Discovery of a New Inter‐Basin Climate Pattern: Australia Boundary Current Dipole

A new inter‐basin teleconnection analysis method is proposed to comprehensively investigate the connection among global oceans. The strong negative correlation (−0.59) between sea surface temperature (SST) along Leeuwin Current (LC) and East Australia Current (EAC) is discovered and named as Australia Boundary Current Dipole (ABCD). This new inter‐basin air‐sea coupled mode exhibits remarkable interannual variability, with a peak phase in austral summer. The related SST anomalies are mainly generated by anomalous Australia High induced opposite alongshore surface winds to enhance/suppress the heat transport of the two warm currents, in turn, reinforce the wind anomalies by heating/cooling the overlying atmosphere and changing regional pressure anomalies. The meridional ocean advections control the anomalous SST, indicating the essential role of coastal Bjerknes feedback. The kinetic energy analysis further demonstrates that ABCD represents the reverse variation pattern of LC and EAC.

2 of 10 throughflow and Tasman Leakage respectively, are affected by basin-scale winds and impact the surrounding region through local air-sea interaction (Hill et al., 2008;Shinoda et al., 2020;S. Wang et al., 2021). Our IBT results may reveal the strong connection between EAC and LC, which may jointly regulate the heat and material transport between the Indian and Pacific Oceans. We name the SST dipole pattern as the Australia Boundary Current Dipole (ABCD) and hypothesize it as a unique inter-basin mode. Identifying and depicting this unique dipole pattern with an appropriate definition is the main motivation of this paper.

Data
The SST datasets used are the improved Extended Reconstructed SST version 5 (ERSST v5), the Centennial in situ Observation-Based Estimates SST (COBE SST) during 1950-2020 (Huang et al., 2017;Ishii et al., 2005), and the Optimum Interpolation SST analysis version 2 (OISST) during 1982-2020 (Reynolds et al., 2002). The monthly surface wind and sea level pressure for the period 1950-2020 are derived from the National Centers for Environmental Prediction reanalysis-1 (NCEP Reanalysis 1) data set (Kalnay et al., 1996). The three-dimensional ocean circulation and ocean subsurface temperature from the Simple Ocean Data Assimilation system (SODA 2.2.4) during 1979-2010 and Global Ocean Data Assimilation System (GODAS) during 1980-2015 are used for understanding the oceanic processes (Behringer & Xue, 2004;Carton & Giese, 2008). The sea surface height (SSH) is derived from GODAS. The Objectively Analyzed air-sea Fluxes version 3 (OAflux) data set for 1984(L. Yu et al., 2008 is employed to investigate atmosphere-ocean heat exchange. All surface heat flux components are positive downward. To calculate kinetic energy (KE) of ocean currents around Australia, we used global ocean reanalysis of altimeter data derived from Australia Integrated Marine Observing System AODN data set for 1993-2017 (IMOS-OceanCurrent-Gridded sea level anomaly-Delayed mode). The linear trend and long-term mean climatology have been removed from each data set.
The Ningaloo Niño/Niña index (NNI) is calculated from the SST anomalies (SSTAs) are averaged over 108-115°E, 28-22°S and the definition of the Ningaloo Niño/Niña year is based on the preview research (Table  S1 in Supporting Information S1) (Kataoka et al., 2014). The ENSO is represented by the Niño 3.4 and Niño 4 index and Indian Ocean Dipole (IOD) is represented by the Dipole Mode Index, which are obtained from NOAA Physical Sciences Laboratory.

Methodology
To find out the inter-basin teleconnection between one specific ocean and the other oceans, we develop a new IBT analysis based on the Teleconnectivity analysis (Wallace and Gutzler., 1981). All the grid points in one ocean basin are classified into Set A, and all the grid points in the other oceans are classified into Set B. After that, the cross-set correlation coefficients R ab (base point a in Set A, and every grid point b in Set B) between the grid points in Set A and all the grid points in Set B are calculated. Since the inter-basin connections could be characterized by negative and positive correlations, we could study the negative inter-basin teleconnectivity (nIBT) and positive inter-basin teleconnectivity (pIBT) based on the positive and negative orientation of R ab . Comparing the strongest positive or negative correlation on cross-set one-point correlation maps of different grids in the Set A contrast the strengths of teleconnection patterns in Set A. An IBT is defined as: Moreover, IBT not only considers simultaneous inter-basin connections, but also takes the lagged relationships into account. The lead-lag IBT (LIBT) is defined as: simultaneous correlation, when = 0 lead correlation ( leads in timesteps), when 0 lag correlation ( lags in timesteps), when 0 In this study, we focused on simultaneous nIBT. Repeat this technique continuously to obtain the strongest negative teleconnection pattern of the three different oceans. This technique presents the geographical distribution of coherent structures of multiple cross-set one-point correlation patterns in a single map ( Figure S1 in Supporting Information S1).
Following previous studies (Kug et al., 2009;Y. Zhang et al., 2020), the heat budget analysis is applied to determine the contributions of the oceanic and atmospheric processes to the ABCD. These contributions are expressed as following equation: where overbars and primes present the monthly mean climatology and deviations from climatology, respectively. The variables T, u(U), v(V), w(W), Q' and Res denote the mix-layer temperature, zonal current, meridional current, vertical current, thermal forcing, and residual terms, respectively. And ′ net , C p , ρ, and H represent the net surface heat flux, the specific heat of sea water, the density of sea water, and the mixed layer depth, respectively. The criterion of H selected in this study is a threshold value of temperature from a near-surface value at 10 m depth (ΔT = 0.1°C) (Sprintall & Roemmich, 1999 where, u and v were the seasonally mean of zonal and meridional geostrophic current components respectively (Caballero et al., 2008).

The Identification of ABCD and Its Air-Sea Coupled Variation
Based on IBT, the strongest teleconnectivity patterns among different oceans in the austral summer are identified ( Figure S1 in Supporting Information S1). In this study, we focus on the strong negative SST correlation between the west and east coast regions of Australia. As Figure 1a shows, SST in the boxed regions (114-130°E, 30-44°S; 150-158°E, 20-30°S) are strongly negative correlated; and their correlation coefficient is −0.56, that is significant at the 95% confidence level (effective number of degrees of freedom is used; Bretherton et al., 1999;Li et al., 2013). Similar significant correlation results also are obtained from the OISST V2 (−0.49) and COBE SST (−0.39) datasets, further demonstrating the robustness of their negative correlation. This inter-basin teleconnection pattern is shown as the warming in the east boundary current region of South Indian Ocean by colder SSTAs in the west boundary current region of South Pacific Ocean ( Figure 1a). Due to the two regions are located off the Australian coasts, we named it the ABCD and define the index (ABCDI) as follows: where the WPI and EPI denote the area-averaged SSTAs over west pole (114-130°E, 30-44°S) and east pole of ABCD (150-158°E, 20-30°S), respectively ( Figure 1a). This ABCD-like SST pattern also can be extracted as the leading Empirical Orthogonal Function mode of SSTAs near Australia, which explains 26% of the total variance ( Figure S2 in Supporting Information S1). As for the ABCD seasonal variability, the seasonal correlation shows the significant negative correlation exists nearly every season, especially the austral summer from December to February (DJF), which displays the strongest correlation season ( Figure S3 in Supporting Information S1). The seasonally stratified standard deviation (Text S1 in Supporting Information S1) of ABCDI and its general seasonal evolution further indicate ABCD is phase locked to the annual cycle, with the largest interannual variability in austral summer (DJF) (Figure 1c; Figure S4 in Supporting Information S1). On this basis, we concentrate our analysis on the DJF ABCD variability and define the ABCD event as the years in which the DJF ABCDI exceeds one standard deviation over the entire analysis period (Table S2 in Supporting Information S1). Figure 1b depicts the time series of the normalized DJF ABCDI, and the power spectrum analysis demonstrates that its possible cycle is 6-year ( Figure S5 in Supporting Information S1).
The coevolution of atmospheric circulation and SSTAs are derived by calculating lead-lag regressions onto the DJF ABCDI. As depicted in Figure 2, the atmospheric anomalies act as the precursor 3-4 months prior to the peak of ABCD. At the early stage of the ABCD, an anomalous cyclone-anticyclone pair emerged over the south Indian-Pacific Oceans (Figure 2a). This cyclone-anticyclone pair induced southward alongshore surface wind anomalies on the west coast and northward alongshore surface wind anomalies on the east coast of Australia. After that, the cyclonic anomalies over the South Indian Ocean disappeared quickly. The high sea level pressure anomalies and associated anticyclonic anomalies further intensified and moved northward to Australia. This atmospheric precursor is associated with the variability of the Australia High. The anomalous anticyclone corresponds to southward surface wind anomalies on the west coast, which lead to the weakening of offshore Ekman transport and the increasing SSH ( Figure S6 in Supporting Information S1). The positive anomalies in SSH correspond to anomalous southward ocean currents along the west coast, which increase the poleward advection of warm water by the LC. While the anomalous northward surface wind anomalies on the east coast strengthen the regional Ekman transport and reduce EAC transportation ( Figure S6 in Supporting Information S1). Once the coastal ocean is anomalously warmed/cooled through oceanic processes, it starts heating/cooling the overlying atmosphere and induces baroclinic responses over the ocean with negative/positive geopotential height anomalies in the lower troposphere (Xue et al., 2020). This further strengthens the alongshore wind anomalies in turn. Consequently, the initial SSTAs grow and develop. Figure S7 in Supporting Information S1 shows the SSTAs of west pole and east pole evolved synchronously. Moreover, leadlag correlations of DJF ABCDI with the alongshore wind index indicate the alongshore surface winds in the two regions coevolve with the SSTAs and reach their peak phase in DJF. The coevolution of these variables demonstrates the coastal Bjerknes feedback plays an essential role in driving the development of ABCD.

Atmospheric and Oceanic Processes Related to the ABCD
Atmospheric and oceanic processes play an important role in the formation of ABCD events. To determine the relative importance of various processes that contribute to the ABCD evolution, we analyzed the heat budget in detail over the two nearshore regions (the boxes in Figure 1a). Figures 3a and 3b display heat budget analysis over the two nearshore regions. Over the west pole of ABCD, during the beginning phase (4-6 months before the peak), the ocean zonal advection favors the warming temperature, which is counteracted by the ocean meridional process. Thus, the surface heating contributes to the initial SST warming. This surface heating is attributed to sensible heat flux (SHF), and latent heat flux (LHF) (Figure 3c). In the beginning phase, southward surface wind anomalies are opposite the climatological northward surface wind over the west pole ( Figures S8 and S9 in Supporting Information S1). Accompanying the initial decrease in surface wind speeds, the evaporation over the west pole was suppressed, which led to increased LHF and thus the warm SSTAs. During the developing and mature phase, the tendency of SST is highly positive and reaches a maximum 1 month before the peak. The ocean meridional advection and surface heating are two major contributors to the positive SST tendency (Figure 3a). Significantly, the positive shortwave radiation (SW) become to dominate the surface heating ( Figure S10 in Supporting Information S1). This is due to the anomalous Australian High promoting the downward SW ( Figure 2e). As for the east pole, the ocean meridional advection is the primary contributor to the negative SST tendency during the beginning and mature phases (Figure 1b). Also, the decreased LHF helps to cool the SST when northward surface anomalies accelerate local wind speed during austral spring and summer (Figure 3d; Figures S8, S9, and S10 in Supporting Information S1). Increased winds enhance evaporation, resulting in the cooling SSTAs. This implies wind-evaporation-SST mechanism occurs in this region and is crucial for the ABCD developing (Xie & Philander, 1994).
Since the heat budget analysis from GODAS data also confirms that ocean meridional advection is the primary contributor to the growth of ABCD ( Figure S11 in Supporting Information S1), Figures 3e and 3f further decompose the ocean meridional advection into three components: the anomalous advection of the anomalous temperature by the mean meridional current (−v∂T′/∂y, anomalous temperature effect), the anomalous advection of the mean temperature by the anomalous meridional current (−v′∂T/∂y, anomalous current effect), and the nonlinear advection of the anomalous temperature by the anomalous meridional current (−v'∂T'/∂y, nonlinear effect). Over the west pole, during the developing and mature phases, the anomalous current effect (−v′∂T/∂y) has a larger magnitude than the anomalous total meridional advection and varies synchronously with the SST tendency, indicating LC carrying a large amount of warm seawater is crucial to the development of ABCD. As for the east pole, anomalous current effect (−v′∂T/∂y) induces initial SST cooling in the beginning phase. In the mature phase, the three components have a combined effect to the anomalous temperature. This result also indicates the SST development of east pole is related to the heat transport by EAC. Both LC and EAC are important contributors to the development of ABCD, which further verifies that coastal Bjerknes feedback is the most significant mechanism for ABCD growth. The anomalous anticyclone induced opposing surface wind anomalies alter the heat transport of LC and EAC, and further lead to the ABCD-like SST pattern in the Pacific and Indian Oceans.
The ABCD reaches its peak in austral summer and gradually decays in subsequent seasons. As the anomalous anticyclone dissipates, the SW flux over the west pole decreases, and the LHF acts as a damping factor for the formation of ABCD. Moreover, the colder SSTAs in the east pole induce less longwave radiation (LW) to be released to the atmosphere, thus damping the cold SST further (Figure 3d). Various mechanisms take the lead in different phases, indicating that ABCD is a complicated air-sea coupled phenomenon.

The Relation Between LC and EAC
There are two strong southward flowing boundary currents along the west and east coasts of Australia (Figure 4a), which have a profound impact on the heat and material transport in the southern hemisphere (Sprintall et al., 1995;Su et al., 2019). Above analysis shows that the evolution of the ABCD is related to heat transport of LC and EAC. To determine whether these two currents display the same opposite variability features as the west and east pole SSTAs, we compare their KE based on the ABCD events. We calculate the mean KE in the two boxed regions in Figure 4a to represent LC (113-116°E, 30-36°S) and EAC (150-156°E, 20-30°S). In the ABCD positive phases, mean KE of LC and EAC are 3.0 cm/s and 7.1 cm/s, respectively. In the ABCD negative phases, the mean KE of LC decreases to 1.5 cm/s, while mean KE of EAC increases to 8.2 cm/s. Moreover, the correlation coefficient between the meridional velocity of these two boundary currents is −0.53 that is significant at the 95% confidence level (Figure 4b). This result further demonstrates that LC and EAC change oppositely in the austral summer. Referring to the ABCDI definition, the Boundary Current Dipole index (BCDI) is defined as the difference between the meridional velocity anomalies (va) averaged over the two boxed regions in Figure 4a: The BCDI and ABCDI change simultaneously and their correlation coefficient reaches 0.44 that is significant at the 95% confidence level, indicating ABCDI can well represent the reverse variation pattern of the two currents ( Figure 4c).
As discussed above, the ABCD precursor in the atmosphere is the anticyclonic anomalies over the Australia. The positive sea level pressure anomalies induce the opposite surface wind anomalies to modulate LC and EAC transports simultaneously. Changes in the LC and EAC will have additional effects on the meridional current heat advection at the two poles, resulting in ABCD-like SST patterns. Hence, ABCD is the result of both oceanic and atmospheric processes.

Discussion and Summary
In this study, we propose a new method IBT for comprehensively examining the connection between different oceans, which takes every grid point into account, regardless of its relationship to the major climate variability in that specific ocean. Based on IBT results, the strong negative correlation between SST off the west and east coasts of Australia is revealed. The west coast of Australia was warmer than normal, but it was accompanied by colder SSTAs to the east coast of Australia. We name the phenomenon as ABCD.
As the subtropical SST variability near Australia, ABCD seems to be related to Ningaloo Niño/Niña that dominates in the west coast of Australia (Figures S12 and S13 in Supporting Information S1) (Feng et al., 2013). However, careful analyses demonstrate the SSTAs around Australia include those independent of Ningaloo Niño/ Niña. The spatial characteristics and occurrence of ABCD and Ningaloo Niño/Niña manifest their substantial difference. Ningaloo Niño, as a typical coastal climate mode, is characterized by warm SSTAs off the northwest coast of Australia and no substantial SSTA variations along EAC. However, ABCD is characterized by positive (negative) SSTAs along the two warm currents, showing an inter-basin dipole pattern. In addition, more than half of ABCD events occur without preceding or simultaneous Ningaloo Niño/Niña events. More importantly, their developing mechanisms are distinct. Previous studies show the Ningaloo Niño/Niña development is related to ENSO events that induced low-level cyclonic wind anomalies over the central Indian Ocean and LC variations (Feng et al., 2013;Kataoka et al., 2014;Marshall et al., 2015;L. Zhang & Han, 2018). However, the ABCD seems unrelated to ENSO and IOD signal (Figures S14, S15, and S16 in Supporting Information S1). It is triggered by Australia High induced anti-cyclonic wind anomalies. The opposite alongshore winds change local evaporation and two warm currents heat transport to initial SSTAs develop in the two regions. The mixed-layer heat budget analyses show the anomalous coastal SST warming/cooling in the two regions are mainly determined by the LHF, SW, and the anomalous meridional ocean advection. The wind-evaporation-SST feedback and coastal Bjerkenes feedback amplified the ABCD signal. As an inter-basin climate pattern, ABCD is strongly correlated to WPI and EPI ( Figure S17 in Supporting Information S1). However, Ningaloo Niño/Niña is not significantly correlated with EPI. It was demonstrated that the significant correlation between Ningaloo Niño/Niña and ABCD may be attributed to the overlap of their domains off the west coast of Australia. Moreover, the inter-basin relationship between the two poles of ABCD is not affected by Ningaloo Niño/Niña signal according to the partial correlation analysis (−0.53 is significant at the 95% confidence level). To sum up, ABCD is a new inter-basin pattern independent of Ningaloo Niño/Niña.
The anomalous anticyclone above Australia appears to be a significant influence in the initiation of ABCD events. Nonetheless, its emergence and sustenance remain unanswered questions. Due to ABCD's independence from the main tropical climate variability in the Pacific and Indian Oceans, the anomalous anticyclone may serve as the key factor in adjusting the time scale of ABCD. More investigation is required. In addition, understanding ABCD has immediate implications for human lives and regional marine ecosystems because it reflects the reverse variation pattern of the two major boundary currents in the Southern Hemisphere, which transport heat from the tropics to the extratropics, and their temperature variations may affect climate change in the Southern Hemisphere (Shi et al., 2008;Sprintall et al., 1995;Su et al., 2019). The ABCD, as a climate mode in the subtropical Indian-Pacific Oceans, may exert significant and independent impacts on regional and global climate variability, for example, the surface temperature and precipitation over Australia, the local coral growth, and the surface temperature in South America (not shown). The underlying physical mechanism will also be investigated in depth in the future.

Data Availability Statement
The atmospheric and oceanic datasets used in this study include: