Interannual relationship between the Asian–Pacific Oscillation and summer sea surface temperature in the North Atlantic

Over the last two decades, there has been increasing interest in investigating the connection between the Asian-Pacific Oscillation (APO) and weather and climate on regional and global scales, but the impacts of the APO on sea surface temperature (SST) remains unclear. Using the multisource reanalysis dataset and observed SST data, we evaluated the interannual relationship between the APO and SST in the North Atlantic (NASST) during the period 1979–2016. The results show that there exists a statistically significant positive interannual relationship between APO and NASST and this connection can be attributed to the Rossby wave train that originates in Asia and propagates to Europe, which is triggered by the APO forcing. Further examination revealed that the cloud radiation, air–sea heat exchange and oceanic dynamic process induced by APO are crucial in modulating the interannual variability of the NASST. Additionally, the numerical simulation results from the linear baroclinic model also provide additional evidence for this linkage.


Introduction
The interannual variability of sea surface temperature (SST) in the North Atlantic (NASST), which is often characterized by the tripole and horseshoe SST anomaly pattern (figures S1(a) and (b)), substantially impacts weather and climate anomalies over both nearby and remote regions (e.g.Cayan 1992, Czaja and Frankignoul 2002, Zuo et al 2013, Chen et al 2019, Han et al 2022).Investigating the factors and associated mechanism that influence the interannual variability of NASST under the context of global warming has therefore always been a challenging topic in ocean and climate research over the past few decades.
It is widely recognized that air-sea-land interaction processes plays an important role in modulating the interannual variability of NASST (Cayan 1992, Deser andTimlin 1997).For example, previous studies have clarified that the interannual variability of NASST is found to be closely linked to the zonal wind stress forcing and atmospheric fluctuations (Czaja and Frankignoul 2002, Richter et al 2013, Groth et al 2017).Furthermore, the Saharan dust could arouse an anomaly in surface shortwave radiation, via altering the aerosol optical depth and air-sea heat exchange around North Atlantic and hence moderate the interannual variability of NASST (Foltz andMcPhaden 2008a, 2008b).Evan et al (2011) also found that the heat content of the tropical Atlantic ocean is directly related to land-use change induced dust emissions over West Africa on interannual to decadal timescales.In addition, the interaction between Atlantic SST and other oceans such as Pacific (Choi et al 2019, Sun et al 2021) and Indian Ocean (Yang et al 2022) also exerts a significant impact on atmospheric and oceanic teleconnections and governs the variability in the SST on a wide range of spatial and temporal scales.
As one of the basic forms of atmospheric circulation, atmospheric teleconnection is crucial in modulating Atlantic SST modes (e.g. Marshall et al 2001, Alexander et al 2002, Pan 2005, Kwon et al 2010, Linderholm et al 2011).For example, Robertson et al (2000) showed that the North Atlantic Oscillation (NAO) correlates positively with SST over the tropical and subtropical South Atlantic on an interannual scale.Jing et al (2020) evaluated the response of winter NASST to the simultaneous NAO in Coupled Model Intercomparison Project Phase 5 models and found that basically all models examined are able to reasonably simulate the tripole pattern of the response of the NASST to the winter NAO on an interannual scale.Tao et al (2023) reported that the strong North Atlantic tripole SST events, which is primarily driven by the NAO during early winter and peaking in January, could continue to actively influence the atmosphere through diabatic heat and transient eddy forcing untill late winter.Furthermore, being different from tripole SST anomalies, the formation of summer horseshoe-shaped SST over North Atlantic is largely forced by the anomalous tropical convection associated with displacement of intertropical convergence zone (Cassou et al 2005).In addition, atmospheric teleconnection over North Atlantic also exerts strong impacts on climate variability at regional to global scales, particularly including precipitation, temperature and monsoon over the regions around the East Asia through the Rossby wave propagation mechanism (Linderholm et al 2011, Bollasina and Messori 2018, Liu et al 2021).
The Asian-Pacific Oscillation (APO) is a dipole pattern over Asia and the North Pacific region during the summer, characterized by a zonal seesaw on various time scales (Zhao et al 2007, figure S1(c)).A positive phase of APO denotes a positive uppertropospheric eddy temperature over the Asian continent and anomalies with opposite signs over the North Pacific, with opposite patterns for negative APO values.Recently, considerable progress has been made in understanding the influence of APO on weather and climate worldwide (e.g.Zhao et al 2007, 2011, 2012a, Zhou et al 2008, Liu et al 2011, Hua et al 2019, Lin et al 2019, 2021), and many studies have clarified that APO could exert significant impacts on interannual variability of SST, particularly the El Niño-Southern Oscillation (ENSO) phenomenon (Zhao et al 2010).Despite the significant progress in understanding the interannual connection between the APO and Pacific SST, little is known about whether the interannual variability of SST in other oceans can be influenced by APO.For example, does the APO impacts interannual variability of NASST?If this is the case, what is the underlying physical process behind this connection?
We aimed to investigate the linkage between the summer APO and NASST on an interannual scale, as well as the underlying mechanisms, through the utilization of statistical analysis and numerical experiments.

Data
The monthly reanalysis data used include the fifth version of the European Centre for Medium-Range Weather Forecasts reanalysis (ERA5) (Hersbach et al 2020) and the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) (Kalnay et al 1996).The monthly SST data used in this study was the National Oceanic Atmospheric Administration (NOAA) Extended Reconstructed SST V5b data (ERSST.v5v)(Huang et al 2017).
The NAO index (NAOI) was obtained from the Climate Prediction Center of the NOAA (Van den Dool et al 2000, Chen and Van den Dool 2003).Following previous study of Zhao et al (2007), the APO index (APOI) was defined as the summer mean upper-tropospheric (500-200 hPa) eddy air temperature difference between Asia (15 in which T is the air temperature and T denotes the zonal mean of T.

Methods
Empirical orthogonal function (EOF) analysis was utilized on the summer NASST to extract its principal mode.The T-N wave activity flux and stream functions were employed to diagnose the wave propagation excited by atmospheric oscillation (Takaya and Nakamura 2001).The T-N wave activity flux is an improvement of the stationary wave activity proposed by Plumb (1985), with its meridional component stronger than that of the Plumb wave activity flux, which enables a better representation of the Rossby long-wave disturbances with larger amplitude in the non-uniform westerly jet.The formula is as follows: (1) The method developed by An (2003) was adopted to eliminate the influence of the NAO according to the following formula: (2) The linear trend of all data was removed before analysis to highlight the interannual variability.All data spanned 1979-2016 unless otherwise noted, and the summer refers to June to August.Two-tailed Student's t test was adopted unless otherwise stated.

Model
The dry version of linear baroclinic model, version 2.2 (LBM, Watanabe and Kimoto 2000) was used to investigate the influence of the APO on atmospheric circulation.This model was developed to expand understanding of the complicated sequence of feedback in the dynamical atmosphere, by removing nonlinearity in their processes.The dynamic framework of LBM was based on a dynamical core of atmospheric general circulation model cooperatively developed at the Center for Climate System Research, University of Tokyo, and the National Institute for Environmental Studies, Japan.The resolution of the model version used in this study is T42 triangular truncation in horizontal and 20 layers in vertical sigma coordinate.The sensitivity experiment was run for 30 model days, where the outputs of the last 15 model days were averaged as the steady response to the prescribed heat forcing.The dataset used for the basic pattern of summer climatology in the model was determined from the NCEP/NCAR reanalysis I data from 1979 to 2016 (Kalnay et al 1996).Additional details are given in section 3.2.

The interannual relationship between APO and NASST
The correlation coefficient between the summer APOI and simultaneous NASST exhibited a clear horseshoe pattern, with significant positive (negative) correlation coefficient occurs over the midlatitude (subtropical) North Atlantic, suggesting a close interannual linkage between the two variables (figure 1(a)).The results from a composite analysis between strong and weak APO cases also revealed the similar results (figure S3(a)).To further address this linkage, the NASST index (NASSTI) was defined as SST anomalies averaged in key area 1 (38 Considering the significant influence of the NAO on the interannual variability of NASST (Deser andTimlin 1997, Marshall et al 2001), we hence removed the simultaneous summer NAO signal from the SST dataset and calculated the correlation pattern again.As shown in figure 1(b), the result also presented a similar pattern between APO and NAO-independent NASST, indicating that the NASST could be affected by atmospheric teleconnection such as APO.It is interesting that the spatial correlation coefficient between figure 1(b) and North Atlantic horseshoe SST pattern (figure S1(b)) was 0.77, suggesting a potential influence of the APO in regulating the interannual variability of the second EOF mode of the NASST.Similar correlation patterns discussed above are also observed in the NCEP/NCAR reanalysis (figure S5).In order to confirm the above results, we further examine the correlation coefficients  between the APOI and NASSTI before and after removing the preceding spring and simultaneous summer NAO signal (table 1).The correlation coefficients are 0.41 (0.40), 0.52 (0.48), and 0.60 (0.49) for April to June (AMJ), May to July (MJJ), and June to August (JJA), respectively, and all passed the significance test at the 99% confidence level, indicating that APO seems independent of NAO in influencing the North Atlantic horseshoe SST pattern.

Possible physical mechanism
To understand the underlying processes that link APO to the NASST, we performed regression analysis of atmospheric circulation on APOI and NASSTI (figure 2).The APO-induced positive anomalous stream function appeared over Asia, the North Pacific and midwestern North America to the North Atlantic, and negative anomalies occurred over the western coast of North America, with wave activity flux emanating from East Asia to the North Atlantic through the North Pacific and America (figure 2(a)).The NASST-related regression pattern showed similar results, which demonstrates the key role of the APOtriggered wave train in driving the interannual variability of the NASST (figure 2(e)).Furthermore, the APO-triggered wave activity flux convergences/divergences along East Asia to the North Atlantic were accompanied by positive geopotential height anomalies appearing over East Asia, the North Pacific, and midwestern North America to the North Atlantic areas and anomalous negative geopotential heights occurring over the western coast of North America (figure 2(b)).Correspondingly, in association with a stronger APO, significant anticyclonic circulation anomalies appeared over the Pacific and North Atlantic, and pronounced anomalous cyclonic circulation were mainly seen in the polar region and North America at low level (figure 2(c)).Notably, the NASST-related geopotential height and wind anomalies in the mid-and low-levels (figures 2(f) and (g)) were found to be similar to those shown in figures 2(b) and (c).This close resemblance supported the abovementioned conclusion of a significant interannual relationship between the APO and NASST, and we therefore hypothesized that the APO may significantly influence the interannual variability in the NASST by triggering a quasistationary wave train and related atmospheric circulation anomalies.An elliptic heat forcing (centered at 32 • N, 82 • E) over the Tibetan Plateau (TP) (figure S2(a)) was prescribed in the LBM to verify the interannual linkage between the APO and NASST because the formation and interannual variability of the APO is closely associated with the thermal condition over the TP (Liu et al 2017).The maximum heating rate was set to 1 K d −1 and the vertical heating profile was configured as a sine function, with the maximum heating centered around 0.4 sigma level (approximately 400 hPa) (figure S2(b)).Figure 2(d) shows the response of 500 hPa geopotential height anomalies to diabatic heating forcing of the APO.The diabatic heating over the TP excited a wave train emitted from the TP toward the North Pacific, North America and North Atlantic, with increased geopotential height over Asia, North Pacific, North America and North Atlantic and decreased geopotential height over the eastern and western North Pacific, which exhibits a basically similar pattern to the observations.The APO-triggered atmospheric responses at low-level (figure 2(h)) are also generally similar to the regression pattern for APOI (figure 2(c)) and NASSTI (figure 2(g)) and the same results were also found in NCEP/NCAR reanalysis (figure S6).However, the APO-related anticyclonic circulation center simulated by the LBM model tends to be positioned to the south, compared to the regression pattern for APOI.This discrepancy may be attributed to variations in the level of the heating center and the intensity of heating rates across different studies (Zhu and Ren 2019).
Previous studies have shown that anomalous atmospheric circulation affect the ocean by thermal and dynamic processes (Cayan 1992, Deser andTimlin 1997).A question arises here: how does APOdriven anomalous circulation influence the NASST?To answer this question, we performed a regression analysis of the total cloud cover, net shortwave radiation, surface net heat flux (sum of surface sensible heat and latent heat), and zonal wind stress against the APOI and NASSTI.From the perspective of cloud radiation forcing, when the APO was stronger than normal, the total cloudiness decreased in the mid-latitudes and increased in the subtropical North Atlantic (figure 3(a)) due to anomalous high-pressure circulation, thereby leading to enhanced shortwave radiation over the mid-latitudes and reduced shortwave radiation over the subtropical North Atlantic (figure 3(b)), which favored increased (decreased) SSTs over the region and then induced a horseshoe SST pattern.
Furthermore, the sensible and latent heat fluxes greatly influences the SST anomalies.Figures 3(c) and (g) present the regressions of the surface net heat flux against APOI and NASSTI.Figure 3(c) shows that significant negative surface net heat fluxes were present in the subtropical North Atlantic, while positive values were observed in the mid-latitude North Atlantic.Specifically, such a pattern could be attributed to the anomalous anticyclonic circulation in the North Atlantic.changes in the surface heat flux in the Northern Ocean are greatly influenced by variations in wind directions (Bunker and Worthington 1976, Cayan 1992, Zhou et al 2010).Therefore, when the APO was stronger, anomalous northeasterly winds over the subtropical North Atlantic in response to cold and dry advection from north to south should be expected, which favors that the atmosphere gains heat from the ocean and eventually cools the SST in the subtropical North Atlantic.Correspondingly, when the APO was weaker, positive surface net heat flux anomalies appeared over the mid-latitude, implying that the ocean received heat from the atmosphere and thus warmed the SST over the region.Subsequently, we performed composite analysis on the surface net heat flux, shortwave and longwave radiations, and sensible and latent heat fluxes over the North Atlantic between strong and weak APO years (figures S3(b) and S4).The results indicated that the APO-related surface net heat flux anomalies profoundly influenced the NASST and that the contribution of shortwave radiation was predominant for the surface net heat flux anomalies in key area 1, while the latent heat flux had a larger contribution to the changes in surface net heat flux over key area 2. However, since there are some regions where the surface net heat flux anomalies did not pass statistical significance tests, the above analysis could only partially explain the linkage between the APO and NASST.
We further investigated the possible oceanic dynamic processes in modulating the NASST anomalies associated with the APO because the heat anomalies are closely connected to the zonal wind stress with anomalous Ekman heat transport (Alexander and Scott 1997).Figures 3(d) and (h) show the regression of the zonal wind stress on the APOI and NASSTI.Anomalous easterly wind stress predominated in the region of 80 • W-10 • W, 20 • -35 • N. In this region, the anomalous Ekman flow were induced by the easterlies, leading to warm advection from south to north, which subsequently warmed the NASST over the mid-latitude.Basically, the results exhibited similar patterns for both ERA5 (figure 3) and NCEP/NCAR reanalysis (figure S7).

Conclusions and discussion
We investigated the phenomenon and mechanisms of how the APO influences the NASST on an interannual scale.Our results suggest a significant connection between the two variables, with a significant positive correlation appearing in the mid-latitude North Atlantic and significant negative correlation occurring in the subtropics.The statistical results indicate that the APO-related Rossby wave train propagates from East Asia to Europe through the North Pacific, inducing anticyclonic anomalies in the North Atlantic, which leads to changes in cloud radiation, air-sea heat exchange and oceanic dynamic processes and eventually gives rise to SST variations over the region.Using the linear baroclinic model, we verified how the APO elicits a response of anomalous atmospheric circulation.
We propose that the APO can trigger an eastward propagating Rossby wave train and then significantly influence circulation and oceanic thermal and dynamic processes in the downstream region, consequently providing variations in the NASST.Some studies also emphasized the upstream effect of Asian heating on the climate in the North Atlantic (Zhao et al 2012b, Lu et al 2019).For example, Zhao et al (2012b) suggested that the surface heating of the Asian continent leads to the westward propagation of anomalous tropospheric temperatures, resulting in interannual variations in SSTs over the North Atlantic.Lu et al (2018) pointed out that the surface heating of the TP strengthens the South Asian high, causing it to extend westward, and elicits a response in the troposphere over the North Atlantic, characterized by anomalous high pressure due to the heating-induced Rossby wave.Hence, it remains unclear whether the upstream or downstream waves triggered by APO more effectively modulate NASST.Considering the complex ocean-atmosphere interactions, further analysis with advanced coupled models is needed in the future.
In addition, we should note that both the APO (Zhao et al 2007, Hua et al 2019) and NASST (Delworth et al 1993, Eden andJung 2001) have experienced significant decadal variability in recent decades and the question naturally arises as to whether there is an interdecadal shift in the interannual relationship between APO and NASST.In fact, the results from 21 and 25 years running correlation between APO and NASST indicate that the interannual relationship between APO and NASST is stable on interdecadal timescales (figure S8).Furthermore, since the APO can significantly influence the atmospheric circulation on regional and global scales, which is particularly crucial to the global SST modes, it is worth to extend the analysis from NASST to global SST or mixed layer in the future.

Figure 1 .
Figure 1.(a) Distribution of correlation coefficient between the summer APOI and simultaneous NASST.The black rectangle indicates the key area.Regions above the 95% significance level are dotted.(b) As in (a), but the simultaneous NAO signal has been removed.

Table 1 .
The correlation coefficients between the APOI and NASSTI before and after removing the preceding spring and simultaneous summer NAO signal.
a represent statistical significance above the 99% confidence level.