M0, and M1S. The North Equatorial Countercurrent (NECC), South Equatorial Current (SEC), New Guinea Coastal Current (NGCC), Equatorial Undercurrent (EUC), and New Guinea Coastal Undercurrent (NGCUC) are highlighted.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The WEPO regional circulation variability is mainly governed by equatorial Rossby–Kelvin wave dynamics forced by equatorial winds (Boulanger et al. 2004; Yuan et al. 2004; D. Chen et al. 2015; Chiodi et al. 2014; Melet et al. 2013; Zhang and Clarke 2017) and modulated by the complex topography the western boundary (e.g., du Penhoat and Cane 1991; Clarke 1991). Changes in either strength or transported seawater properties can dramatically influence the heat budget of the western Pacific warm pool and ocean–atmosphere interaction of the tropical Pacific (Meyers and Donguy 1984; Lukas et al. 1996; Wang and McPhaden 2000; Picaut et al. 2001; Hu et al. 2015). The WEPO is also linked to the Indian Ocean through the Indonesian Throughflow (ITF), which is a key component of the global ocean conveyor belt (e.g., Gordon 1986; Gordon and Fine 1996; Sprintall et al. 2014). It is also connected with the midlatitude North Pacific via the Mindanao Current and the Kuroshio (e.g., Nitani 1972; Toole et al. 1990; Li and Wang 2012; Gordon et al. 2014) and with the midlatitude region South Pacific via the NGCUC across the Vitiaz Strait (e.g., Fine et al. 1994; Qu and Lindstrom 2002; Melet et al. 2010b, 2013; Davis et al. 2012; Grenier et al. 2011; Kessler and Cravatte 2013), thus is involved in the subtropical–tropical coupling of the Pacific climate (e.g., Latif and Barnett 1996; Gu and Philander 1997; McPhaden and Zhang 2002). With these regards, investigating the variability of the WEPO circulation is useful for the integration of our knowledge in tropical ocean and climate dynamics in both regional and global senses.
Under the dominant influence of El Niño–Southern Oscillation (ENSO), the WEPO circulation exhibits strong interannual variability and exerts important feedback effects on ENSO events, as, for example, illustrated within the recharge–discharge paradigm (Jin 1997; Jin and An 1999). Interannual variations of the upper WEPO circulation have been reported and extensively studied in existing literature using various sources of observational data (e.g., Taft and Kessler 1991; Delcroix et al. 1992; Qiu and Joyce 1992; Johnston and Merrifield 2000; Bonjean and Lagerloef 2002; Johnson et al. 2002; Shinoda et al. 2011; Hsin and Qiu 2012; Zhao et al. 2013), and these studies underscore the ENSO’s surface wind forcing through both localized dynamics or far-reaching processes such as Rossby wave propagation (e.g., Meyers 1979; Kessler 1990; Masumoto and Yamagata 1991; Qiu and Joyce 1992; Qiu and Lukas 1996). Generally, the weakened Pacific trade winds under El Niño condition cause reductions of the SEC and EUC in strength (e.g., Firing et al. 1983; Delcroix et al. 1992; Johnson et al. 2002; Izumo 2005; Qin et al. 2015), and the EUC may disappear during strong El Niño events (e.g., Halpern 1987; McPhaden et al. 1990; Seidel and Giese 1999; Johnson et al. 2000; Izumo 2005; Stramma et al. 2016). The variability of NGCC and NGCUC is particularly complex and less appreciated, as compared with zonal currents (e.g., Melet et al. 2010a,b, 2013; Qin et al. 2015, 2016; Kessler et al. 2019). Mooring observations showed that the NGCUC were significantly enhanced during the 1997–98 El Niño, and the NGCC, which was supposed to flow southeastward, disappeared in that winter (Ueki 2003).
Despite existing studies reviewed above, our knowledge of the WEPO circulation variability during the strong ENSO events is still far from complete, especially for the far western Pacific region. The lack of direct observation hinders a complete understanding of the equatorial ocean dynamics and its impacts on ENSO. To obtain continuous observational records of the upper-ocean circulation in the far western Pacific, multiple subsurface moorings were deployed since 2014 by the Institute of Oceanography, Chinese Academy of Sciences (IOCAS), under the framework of the Scientific Observing Network of the Chinese Academy of Sciences (CASSON; Wang et al. 2016a,b; Lyu et al. 2018; Ma et al. 2019). These mooring observations successfully captured variations of the WEPO circulation during the 2015–16 super–El Niño. As one of the strongest events in observational history, the 2015–16 El Niño is comparable in amplitude to the previous 1982–83 and 1997–98 events and exhibits unique characteristics (Chen et al. 2017; Corbett et al. 2017; L’Heureux et al. 2017; Paek et al. 2017; Ren et al. 2017; Xue et al. 2017). For instance, the 2015–16 event was developed upon a relatively warmer background condition of the equatorial Pacific as the remnants of the 2014 warming and had stronger warming signatures in the central Pacific (e.g., Lim et al. 2017; Chiodi and Harrison 2017; Hu and Fedorov 2017; Levine and McPhaden 2016; Min et al. 2015); intraseasonal WWEs occurred in the central Pacific Ocean and promoted the development of warm SSTAs, which were not the case in the 1982–83 and 1997–98 event (Marshall et al. 2016; Hong et al. 2017; Puy et al. 2017; Lyu et al. 2018). The subsurface moorings of CASSON documented unexpected, intriguing variability signatures during the decaying stage of this event (2016 summer) that deserves a particular investigation.
The present study mainly analyzes the mooring observational data of CASSON to understand the dynamical response of the WEPO upper circulation to the 2015–16 El Niño. The observed circulation anomalies are quantified and described, and the underlying mechanisms are explored through sensitive experiments of a reduced-gravity ocean (RGO) model. The rest of the paper is organized as follows. Section 2 outlines the datasets, model, and methods utilized in this study. Section 3 presents the variations of the WEPO circulation observed by CASSON. Section 4 explores the mechanisms of the WEPO circulation variability using the RGO model experiments. Section 5 provides a summary and discussion for the main findings of the paper.
2. Data and methods
a. Mooring data
Ocean current records of four CASSON subsurface moorings during 2014–18 are analyzed (Fig. 1). These moorings are located respectively at 1°S, 142°E (M1S); 0°, 142°E (M0); 2°N, 140°E (M2N); and 3°N, 143.55°E (M3N). Each mooring was equipped with one upward-looking and one downward-looking Teledyne RD Instruments (TRDI) 75-kHz ADCP mounted on the main float at the design depth of 400–500 m. The accuracy of the 75-kHz ADCP is within ±0.5 cm s−1 in velocity magnitude and ±5° in direction. M1S was initially deployed in November 2015 and was maintained in December 2017 and December 2018. Mooring M0 was deployed in August 2014 and was maintained in November 2015, December 2016, December 2017, and December 2018. M2N was deployed in January 2014 and was maintained in August 2014, November 2015, December 2016, December 2017, and June 2018. M3N was deployed in October 2015 and recovered in December 2016. The mooring provides zonal and meridional ocean velocities (U and V) over 50–1000 m with a vertical bin size of 8 m and a measurement frequency of 35 pings per hour. We selected ADCP records meeting the criteria of 1) at least three of the four correlation magnitudes > 64, 2) echo intensity >30 counts, 3) percent good > 50%, and 4) error velocity in each bin < 15 cm s−1, after which we further removed obvious incorrect data points through hand editing. Among them, the correlation magnitude is a measure of the pulse-to-pulse correlation in a ping for each bin depth. The echo intensity is a measure of the signal strength intensity returned to the transducer, and a test is applied by comparing the echo intensity at a particular bin to the echo intensity of the previous bin. Percent good is the ratio of good pings per total pings for each ensemble. Error velocity is a measure of the disagreement of measurement estimates of opposite beams. ADCP data from the four moorings were interpolated vertically to 1-m intervals and averaged into daily mean data. To eliminate the effects of tides, a 9-day running average was applied.
In addition to CASSON moorings, a mooring with also ADCP measurements was deployed by the Japan Marine Science and Technology Center (JAMSTEC) at approximately the same location to M0 (0°, 142°E) during the period of 1994–98 (Kuroda 2000; Matsuura 2002; Ueki 2003). This mooring fortunately captures the evolution of the 1997–98 super–El Niño and provides a valuable reference for CASSON mooring records. Combining JAMSTEC and CASSON mooring data, we have nearly 10 years of current data at M0, which are used to calculate the monthly climatology and obtain monthly anomaly. Monthly ADCP measurements of Tropical Atmosphere Ocean/Triangle Trans-Ocean Buoy Network (TAO/TRITON) array for 0–250 m are available at five sites (147°E, 165°E, 170°W, 140°W, and 110°W) along the equator, covering the interior Pacific basin from 1990 through 2016 (McPhaden et al. 1998). These ADCP data are also utilized to aid the analysis of WEPO circulation variability.
b. Satellite and reanalysis datasets
The ocean surface current data obtained from Ocean Surface Current Analyses Real-Time (OSCAR; Bonjean and Lagerloef 2002; Johnson et al. 2007) and the Global Ocean Data Assimilation System (GODAS; Behringer and Xue 2004) were also used to understand the WEPO circulation. The OSCAR data are derived from multisatellite data of sea surface height, sea surface winds, sea surface temperature (SST), and mean dynamic topography, using geostrophic, Ekman, and Stommel shear dynamics. OSCAR provides ocean current to estimate the total ocean velocity of the upper 30 m on global grids of ⅓°. GODAS is based on the Geophysical Fluid Dynamics Laboratory (GFDL) Modular Ocean Model (MOM), version 3 (Pacanowski and Griffies 1999), simulation forced by National Centers for Environmental Prediction (NCEP) atmospheric Reanalysis 2 fields (Kanamitsu et al. 2002), covering from 75°S to 65°N and having horizontal resolutions of 1° × ⅓° and 40 vertical layers. GODAS adopts three-dimensional variational data assimilation (3DVAR) scheme to assimilate the temperature profiles from expendable bathythermographs (XBTs), buoy arrays, and Argo profiling floats. For our analysis purpose, the 5-day U component of the OSCAR surface current of 1993–2016 and the 5-day 0–350-m average U component of GODAS of 1981–2016 are used. Validation for the GODAS data in the simulation of the EUC and its water source has been provided in several existing papers (e.g., Wang and Wu 2013; Xue et al. 2017). The 0.25° × 0.25° monthly sea surface height data from Archiving Validation and Interpretation of Satellite Oceanography (AVISO) (Le Traon et al. 1998; Ducet et al. 2000) for 1993–2017 and 0.75° monthly wind stress of the European Center for Medium-Range Weather Forecasts (ECMWF) ERA-Interim data for 1979–2017 (Dee et al. 2011) were also used.
c. 1.5-layer nonlinear RGO model
A 1.5-layer nonlinear RGO model is employed to simulate the observed WEPO variability and explore the underlying dynamical processes. This model is configured to the closed Pacific Ocean basin between 100°E and 70°W and 40°S and 40°N with horizontal resolutions of 0.25° × 0.25° (Fig. 2a). The South China Sea and the Maritime Continent are masked as landmass, and therefore the western boundary is closed, without outflows of the ITF and Luzon Strait transport. The islands surrounding Solomon Sea, including New Britain, Buka Island, Bougainville Island, and Solomon Islands are all removed. Such a simplified western boundary may affect the WEPO circulation and its variability in a dramatic manner, which is discussed in section 4 using two experiments with more realistic topography (Figs. 2b,c).
(a) Land–sea distribution of the 1.5-layer RGO model. The gray shading denotes the simplified land areas in CTR, and the yellow shading combined with the gray shading denotes the land areas in the “Island” experiment. (b) A zoomed-in view of the Solomon Sea region in (a). New Britain, Buka Island, Bougainville Island, and the Solomon Islands are taken into account in the Island experiment. (c) The land–sea distribution of the Indo-Pacific experiment. The Makassar Strait is opened as the only channel for the ITF.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
(a) Land–sea distribution of the 1.5-layer RGO model. The gray shading denotes the simplified land areas in CTR, and the yellow shading combined with the gray shading denotes the land areas in the “Island” experiment. (b) A zoomed-in view of the Solomon Sea region in (a). New Britain, Buka Island, Bougainville Island, and the Solomon Islands are taken into account in the Island experiment. (c) The land–sea distribution of the Indo-Pacific experiment. The Makassar Strait is opened as the only channel for the ITF.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
(a) Land–sea distribution of the 1.5-layer RGO model. The gray shading denotes the simplified land areas in CTR, and the yellow shading combined with the gray shading denotes the land areas in the “Island” experiment. (b) A zoomed-in view of the Solomon Sea region in (a). New Britain, Buka Island, Bougainville Island, and the Solomon Islands are taken into account in the Island experiment. (c) The land–sea distribution of the Indo-Pacific experiment. The Makassar Strait is opened as the only channel for the ITF.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
This model is able to represent the ocean’s first baroclinic mode response to surface wind forcing that is the dominant form of interannual variability in sea level and upper-ocean circulation of the WEPO. The initial upper-layer thickness (ULT) is set to H = 350 m. The coefficient of horizontal eddy viscosity is 1500 m2 s−1 between 30°S and 30°N and increases to 7500 m2 s−1 poleward of 30° to damp artificial coastal Kelvin waves along the northern and southern boundaries (Duan et al. 2019). For the model equations and more parameterizations of the RGO model, readers can refer to Chen and Wu (2011). The model’s spinup run is 10 years, under monthly climatological wind stress forcing of ERA-Interim. Subsequent to the spinup, the control run (CTR) is further integrated with ERA-Interim monthly wind stress from 1979 to 2017. The monthly ULT and upper-layer currents of CTR are used for our analysis and compared with observational data. Sensitivity experiments of the RGO model are used for mechanism investigation and described in section 4.
3. Observed variability of the WEPO circulation in 2016 summer
Figure 3 shows zonal current U evolutions observed by four moorings from August 2014 to December 2016. The strong subsurface eastward flows centered at ~200 m in 2016 boreal summer (“boreal” omitted hereinafter) immediately stands out as the most striking feature during the observation period, indicating the enhancement of the EUC in its formation region (e.g., Tsuchiya et al. 1989; Goodman et al. 2005; Qin et al. 2015; Radenac et al. 2016; Wang et al. 2019). This phenomenon was observed in M2N, M0, and M1S, with the maximum speeds of 90, 67, and 63 cm s−1, respectively. During this season, the eastward current occupies the entire upper 500 m at M2N and the upper 350 m at M0, even though the top 50 m is not covered by the moorings’ ADCP measurements. The surface westward currents of ~20 cm s−1 in climatology (e.g., Wang et al. 2016b; Song et al. 2018b), which largely represent the SEC (see also Fig. 1a), are nearly disappeared. The enhancement was, however, unseen at M3N station, and in fact, the subsurface eastward current was weaker in strength than that at the end of 2015. The different changes at these closely distributed mooring sites indicate a complex spatial structure of the WEPO circulation anomalies in the decaying stage of the 2015–16 El Niño event. This phenomenon is worthy of a thorough investigation. To better quantify the mooring-observed variations, we plot the time series of daily and monthly zonal currents averaged over the upper 350 m of the four CASSON moorings (Fig. 4). M1S, M0, and M2N show evident U enhancements in the 2016 summer. The eastward velocity of M1S, M0, and M2N increased from −3, 25, and 13 cm s−1 in late 2015 to 23, 37, and 59 cm s−1 in 2016 summer [June–August (JJA)], respectively. By contrast, the velocity at M3N decreased from 34 cm s−1 in 2015 winter to 25 cm s−1 in 2016 summer. The changes described above, representing the interannual variations during the 2015–16 super–El Niño (black curves in Fig. 4), are much stronger in amplitude than the intraseasonal fluctuations contained in daily time series (gray curves). These interannual variations were replicated by GODAS data, although showing discrepancies in magnitude (blue curves in Fig. 4).
Time–depth plots of daily zonal current U (m s−1) for August 2014–December 2016 derived from the moorings of (a) M3N, (b) M2N, (c) M0, and (d) M1S; U is smoothed using a 9-day-moving-average filter. For daily data, tick marks on the x axis indicate the first day of each month (hereinafter the same). Also shown are the mean zonal (blue) and meridional (red) velocity profiles derived from the moorings of (e) M3N, (f) M2N, (g) M0, and (h) M1S.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–depth plots of daily zonal current U (m s−1) for August 2014–December 2016 derived from the moorings of (a) M3N, (b) M2N, (c) M0, and (d) M1S; U is smoothed using a 9-day-moving-average filter. For daily data, tick marks on the x axis indicate the first day of each month (hereinafter the same). Also shown are the mean zonal (blue) and meridional (red) velocity profiles derived from the moorings of (e) M3N, (f) M2N, (g) M0, and (h) M1S.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–depth plots of daily zonal current U (m s−1) for August 2014–December 2016 derived from the moorings of (a) M3N, (b) M2N, (c) M0, and (d) M1S; U is smoothed using a 9-day-moving-average filter. For daily data, tick marks on the x axis indicate the first day of each month (hereinafter the same). Also shown are the mean zonal (blue) and meridional (red) velocity profiles derived from the moorings of (e) M3N, (f) M2N, (g) M0, and (h) M1S.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time series of daily ADCP (gray), monthly ADCP (black), and monthly GODAS (blue) zonal current averaged over the upper 350 m U350 (m s−1) at (a) M3N, (b) M2N, (c) M0, and (d) M1S. The period of JJA 2016 is highlighted with gray shading. Monthly data are plotted on the 15th day of each month.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time series of daily ADCP (gray), monthly ADCP (black), and monthly GODAS (blue) zonal current averaged over the upper 350 m U350 (m s−1) at (a) M3N, (b) M2N, (c) M0, and (d) M1S. The period of JJA 2016 is highlighted with gray shading. Monthly data are plotted on the 15th day of each month.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time series of daily ADCP (gray), monthly ADCP (black), and monthly GODAS (blue) zonal current averaged over the upper 350 m U350 (m s−1) at (a) M3N, (b) M2N, (c) M0, and (d) M1S. The period of JJA 2016 is highlighted with gray shading. Monthly data are plotted on the 15th day of each month.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
By removing the monthly climatology at M0 based on CASSON and JAMSTEC mooring records, we are able to obtain monthly anomalies during the 1997–98 and 2015–16 El Niño events (Fig. 5). It is revealed that the eastward current anomalies in JJA 1998 were up to 80 cm s−1 (Fig. 5b) and even stronger than those in JJA 2016 (Fig. 5a). Different from the 2015–16 case, eastward current anomalies in 1997–98 were persistent from 1997 winter to the end of 1998, reaching the maximum magnitude in early autumn of 1998. These results suggest that the enhanced eastward equatorial currents observed in 2016 summer may be a common feature during the decaying stage of El Niño. Acknowledging this, we extract the U and U anomalies during the El Niño events derived from TAO/TRITON and GODAS data at the longitudes of 147°E, 165°E, 170°W, 140°W, and 110°W along the equator (Fig. 6). In addition, we also show GODAS results at 142°E to match M0 for comparison (Fig. 6f). ENSO events are identified using the oceanic Niño index (ONI), which is the 3-month (one season) running mean of SST anomaly in the Niño-3.4 region (5°–5°N, 120°–170°W) relative to a multiple centered 30-yr base period. An El Niño is defined by a period with the ONI exceeding +0.5°C for at least five consecutive overlapping seasons (https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_v5.php). The 1986–87 event, which is followed by another warming event of the 1987–88 winter, is excluded, because there was no decaying stage of El Niño in 1987 as in 2016 summer. Ten events were identified during the 1980–2016 period: 1982–83, 1987–88, 1991–92, 1994–95, 1997–98, 2002–03, 2004–05, 2006–07, 2009–10, and 2015–16 (Fig. 5c). Among them, the 1982–83, 1987–88, and 2015–16 events were not covered by ADCP measurements of TAO/TRITON array but included in the composite of GODAS data.
(a) Time–depth plots of monthly U anomaly (m s−1) over the upper 350 m for (a) January 2015–December 2016 derived from M0 and (b) January 1997–December 1998 derived from the JAMSTEC mooring at the same location. Monthly data are plotted on the tick marks of the x axis (hereinafter the same). (c) Time series of the oceanic Niño index (ONI); the red- and blue-filled parts are greater than 0.5°C and less than −0.5°C, respectively. The red numbers marked the El Niño events used for composite.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
(a) Time–depth plots of monthly U anomaly (m s−1) over the upper 350 m for (a) January 2015–December 2016 derived from M0 and (b) January 1997–December 1998 derived from the JAMSTEC mooring at the same location. Monthly data are plotted on the tick marks of the x axis (hereinafter the same). (c) Time series of the oceanic Niño index (ONI); the red- and blue-filled parts are greater than 0.5°C and less than −0.5°C, respectively. The red numbers marked the El Niño events used for composite.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
(a) Time–depth plots of monthly U anomaly (m s−1) over the upper 350 m for (a) January 2015–December 2016 derived from M0 and (b) January 1997–December 1998 derived from the JAMSTEC mooring at the same location. Monthly data are plotted on the tick marks of the x axis (hereinafter the same). (c) Time series of the oceanic Niño index (ONI); the red- and blue-filled parts are greater than 0.5°C and less than −0.5°C, respectively. The red numbers marked the El Niño events used for composite.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–depth plots of the 2-yr composite U (black contours; m s−1) and U anomaly (color shading) at (a) 147°E, (b) 165°E, (c) 170°W, (d) 140°W, and (e) 110°W along the equator for El Niño events, observed by TAO/TRITON data of 1990–2016 and (f) 142°E, (g) 147°E, (h) 165°E, (i) 170°W, (j) 140°W, and (k) 110°W, based on GODAS data of 1980–2016. For GODAS, 10 events were used to composite during the 1980–2016 period, i.e., the 1982–83, 1987–88, 1991–92, 1994–95, 1997–98, 2002–03, 2004–05, 2006–07, 2009–10, and 2015–16 events. For ADCP, the 1982–83, 1987–88, and 2015–16 events were excluded from the composite.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–depth plots of the 2-yr composite U (black contours; m s−1) and U anomaly (color shading) at (a) 147°E, (b) 165°E, (c) 170°W, (d) 140°W, and (e) 110°W along the equator for El Niño events, observed by TAO/TRITON data of 1990–2016 and (f) 142°E, (g) 147°E, (h) 165°E, (i) 170°W, (j) 140°W, and (k) 110°W, based on GODAS data of 1980–2016. For GODAS, 10 events were used to composite during the 1980–2016 period, i.e., the 1982–83, 1987–88, 1991–92, 1994–95, 1997–98, 2002–03, 2004–05, 2006–07, 2009–10, and 2015–16 events. For ADCP, the 1982–83, 1987–88, and 2015–16 events were excluded from the composite.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–depth plots of the 2-yr composite U (black contours; m s−1) and U anomaly (color shading) at (a) 147°E, (b) 165°E, (c) 170°W, (d) 140°W, and (e) 110°W along the equator for El Niño events, observed by TAO/TRITON data of 1990–2016 and (f) 142°E, (g) 147°E, (h) 165°E, (i) 170°W, (j) 140°W, and (k) 110°W, based on GODAS data of 1980–2016. For GODAS, 10 events were used to composite during the 1980–2016 period, i.e., the 1982–83, 1987–88, 1991–92, 1994–95, 1997–98, 2002–03, 2004–05, 2006–07, 2009–10, and 2015–16 events. For ADCP, the 1982–83, 1987–88, and 2015–16 events were excluded from the composite.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composites based on TAO/TRITON and GODAS are broadly consistent with each other in showing the subsurface eastward current anomalies in the western Pacific (142° and 147°E) during JJA(1) (“1” denotes the year following the El Niño event). Subsequently, these eastward anomalies gradually move to the east and manifest as the enhancement of the EUC in the central and eastern Pacific, which is the typical response of the equatorial ocean to a developing La Niña (e.g., Izumo et al. 2002; Izumo 2005; Qin et al. 2015). Clarke and Zhang (2019) demonstrated that equatorial current anomalies left by a decaying El Niño can modulate SST distribution over the Pacific and thereby facilitate the formation of a La Niña condition. It is discernible that the composite anomalies of GODAS are weaker in magnitude than those of ADCP measurements.
Spatial structures of the variations are also of interest. Figures 7a and 7d show the U anomalies in JJA 2016 derived from the OSCAR surface current data and the GODAS 0–350-m average current data. OSCAR U is noisier and exhibits more details than GODAS U, but the two show broadly consistent spatial structure with eastward current anomalies along the NGI and westward equatorial current anomalies in the ocean interior. This pattern is to some degree in the opposite sign to the JJA climatology at the depth of 250 m, which shows westward currents along the NGI coast characterizing the NGCUC and prevailing eastward currents near the equator featuring the EUC (Fig. 1b). The U anomalies in JJA 2016 generally show a counterclockwise circulation structure straddling the equator, with the westward equatorial flows detouring to the north in the vicinity of CASSON moorings and retroflects back at the western boundary to feed the southeastward flows along the NGI. Under the influence of this anticlockwise anomalous circulation, eastward U anomalies were observed at M1S, M0, and M2N, while westward anomaly was observed at M3N (marked as black dots in Fig. 7). It is obvious that neither OSCAR nor GODAS can precisely represent the anomalies observed by CASSON moorings in JJA 2016 with all details. Specifically, M2N is supposed to be enveloped by eastward U anomalies rather than westward anomalies. In OSCAR, M2N locates near the boundary between eastward and westward flow anomalies. It is possible that the surface current presented by OSCAR slightly deviates from the upper-ocean current observed by the mooring, giving that the upper 50 m was not monitored by ADCPs (Fig. 3). Apparently, it is also difficult for GODAS to realistically reproduce the exact position and detailed structures of this anomalous circulation. GODAS cannot fully represent the impacts of nonlinearity, topography, and ocean internal instabilities that are important in the WEPO region (e.g., Lukas et al. 1996; Wang and Yuan 2012; Hu et al. 2015; Wang et al. 2016a).
The U anomalies in JJA 2016 based on (a) OSCAR surface current data, (d) GODAS 0–350-m average, and (g) upper-layer current of the CTR run of the 1.5-layer RGO model. (b),(e),(h) As in (a), (d), and (g) but for the composite U anomaly during JJA(1) of El Niño events (“1” denotes the year following the El Niño event). GODAS and RGO composite events are the same as the 10 events used by GODAS in Fig. 6. For OSCAR data, seven events were used to composite. (c),(f),(i) The STD of the U anomaly among events during JJA(1) obtained from OSCAR, GODAS, and RGO.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The U anomalies in JJA 2016 based on (a) OSCAR surface current data, (d) GODAS 0–350-m average, and (g) upper-layer current of the CTR run of the 1.5-layer RGO model. (b),(e),(h) As in (a), (d), and (g) but for the composite U anomaly during JJA(1) of El Niño events (“1” denotes the year following the El Niño event). GODAS and RGO composite events are the same as the 10 events used by GODAS in Fig. 6. For OSCAR data, seven events were used to composite. (c),(f),(i) The STD of the U anomaly among events during JJA(1) obtained from OSCAR, GODAS, and RGO.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The U anomalies in JJA 2016 based on (a) OSCAR surface current data, (d) GODAS 0–350-m average, and (g) upper-layer current of the CTR run of the 1.5-layer RGO model. (b),(e),(h) As in (a), (d), and (g) but for the composite U anomaly during JJA(1) of El Niño events (“1” denotes the year following the El Niño event). GODAS and RGO composite events are the same as the 10 events used by GODAS in Fig. 6. For OSCAR data, seven events were used to composite. (c),(f),(i) The STD of the U anomaly among events during JJA(1) obtained from OSCAR, GODAS, and RGO.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The general consensus between OSCAR surface U and GODAS upper-ocean U indicates that the anomalous counterclockwise circulation emerging in JJA 2016 may be largely explained by the ocean’s first baroclinic mode response to wind forcing, as suggested by previous studies for interannual upper-ocean circulation variability of this region (e.g., Kessler 1990; Qiu and Lukas 1996; Yuan et al. 2004; Qiu and Chen 2010; Li et al. 2012; Zhao et al. 2013; Song et al. 2018a). The 1.5-layer nonlinear RGO model can be used to understand this phenomenon. Figure 7g shows the upper-layer U anomaly in JJA 2016 simulated by the CTR of RGO model, which can also to a large degree reproduce the anomalous counterclockwise circulation seen in OSCAR and GODAS. Interestingly, the RGO model shows a better fidelity in representing the direction of anomalous flow than GODAS, and the anomaly magnitudes are also closer to mooring measurements. Figures 7b, 7e, and 7h show the composite U anomalies in JJA(1) of El Niño events, in which the GODAS and RGO model composite events are the same as the 10 events used by GODAS in Fig. 6. Since OSCAR data began in 1992, seven events were used in its composite. These figures are generally consistent in spatial pattern with the anomalies of JJA 2016 (Figs. 7a,d,g), albeit with weaker amplitudes. Because of strong event-by-event variability, composite anomalies show evident differences from the 2016 anomalies observed by the moorings, which can be seen from the patterns of U anomaly in JJA(1) for all El Niño events (no shown) and its standard deviations (STD; Figs. 7c,f,i).
The success of the 1.5-layer RGO model in reproducing the sketch of the anomalous circulation (Figs. 7g,h) indicates the fundamental role played by the wind-driven long-wave dynamics in regulating the general features of the observed variability, and in the following section, we will use the RGO model to explore the underlying dynamical processes. We are also aware that the model has merely captured the basic features of the mooring-observed current variability and missed many specific characteristics such as the vertical structure and time lag between moorings. The intensity of RGO-simulated current anomalies is also much weaker than mooring observations. The discrepancies reflect the impacts from the complex topography, nonlinearity of ocean circulation, and high-order baroclinic modes. The analysis presented in the following section using the RGO model elucidates only the lowest-order dynamics of the WEPO anomalous circulation.
4. Dynamical processes
To characterize surface wind changes associated with El Niño, the 5°S–5°N zonal wind stress τx anomalies of the 2-yr composite El Niño event was plotted in Fig. 8. Since the equatorial circulation anomaly mainly responds to the wind stress forcing (Kessler 1990; Qiu and Lukas 1996; Qiu and Chen 2010), we divide the composite event into three stages according to τx evolution (Gebbie et al. 2007; Hu et al. 2014; Menkes et al. 2014; D. Chen et al. 2015; Fedorov et al. 2015; Fang and Mu 2018). This El Niño stage definition may be different from that classified by Niño index (Clarke 2014), but it is more suitable for studying this phenomenon. The developing stage is characterized by strengthening westerly wind anomalies in the western Pacific. The mature stage features fully developed westerly wind anomalies over the western and central Pacific and weaker easterly wind anomalies in the eastern Pacific. During the decaying stage, easterly wind anomalies emerge in the western Pacific and gradually strengthen, indicating a tendency toward La Niña–like condition over the Pacific basin. Using the composite winds stress anomalies, we perform four 2-yr sensitivity experiments (after spinup) to answer the question as to which stage is more important in causing the observed WEPO variability through wind forcing. These experiments are described in Table 1. For case 0, the RGO model is integrated forward for two years using the composite wind stress anomalies of El Niño (Fig. 8) plus monthly climatology. The result of case 0 in JJA(1) is in accordance with the composite of CTR in JJA(1) (Fig. 9a), showing the well-structured anticlockwise upper-layer current anomalies in the WEPO (Fig. 9b). To examine the effect of wind forcing in different stages, for cases 1, 2, and 3, the El Niño composite wind stress anomalies are exerted only in the developing, mature, and decaying stages, respectively, in addition to climatological winds (Table 1).
Time–longitude plot of 5°S–5°N zonal wind stress τx anomalies for the composite of 10 El Niño events. The gray dashed lines separate the developing [January(0)–August(0)], mature [September(0)–April(1)], and decaying [May(1)–December(1)] periods. The green dashed line denotes the mean longitude of the IOCAS moorings (142°E).
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–longitude plot of 5°S–5°N zonal wind stress τx anomalies for the composite of 10 El Niño events. The gray dashed lines separate the developing [January(0)–August(0)], mature [September(0)–April(1)], and decaying [May(1)–December(1)] periods. The green dashed line denotes the mean longitude of the IOCAS moorings (142°E).
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–longitude plot of 5°S–5°N zonal wind stress τx anomalies for the composite of 10 El Niño events. The gray dashed lines separate the developing [January(0)–August(0)], mature [September(0)–April(1)], and decaying [May(1)–December(1)] periods. The green dashed line denotes the mean longitude of the IOCAS moorings (142°E).
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Summary of the wind stress forcing used in the experiments for cases 0–4.
Upper-layer current anomalies (m s−1; colored vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) derived from (a) the composite of 10 El Niño events of CTR output, (b) case 0, (c) case 1, (d) case 2, and (e) case 3. Black rectangles denote box 1 and box 2. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Upper-layer current anomalies (m s−1; colored vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) derived from (a) the composite of 10 El Niño events of CTR output, (b) case 0, (c) case 1, (d) case 2, and (e) case 3. Black rectangles denote box 1 and box 2. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Upper-layer current anomalies (m s−1; colored vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) derived from (a) the composite of 10 El Niño events of CTR output, (b) case 0, (c) case 1, (d) case 2, and (e) case 3. Black rectangles denote box 1 and box 2. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The results of cases 1–3 are shown in Figs. 9c–e, respectively. Among them, only case 2 can well reproduce the counterclockwise anomalous circulation in the WEPO (Fig. 9d), highlighting the leading role of wind forcing in the El Niño mature stage. The anomalies in case 3 are somewhat weak, as compared with other experiments (Fig. 9e). These results suggest that the observed variability in JJA(1) is not caused by the instantaneous wind forcing but largely the delayed consequences of wind forcing of the El Niño mature stage. In addition, the anomalies in case 2 are in fact stronger than those in case 0. This is likely due to the offsetting effect by the developing-stage wind forcing, as indicated by the clockwise current anomalies in case 1 (Fig. 9c). To facilitate the quantification of anomalous currents, we use two boxes to envelope the equatorial westward current branch (box 1) and the coastal southeastward branch (box 2) of the anomalous circulation, respectively (Fig. 9e). Upper-layer U anomalies of box 1 and box 2 are shown in Fig. 10, and it can be seen that only case 2 can reproduce the negative anomalies (westward) in box 1 and positive anomalies (eastward) in box 2 as in CTR and case 0, confirming the primary role played by the El Niño’s wind forcing of its mature stage in causing the anomalous circulation in the WEPO.
The U anomalies during January(1)–December(1) averaged over (a) box 1 and (b) box 2, derived from the composite of 10 El Niño events of CTR and cases 0–3.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The U anomalies during January(1)–December(1) averaged over (a) box 1 and (b) box 2, derived from the composite of 10 El Niño events of CTR and cases 0–3.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The U anomalies during January(1)–December(1) averaged over (a) box 1 and (b) box 2, derived from the composite of 10 El Niño events of CTR and cases 0–3.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Then we attempt to identify the key region of wind forcing. We first use case 4 to examine the equatorial wind forcing effect, in which El Niño–composite wind stress anomalies are added only to 5°S–5°N. One can see that case 4 can also well reproduce the anomalous circulation in JJA(1) as in case 0 and case 2 (Fig. 11a). Therefore, we can narrow down the key wind forcing region to the equatorial band. To quantify the effects of equatorial wind forcing of different longitudes, we further perform a set of longitudinal experiments in which composite wind stress anomalies are exerted only in a region between the western boundary (100°E) and the longitude X and between 5°S and 5°N, where X changes from 120°E to the eastern boundary of the Pacific basin with an interval of 20°. Note that case 4 is also a member of them with an X reaching the eastern boundary. Figure 11b compares the results of these longitudinal experiments for box 1 and box 2. The counterclockwise anomaly, as indicated by westward anomalies in box 1 and eastward anomalies in box 2, does not emerge until X reaches 160°W. East of 160°W, the current anomalies increase rapidly in magnitude as X approaching the eastern boundary. These results suggest that the key region of the wind forcing is the equatorial central and eastern Pacific. During the mature stage, the central and eastern regions are dominated by westerly and easterly wind anomalies, respectively (Fig. 8). In the following, we attempt to elucidate how these wind changes in the El Niño mature stage drive the WEPO anomalous circulation in the following JJA.
(a) Upper-layer current anomaly (m s−1; colored vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) from case 4. (b) JJA(1) U anomalies of case 4 averaged over box 1 (blue) and box 2 (red), compared with those derived from longitudinal experiments in which wind forcing is exerted over a region from the west boundary (100°E) to a longitude X and from 5°S to 5°N. Black dots in (a) show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
(a) Upper-layer current anomaly (m s−1; colored vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) from case 4. (b) JJA(1) U anomalies of case 4 averaged over box 1 (blue) and box 2 (red), compared with those derived from longitudinal experiments in which wind forcing is exerted over a region from the west boundary (100°E) to a longitude X and from 5°S to 5°N. Black dots in (a) show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
(a) Upper-layer current anomaly (m s−1; colored vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) from case 4. (b) JJA(1) U anomalies of case 4 averaged over box 1 (blue) and box 2 (red), compared with those derived from longitudinal experiments in which wind forcing is exerted over a region from the west boundary (100°E) to a longitude X and from 5°S to 5°N. Black dots in (a) show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite of 10 El Niño events for ULT anomaly (color shading) and current anomaly (m s−1; black vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) derived from case 0. Green dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite of 10 El Niño events for ULT anomaly (color shading) and current anomaly (m s−1; black vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) derived from case 0. Green dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite of 10 El Niño events for ULT anomaly (color shading) and current anomaly (m s−1; black vectors; currents weaker than 0.04 m s−1 are not plotted) of JJA(1) derived from case 0. Green dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The analysis of Fig. 12 underscores the key role played by the positive ULT anomalies in the WEPO, which can be largely explained by the propagation of wind-driven equatorial waves (Fig. 13). The time–longitude plots of ULT anomalies simulated by the RGO model agree well with sea level anomalies from AVISO data, suggesting that the model has well captured the wind-driven long-wave dynamics during El Niño events. In the composite RGO results, the westerly wind anomalies of the equatorial central Pacific during the El Niño mature stage drive equatorial downwelling Kelvin waves that propagate to the east (Fig. 13d) and reflected into equatorial downwelling Rossby waves in both hemispheres at the beginning of the following year (Figs. 13e and 12f). Meanwhile, easterly wind anomalies of the equatorial eastern Pacific during the mature stage also act to drive equatorial downwelling Rossby waves and the downward propagation of Rossby waves that contribute to significantly reducing the reflected signal in the western Pacific. The latter source may be primary, given that in reality reflected equatorial Kelvin waves are not able to reach the western Pacific owing to strong dissipation and instabilities (e.g., Schopf et al. 1981; Fu and Qiu 2002; Qiu et al. 2013). These downwelling Rossby waves propagated to the western Pacific Ocean, causing the high ULT anomalies in the off-equatorial areas of both hemispheres during the decaying stage. These waves, as a typical response to equatorial winds, have been confirmed by previous studies (Boulanger and Menkes 1999; McPhaden and Yu 1999; Clarke et al. 2007; Bosc and Delcroix 2008). Zhang and Clarke (2017) suggested that wind-forced sea level anomalies are closely associated with equatorial current anomalies during El Niño events. The established UTL anomaly distribution leads to the westward current anomalies along the equator from the ocean interior through the equatorial geostrophic balance [Eq. (1)] and the counterclockwise anomalous circulation in the far WEPO with the aid of topographic effect. These processes also highlight active interactions between the WEPO and the central-eastern Pacific: 1) central-eastern Pacific wind forcing in the mature stage of El Niño induce WEPO circulation anomalies in the following summer through westward propagating Rossby waves and 2) subsequently, WEPO circulation anomalies propagate to the central and western basin.
Time–longitude plots of the composite AVISO sea level anomaly (m) at (a) the equator, (b) 5°N, and (c) 5°S. (d)–(f) As in (a)–(c), but for ULT anomalies of CTR. The black contours in (d)–(f) are the composite 5°S–5°N average zonal wind stress anomaly (N m−2) from ERA-Interim data. Dashed lines are as in Fig. 8.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–longitude plots of the composite AVISO sea level anomaly (m) at (a) the equator, (b) 5°N, and (c) 5°S. (d)–(f) As in (a)–(c), but for ULT anomalies of CTR. The black contours in (d)–(f) are the composite 5°S–5°N average zonal wind stress anomaly (N m−2) from ERA-Interim data. Dashed lines are as in Fig. 8.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
Time–longitude plots of the composite AVISO sea level anomaly (m) at (a) the equator, (b) 5°N, and (c) 5°S. (d)–(f) As in (a)–(c), but for ULT anomalies of CTR. The black contours in (d)–(f) are the composite 5°S–5°N average zonal wind stress anomaly (N m−2) from ERA-Interim data. Dashed lines are as in Fig. 8.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The geometry of western boundary in the WEPO, particularly the slanted NGI, is suggested to be critical in regulating the structure of the observed circulation variability. To test this hypothesis, another set of RGO experiments are conducted with the western boundary geometry edited to different degrees (WB1–WB7). In the WB1 experiment, the NGI coast is replaced by a slanted straight line from 1°N, 126°E to 5°S, 148°E points, and the western boundary south of 5°S, 148°E is set to a straight meridional wall (Figs. 14a,b). Using the simplified western boundary, WB1 can still reproduce the structures of ULT and upper-layer U anomalies in JJA(1) that are similar to case 0 (Fig. 12). Subsequently, we examine the effect of the slanted NGI by gradually reducing the tilt of NGI from WB1 to WB7, and the counterclockwise anomalous circulation becomes less discernible. In WB7, the entire western boundary follows a straight meridional line (Figs. 14m,n), and the ULT and zonal current anomalies are nearly asymmetric to the equator, with no anticlockwise circulation structure formed in the WEPO. These experiments clearly show evident impacts of the slanted NGI on the structure of the anomalous circulation. One may notice that, even in WB7, the ULT and U anomalies are not strictly symmetric to the equator, with the maximal westward velocity occurring north of the equator (Fig. 14n). This is likely due to the asymmetry in wind forcing structure. The spatial structure of ENSO winds is regulated by the air–sea interaction due to the north–south asymmetry of the land–sea distribution after the modification. This is confirmed by an additional RGO experiment, in which we used the same western boundary geometry of WB7 and symmetric wind stress anomalies to the equator (by replacing the anomalies of the Southern Hemisphere with those in the Northern Hemisphere) and achieved strictly symmetric anomalies to the equator (figures not shown).
The composite (left) ULT anomalies and (right) U anomalies of JJA(1) for El Niño events simulated by (a),(b) WB1, (c),(d) WB2, (e),(f) WB3, (g),(h) WB4, (i),(j) WB5, (k),(l) WB6, and (m),(n) WB7. The black solid lines in (a) and (b) are the original coasts.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite (left) ULT anomalies and (right) U anomalies of JJA(1) for El Niño events simulated by (a),(b) WB1, (c),(d) WB2, (e),(f) WB3, (g),(h) WB4, (i),(j) WB5, (k),(l) WB6, and (m),(n) WB7. The black solid lines in (a) and (b) are the original coasts.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite (left) ULT anomalies and (right) U anomalies of JJA(1) for El Niño events simulated by (a),(b) WB1, (c),(d) WB2, (e),(f) WB3, (g),(h) WB4, (i),(j) WB5, (k),(l) WB6, and (m),(n) WB7. The black solid lines in (a) and (b) are the original coasts.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The RGO model experiments described above highlight the critical role played by wind-forced equatorial wave dynamics, while the effects of the exchanges with the Indian Ocean through the ITF and the Solomon Sea through the Vitiaz Strait are not properly resolved in the present model settings. To appreciate their possible influences, we performed the Island and Indo-Pacific experiments. By adopting more realistic topography in the southwest tropical Pacific (Fig. 2b), the Island experiment achieves a better representation of summertime flows from the Solomon Sea to the equator (Fig. 15b), as referenced to GODAS data (Fig. 1). Specifically, the equatorward flows appear both in Vitiaz Strait and east of the island of New Britain (Fig. 15c), rather than all occurring along NGI coast in CTR (Fig. 15a). The Indo-Pacific experiment covers the entire Indo-Pacific Oceans and allows interbasin exchange through the Makassar Strait (Fig. 2c), which modifies the circulation pattern near the entrance of the ITF (Fig. 15c). However, composite U anomalies for JJA(1) of the two experiments (Figs. 15d,f) broadly resemble that of CTR (Fig. 15b). The counterclockwise circulation and anomalies at the four mooring sites are roughly consistent among the three runs. These comparisons indicate that the anomalous circulation in the RGO model is generally not sensitive to the tackling of the ITF and WEPO–Solomon Sea exchange.
The composite (a)–(c) U and (d)–(f) U anomalies for JJA(1) for El Niño events simulated by (top) CTR, (middle) the Island experiment, and (bottom) the Indo-Pacific experiment. Gray shading denotes land areas of each experiment. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite (a)–(c) U and (d)–(f) U anomalies for JJA(1) for El Niño events simulated by (top) CTR, (middle) the Island experiment, and (bottom) the Indo-Pacific experiment. Gray shading denotes land areas of each experiment. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite (a)–(c) U and (d)–(f) U anomalies for JJA(1) for El Niño events simulated by (top) CTR, (middle) the Island experiment, and (bottom) the Indo-Pacific experiment. Gray shading denotes land areas of each experiment. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
It is also possible that Indo-Pacific and Island runs are still not able to fully resolve effects of the ITF and exchange at Solomon Sea straits. To confirm the fidelity of RGO runs in representing the wave propagation and reflection processes in the WEPO, we refer to AVISO SLA data. Evolutions of wave signals from March to August of Year(1) are shown in Fig. 16. Positive SLA or ULT anomalies on both sides of the equator first appeared near about 5°S and 5°N of the international date line, and subsequently reach the western boundary and cause equatorial sea level rise near the mooring sites. These processes are in accordance between AVISO data and RGO runs, except for a faster phase speed of Rossby wave in RGO model. Therefore, the RGO model is able to capture the wave dynamics in the WEPO, which is key to the generation of anomalous circulation following El Niño events, and the tackling of the ITF and Solomon Sea straits does not affect the results in a radical manner.
The composite ULT anomalies of (left) March(1)–April(1), (center) May(1)–June(1), and (right) July(1)–August(1) for El Niño events obtained from (a) CTR, (b) the Island experiment, (c) the Indo-Pacific experiment, and (d) AVISO data. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite ULT anomalies of (left) March(1)–April(1), (center) May(1)–June(1), and (right) July(1)–August(1) for El Niño events obtained from (a) CTR, (b) the Island experiment, (c) the Indo-Pacific experiment, and (d) AVISO data. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
The composite ULT anomalies of (left) March(1)–April(1), (center) May(1)–June(1), and (right) July(1)–August(1) for El Niño events obtained from (a) CTR, (b) the Island experiment, (c) the Indo-Pacific experiment, and (d) AVISO data. Black dots show the locations of four subsurface moorings.
Citation: Journal of Physical Oceanography 50, 11; 10.1175/JPO-D-20-0011.1
5. Summary and discussion
The WEPO has complex and highly variable upper-ocean circulation. Because of a shortage of direct observation, the WEPO circulation variability during the El Niño event was rarely explored. CASSON moorings deployed at 2°N, 0°, and 1°S (M2N, M0, and M1S) documented exceptionally strong eastward currents in 2016 summer, the decaying stage of the 2015–16 super–El Niño, showing maximum velocities of 90, 67, and 63 cm s−1, respectively. Another mooring at 3°N (M3N) did not observe similar phenomenon. Analysis of historical ADCP observations and GODAS reanalysis data suggests that such variations also occurred during previous El Niño events in JJA(1). These mooring-observed current anomalies are manifestations of a counterclockwise anomalous circulation straddling the equator, with westward equatorial currents retroflecting near the western boundary and feeding southeastward currents along the NGI. Simulation of a 1.5-layer RGO model was able to crudely capture the basic features of this anomalous circulation, and experiments were performed to understand the wind-driven dynamics on the lowest order.
The results of RGO model experiments suggest that this counterclockwise anomalous circulation is largely the delayed response to equatorial wind forcing over the central-to-eastern Pacific appearing in the mature stage of El Niño. The downwelling Rossby waves, originating from the reflection of equatorial Kelvin waves at the eastern boundary and the generation by easterly wind anomalies in the eastern Pacific, propagate across the Pacific to the WEPO and give rise to two high ULT lobes on both side of the equator. The meridional high–low–high structure of ULT corresponds to westward equatorial currents, conforming to the linear equatorial geostrophic balance relationship. In the WEPO, the southern lobe shifts equatorward as encountering the slanted NGI coast, establishing a high ULT center near the equator at ~142°E. As a result, the westward equatorial currents detour to the north and retroflect to feed the southeastward currents along the NGI. Idealized experiments confirmed the topographic effect in regulating the structure of the anomalous counterclockwise circulation straddling the equator. Results of two additional experiments suggest that the wind-forced equatorial wave dynamics, which is the main driver of this anomalous circulation, is not sensitive to the exchanges through Solomon Sea straits and the ITF.
We are aware that the upper circulation variability in the WEPO in 2016 summer is of complex three-dimensional structure and temporal evolution, as implied by mooring records. These characteristics are regulated by combined effects of topography, nonlinearity, and high-order baroclinic modes near the western boundary. Even ocean reanalysis data of GODAS have difficulties to reproduce these characteristics. Our analysis presented here only addresses the lowest-order dynamics of the WEPO circulation variability, that is, how the wind-driven linear oceanic wave adjustments give rise to a counterclockwise anomalous circulation in the WEPO. Detailed and quantitative characteristics of this anomalous circulation and the event-by-event variability are not explained here. More observational data and model simulations with higher complexity are required to achieve further insights. For instance, high-resolution ocean models can be used to confirm the findings of this study and shed lights on more complicated processes (such as the time lag of the mid-2016 current enhancements from south to north in Fig. 3), although serious difficulties are expected for such efforts. There is still a long way to go for the complete understanding of the interannual variability of this region. With these regards, the lowest-order framework of dynamics proposed in this paper might serve as a good starting point.
Variations in the WEPO also show evident event-by-event differences, such as in intensity, spatial structure, and timing. The 2016 and 1998 events obviously differ from each other (Fig. 5). Here we explore only the general characteristics and mechanisms, that is, the composite results of all events (section 4), which has averaged out many of the specific characteristics of the subsurface moorings observed in 2016. These event-by-event differences to a large extent reflect the diversity of El Niño events. For example, the eastern Pacific and central Pacific El Niño types (e.g., Ashok et al. 2007; Kug et al. 2009; Yeh et al. 2009; Yu et al. 2010; Takahashi et al. 2011; Capotondi et al. 2015; Cai et al. 2018; Timmermann et al. 2018) have been extensively discussed and are likely to exert different signatures on the WEPO circulation (Wang and Wu 2013; Zhao et al. 2013). Contrasting the WEPO variability in different El Niño events and its feedback effects on SST is also an interesting topic. In future research, more observational records should be involved in the study of effects of different types of El Niño events on the upper-ocean circulation of the WEPO such as the ADCP observations of the TAO/TRITON array at 0°, 156°E included two El Niño events, 1991–92 and 1994–95.
It has been seen in this paper that the equatorial current anomalies in WEPO occurring in 2016 summer subsequently transmitted eastward and reached the eastern Pacific by the end of 2016. A strong El Niño is usually followed by a La Niña event in the next year, but the opposite scenario is not frequently observed (McPhaden 1999; Cai et al. 2015). Clarke and Zhang (2019) provided detailed description for such El Niño–La Niña asymmetry by analyzing warm water volume change. We observed a prominent counterclockwise anomalous circulation in the WEPO after El Niño events, but in the summers following La Niña events, clockwise current anomalies are not evident. This asymmetry of the WEPO circulation variability may be an important process of the ENSO asymmetry. Future efforts may be devoted to clarify how the WEPO anomalies propagate eastward (as equatorial Kelvin waves or air–sea interaction) and examine its contribution to SST changes through mixed-layer heat budget analysis, so as to explore its potential role in the establishment of the subsequent La Niña. Another interesting topic is the impact on the source of the EUC. Previous studies have shown that 2/3 of the EUC sources come from the Southern Hemisphere (Blanke and Raynaud 1997; Rodgers et al. 2003; Fukumori et al. 2004; Goodman et al. 2005), while Izumo et al. (2002) suggested that the contributions of the two hemispheres are relatively uniform. The impacts of the WEPO anomalous circulation during El Niño events may play a role in modifying the hemispheric partition of the EUC’s water source and modulate the temperature and salinity variability of the equatorial thermocline.
Acknowledgments
This study is supported by the Strategic Priority Research Program of Chinese Academy of Sciences (XDB42000000), National Natural Science Foundation of China (41730534), the National Program on Global Change and Air-Sea Interaction (GASI-IPOVAI-01-01) and the Key Research Program of Frontier Sciences, CAS (Grant QYZDB-SSW-SYS034). The crew of R/V KEXUE (Science) is thanked for their assistance with the deployment and retrieval of the moorings.
Data availability statement
IOCAS mooring ADCP data are available as online supplementary material of this paper. JAMSTEC mooring ADCP data are available at http://www.jamstec.go.jp/rigc/j/tcvrp/ipocvrt/adcp_data.html. TAO/TRITON data are available at https://www.pmel.noaa.gov/tao/drupal/disdel/. OSCAR data are from https://www.esr.org/research/oscar/oscar-surface-currents/. GODAS data are from http://www.cpc.ncep.noaa.gov/products/GODAS/. ERA-Interim data are from https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era-interim. AVISO data are from http://marine.copernicus.eu/. ONI index data are from https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_v5.php.
REFERENCES
Ashok, K., S. K. Behera, S. A. Rao, H. Weng, and T. Yamagata, 2007: El Niño Modoki and its possible teleconnection. J. Geophys. Res., 112, C11007, https://doi.org/10.1029/2006JC003798.
Behringer, D., and Y. Xue, 2004: Evaluation of the global ocean data assimilation system at NCEP: The Pacific Ocean. Eighth Symp. on Integrated Observing and Assimilation Systems for Atmosphere, Oceans, and Land Surface, Seattle, WA, Amer. Meteor. Soc., 2.3, https://ams.confex.com/ams/84Annual/techprogram/paper_70720.htm.
Blanke, B., and S. Raynaud, 1997: Kinematics of the Pacific Equatorial Undercurrent: An Eulerian and Lagrangian approach from GCM results. J. Phys. Oceanogr., 27, 1038–1053, https://doi.org/10.1175/1520-0485(1997)027<1038:KOTPEU>2.0.CO;2.
Bonjean, F., and G. S. E. Lagerloef, 2002: Diagnostic model and analysis of the surface currents in the Tropical Pacific Ocean. J. Phys. Oceanogr., 32, 2938–2954, https://doi.org/10.1175/1520-0485(2002)032<2938:DMAAOT>2.0.CO;2.
Bosc, C., and T. Delcroix, 2008: Observed equatorial Rossby waves and ENSO-related warm water volume changes in the equatorial Pacific Ocean. J. Geophys. Res., 113, C06003, https://doi.org/10.1029/2007JC004613.
Boulanger, J.-P., and C. Menkes, 1999: Long equatorial wave reflection in the Pacific Ocean from TOPEX/POSEIDON data during the 1992-1998 period. Climate Dyn., 15, 205–225, https://doi.org/10.1007/s003820050277.