Remote tropical central Pacific influence on driving sea surface temperature variability in the Northeast Pacific

The Northeast Pacific (NEP) had two record-breaking marine heatwave events in the winters of 2013–2015 and summer of 2019, which had a detrimental impact on the fisheries, marine ecosystems, and climate in North America. Here, we investigated the cause of sea surface temperature (SST) variability in NEP during late spring–summer of 1981–2020. The regression circulation anomalies to the principal component of leading empirical orthogonal function mode suggested that the warm NEP SST were characterized by a cyclonic circulation anomaly in the midlatitude North Pacific and a warming SST center in the Gulf of Alaska. We noted that this cyclonic circulation anomaly, attributable to a barotropic atmospheric wave originating from the tropical central Pacific (CP) in the preceding spring, reduced the surface heat flux loss from the ocean to the atmosphere in the NEP and led to the warm SST anomalies in summer. This finding was confirmed by not only empirical diagnosis but also long-term numerical simulations forced by the observed SST perturbations in the tropical CP. Our results highlight the role of the tropical CP SST in driving the summertime North Pacific SST variability through the atmospheric bridge in recent decades.


Introduction
In 2013-2015, a Northeast Pacific (NEP; 150 • W-130 • W, 40 • N-55 • N) marine heatwave event (MHW) (Di Lorenzo and Mantua 2016), also known as the warm Blob (Bond et al 2015), received considerable research attention due to its prolonged lifetime (Amaya et al 2016, Di Lorenzo and Mantua 2016, Joh and Di Lorenzo 2017, Xu et al 2021 and its devastating impact on of North America's local fisheries (McCabe et al 2016), marine ecosystems (Cavole et al 2016, Jones et al 2018, and regional temperature (Bond et al 2015, Hartmann 2015. The warm sea surface temperature (SST) anomalies of the MHW were initiated in the Gulf of Alaska in the winter of 2013-2014, with a peak SST anomaly of >2.5 • C in the NEP, and the anomalies developed into an arc-shaped pattern along the North American west coast (NAWC) in the winter of 2014-2015 (Amaya et al 2016, Di Lorenzo andMantua 2016). In the summer of 2019, another MHW emerged in the NEP (Amaya et al 2020)-which implied the possibility of MHW recurrence in the NEP in the future.
Recent studies have focused on understanding the mechanisms underlying NEP MHWs. The occurrence of the 2013-2015 MHW was attributable to a long-persistent high-pressure ridge in the NEP; this weakened the local prevailing wind of the Aleutian Low and the evaporation process, reduced the heating loss from the ocean to the atmosphere, and decreased ocean temperature advection (Bond et al 2015, Hartmann 2015, Amaya et al 2016, Di Lorenzo and Mantua 2016, Liang et al 2017. This warming SST anomaly later evolved from the Gulf of Alaska to the extratropical central Pacific (CP) along the NAWC under the influence of atmospheric forcing associated with the Pacific-North American (Wallace and Gutzler 1981) and North Pacific Oscillation (NPO) (Rogers 1981) pattern in 2014 (Amaya et al 2016, Di Lorenzo andMantua 2016). Through statistical analyses based on historical MHWs, Liang et al (2017) indicated that the circulation anomalies associated with the tropical Northern Hemisphere pattern (Mo and Livezey 1986), which has emerged as an influential teleconnection pattern in recent years, could be responsible for the generation of the Pacific Blob. The prolonged warm NEP anomaly in the winter of 2014-2015 has been widely attributed to the tropical teleconnection associated with the El Niño Southern Oscillation (ENSO) (Di Mantua 2016, Xu et al 2021). Compared with the 2013-2015 MHW, the 2019 MHW has been investigated less. Amaya et al (2020) suggested that the warm SST anomaly of the 2019 MHW was triggered by a NPO-like anomalous sea level pressure (SLP) dipole, which weakened evaporative cooling and upper ocean mixing in the NEP. The reduction in the low-cloud fraction due to the warm ocean condition prolonged the warm SST anomaly through the positive lowcloud feedback. The atmosphere-only model simulation in the aforementioned study emphasized the influence of SST anomalies in the equatorial and North Pacific on weakening the summertime North Pacific High. To the best of our knowledge, a comprehensive study involving atmosphere-ocean coupling on the historical MHWs in summer over the NEP region is warranted.
Atmospheric forcings resulting in warm anomalies in the extratropic North Pacific may originate from the tropics (Hartmann 2015, Seager et al 2015, Hu et al 2017, Capotondi et al 2019, Amaya et al 2020. The role of the tropical Pacific in the development of SST variability associated with the MHW has been indicated by climate variability studies using statistical and modeling methods (Capotondi et al 2019, Shi et al 2019, Tang et al 2021, Xu et al 2021. For instance, Capotondi et al (2019) reported that tropical sensitivity patterns, including ENSO SST precursors and tropical centralwestern Pacific SST anomalies, can force anomalously warm SST along the US west coast. By considering the presence of ENSO-like idealized forcings in their slab-ocean climate model, Shi et al (2019) demonstrated a Pacific decadal oscillation-like pattern in the North Pacific. By analyzing a large ensemble simulation from a Pacific-basin linear inverse model, Xu et al (2021) indicated that the tropical influence primarily increased MHW durations, whereas the initial anomalies determined MHW intensities. In addition to the warm Blob events, Tang et al (2021) suggested a strong association between the NEP cold Blob events and the La Niña-like SST cooling. The aforementioned studies mainly demonstrated how tropical forcings can contribute to the long-lived SST anomalies and atmospheric patterns resembling the 2013-2015 MHW.
In addition to tropical forcings, studies have also linked the North Pacific variability to Arctic forcings associated with the warming trend in Bering and Chukchi Seas (Yeo et al 2014, Carvalho et al 2021, Song et al 2023. Carvalho et al (2021) identified increasing frequencies and days of extreme warm ocean temperature events in the Bering and Chukchi Seas correlated to sea ice concentration in Chukchi Sea and air temperature of Alaska. Yao et al (2014) showed an NPO-like atmospheric response to the warming of the Bering and Chukchi Seas, where the climate variability revealed strong covariability with the North Pacific Gyre Oscillation and the CP ENSO. These atmospheric and ocean responses in the North Pacific may further lead to a warming CP through a Pacific Meridional Mode-like pathway in the subtropical Pacific (e.g. Hogn et al 2017, Chen and Chen 2022). A very recent study by Song et al (2023) attributed the increasing marine heatwave days in the NEP to the Arctic warming over the past decades, which can contributed to the upward trend of the SST variability of the same Pacific region. However, in this study we only focus on the tropical influence on the North Pacific variability. Sun et al (2022) indicated an increased Pacific decadal variability over the North Pacific in the 21st century, which was likely influenced by internal variability associated with more frequent CP ENSO events (e.g. McPhaden et al 2011, Yu et al 2012, Chung and Li 2013, Wang et al 2019. However, the authors did not explore the seasonal dependence of the Pacific decadal variability and did not dynamically explain its potential tropical influence on the NAWC region. As illustrated in figure 1(c), the seven-year running SST variance of the NEP revealed an abrupt increase after 2010, particularly during the late spring and summer. Therefore, the aim of the current study was to understand the circulation patterns and the dominant forcings associated with the strengthened SST variability in the NEP, with a particular emphasis on the boreal summer.
The paper is structured as follows: section 2 describes the data, methods, and model used in this study. Section 3 presents the changes observed in the SST variability over the NEP in the recent decades. Section 4 provides the modeling results for investigating the remote tropical influence on the extratropical MHW pattern generation. Finally, the present results are summarized and discussed in section 5.

Data
We used the 1981-2020 monthly mean SST data at a horizontal resolution of 2 • from the Extended Reconstructed Sea Surface Temperature version 5 (ERSSTv5) . Data for monthly mean atmospheric diagnostics during the same period were obtained from the ECMWF Reanalysis v5 (ERA5) (Hersbach et al 2020). We derived the stream function based on the zonal and meridional wind in an N48 Gaussian grid. Monthly anomalies were calculated by subtracting climatologies over the analysis period from the original data.

Wavelet analysis
We applied wavelet analysis-a standard tool used for analyzing the dominant signals within a time seriesto monthly mean SST anomalies over the NEP region (150 • W-130 • W, 40 • N-50 • N) to investigate the evolution of SST variability during 1981-2020. We used a Gaussian derivative (Torrence and Compo 1998) as our base function to provide a fine-scale structure in the power spectrum: where η is the nondimensional time parameter and m is the Gaussian derivative. We also used the Mexican hat wavelet setup (m = 2) for our analysis.

Model
We used the Abdus Salam International Centre for Theoretical Physics (ICTP) atmospheric general circulation model (AGCM; version 41) (Kucharski et al 2013), also called the SPEEDY model (short for simplified parameterizations, primitive-equation dynamics), to conduct pacemaker simulations. The SPEEDY model is based on the Geophysical Fluid Dynamics Laboratory's spectral dynamic core (Held and Suarez 1994) at a horizontal resolution of T30 and has eight vertical model levels from 30 to 925 hPa. In the SPEEDY model, various parameterizations were incorporated into the model to represent the processes of condensation, convection, radiation, and fluxes for momentum, heat, and moisture realistically. The SPEEDY model is coupled to a slab-ocean model to calculate the heat flux exchanges between the ocean surface and the atmosphere, enabling the study of circulation and SST responses to given SST forcing.

Enhanced SST variability in the NEP
Here, we first address the recent regime shift in the SST variability over the NEP. To understand the seasonal distribution of the NEP SST variability in the interannual time scale further, we calculated the seven-year running variance of the SST anomalies for each month in 1981-2020. As shown in figure 1(c), during 1981-2005, the SST variance was highest in fall and, sometimes, in summer. However, this seasonal distinction in variance changes considerably after 2006. Although the SST variance generally increased in all seasons, the maximum variance shifted toward late spring and summer, with the most significant increase noted from May to August (right panel of figure 1(c)). We note that a similar result was found when a linear trend of data was removed before the same analysis (figure S1(c)). Figures 2(a) and (b) further show the May-August mean horizontal structures of SST variance for 1981-2005 and 2006-2020, respectively. The difference between the two periods indicates significant increases of SST variance maximize in the north of the Gulf of Alaska and the NEP region in this study (figure 2(c)).
Next, we explored the empirical orthogonal function (EOF) of the late spring-summer (May-August) SST over the North Pacific (180 • W-120 • W, 30 • N-60 • N) associated with the recent regime shift of SST variability. Figures 3(a) and (b) present the dominant EOF for SST during 1981-2020 and the corresponding principal component, respectively. The leading EOF explained 43.7% of total variance and exhibited a maximum variance near the NAWC ( figure 3(a)). The correlation coefficient of the principal component and the mean SST anomalies over the NEP reaches a significantly high value of 0.95, suggesting the dominant role of the leading EOF for the SST in the NEP ( figure 3(b)).
To further understand the corresponding circulation patterns for the SST variability illustrated in figure 3(a), we regressed data anomalies to a normalized principal component of the leading EOF. Figure 3(c) illustrates the regression anomalies of the May-August mean SST (shading), SLP (contour), and 10 m wind (vector) for the leading EOF. In the extratropical North Pacific, the SST exhibited a significant warming structure-extending from the Gulf of Alaska to the NEP along NAWC, with a southwestward tail toward the tropical CP. These SST anomalies were accompanied by an anomalous SLP dipole: a cyclonic circulation anomaly in the midlatitude central North Pacific and an anticyclonic circulation anomaly at the polar latitude. Amaya et al (2020) identified similar circulation anomalies in the 2019 MHW, which initiated the warm SST anomalies by decreasing the wind evaporation and suppressing upper ocean mixing. Figure 3(c) presents significant warming SST anomalies in not only the North Pacific but also the tropical CP (160 • E-150 • W, 10 • S-10 • N). We evaluated the correlation between the May-August mean SST averaged over the CP and NEP regions. The correlation coefficients were 0.48 in 1981-2005, and it increased to 0.78 in 2006-2020; this indicated a strong correlation between the CP and NEP in 2006-2020 ( figure 3(d)). We also calculated the lag correlation between the March-April CP SST and the May-August NEP SST in 1981-2005 and 2006-2020, and a similar increasing correlation from 0.47 to 0.65 was noted in our analysis. Considering the considerable effects of the CP SST, we hypothesized that the CP SST has a driving effect on the SST variability in the NEP. The results of our pacemaker numerical

Role of remote tropical influence in NEP SST variability induction
This section presents the correlation between the CP and NEP SST variability based on model simulations using the pacemaker setup. In our SPEEDY model experiment, we prescribed observed monthly mean SST over the CP domain and allowed for air-sea coupling processes only in the North Pacific region (120 • E-100 • W, 30 • N-65 • N; figure 4(a)). The remaining ocean areas were forced by seasonal varying SST climatologies estimated over 1981-2020. The monthly heat flux climatologies for running the slabocean model were also updated according to the given SST climatologies. The simulations were from January 1980 to December 2020; we discarded the model output of the first year as the model spin-up time. To reduce model uncertainty, we conducted 100 ensemble members of the simulations.
To understand how the NEP SST responded to the CP SST forcing in the model, we calculated the simulated leading EOFs and regressed circulation anomalies onto the principal component. We used the ensemble mean model output in the subsequent analyses. Figure 4(b) presents the leading EOF of May-August SST for 1981-2020, which exhibited an arc-shaped pattern along the NAWC and resembled the observed EOF in figure 3(a). The simulated EOF explained an extremely large amount of variance (91.4%). This result indicated that the simulated SST variability in the NEP was mainly attributable to the CP SST anomalies, which was the only time-varying forcing other than the mean climatological seasonal forcings in the suit of simulations. Furthermore, figure 4(c) presents the simulated circulation anomalies associated with the corresponding principal component. The result suggested that our SPEEDY model experiment reproduced the SST anomalies and the meridional SLP anomalies in the extratropical North Pacific (figure 3(c)). However, we also found some discrepancies between our simulations and observations. For instance, the simulated variance center in figure 4(b) is more equatorward, and the magnitudes of the simulated SST anomalies are smaller than those of the observed anomalies, presumably because of the lack of oceanic dynamics in or the coarse resolution of the SPEEDY model. We also note that the spatial structure of the increased SST variability during 2006-2020 cannot be properly simulated in our numerical experiments (figure not shown) and would need further investigation by taking different approach. In general, our simulation results indicated that the CP SST can drive SST variability and anomalies in the NEP.
We next investigated the origin of the cyclonic circulation anomalies in the North Pacific ( figure 3(c)), which played a crucial role in generating warm SST anomalies during late spring-summer (e.g. Amaya et al 2020). Figures 5(a) and (b) illustrate the regressed anomalies of zonally averaged stream function across the North Pacific (150 • E-150 • W) for the spring (March-April) and the subsequent summer (May-August). The summertime stream function (figure 5(b)) suggested that the cyclonic circulation anomaly between 30 • N and 50 • N in figure 3(c) demonstrated a substantial barotropic structure in the lower troposphere. Figure 5(a) further indicates that the midlatitude circulation anomaly was associated with a barotropic wave train originating in the tropics during the preceding spring. Notably, this wave train was not significant in the preceding winters (January-February; figures S2(a) and (d)). We also simulated the springtime wave train ( figure 5(d)), and the perturbation centers and the tropical baroclinic structure reasonably agree with those observed, albeit the SPEEDY model has a coarse vertical resolution yielding a shallower structure. Therefore, our simulation indicated the effects of CP SST on the generation of a northward-propagating wave train that affected the extratropical North Pacific in spring. Figure 5(c) presents the time series of observed summertime SST in the NEP and its correlation with the net surface heat flux (positive values towards the ocean) in summer as well as in the preceding spring and winter. A significant correlation coefficient of 0.41 was noted between the SST and the springtime surface heat flux, whereas a lower nonsignificant correlation coefficient of 0.21 and −0.06 was found with the wintertime and the summertime heat flux, respectively. We repeated this analysis using the simulated data and obtained a similar result: the correlation coefficient between the summertime SST and the springtime surface heat flux in the NEP was highest (0.97). Our results suggested that summertime warm SST initiation was dominated by reduced local heat flux loss from the ocean to the atmosphere that began in the preceding spring. The correlations for January-February (JF) (0.91) and May-August (MJJA) (0.67) were much higher than the observation. This discrepancy with the observation was likely due to the experimental design that prescribed CP SST as single forcing and ignore effects of other factors such as internal variability and forcing from other SST anomalies that might cause the cooling in the NEP in JF and MJJA. Nevertheless, our simulations suggested that the summertime SST variability in the NEP could be strongly modulated by  the CP SST in spring. The reason for this strong seasonal dependence that is not understood in current study deserves further investigation. Figure 6 illustrates the horizontal structures of circulation and surface heat flux regression during the preceding spring. The stream function anomaly in the upper troposphere exhibited a meridional wave train, which passed through the North Pacific ( figure 6(a)), accompanied by a corresponding cyclonic circulation and negative SLP anomalies at the surface (figure 6(c)). As mentioned in the earlier text, the southerly anomaly along the NAWC weakened the prevailing northerly and the local wind-evaporation-SST feedback, resulting in a reduction in heat flux loss from the ocean to the atmosphere. The result was confirmed by the consistent downward surface heat flux anomalies noted along the NAWC and in the NEP (figure 6(e)). Furthermore, our SPEEDY model experiment reproduced these dynamic and thermal perturbations in spring (figures 6(b), (d) and (f)), indicating that the springtime barotropic wave train associated with the late spring-summer NEP SST variability was of tropical origin.

Summary and conclusion
The present study investigated the enhanced interannual SST variability over the NEP during the summers of 2001-2020. The leading EOF of SST exhibited an MHW-like pattern resembling the 2013-2015 and 2019 MHWs. The composite anomalies for the selected warm SST years associated with the leading EOF demonstrated a significant warming SST center in the NEP, accompanied by an arc-shaped SST tail extending southwestward to the CP. The corresponding atmospheric circulation patterns demonstrated a negative SLP anomaly and a weakened surface northerly along the NAWC as a part of the cyclonic circulation anomalies in the midlatitude North Pacific. We noted that these circulation anomalies in summer were attributable to a barotropic wave train in the preceding spring. The surface wind anomalies associated with the wave train reduced surface heat flux loss from the ocean to the atmosphere in the NEP and NAWC, leading to warm SST anomalies in the subsequent summer. Based on our SPEEDY model experiment with observed SST forcings, we confirmed that the barotropic wave in the spring of selected warm SST years had a tropical origin from the CP.
To the best of our knowledge, this is the first study to investigate historical MHWs in the NEP during summer on the basis of an air-sea coupled model framework. Our results support the role of the tropical Pacific in driving the extratropical MHWs, which has been indicated by previous studies that used either statistical methods or AGCM simulations. We further demonstrated the dynamic processes linking the tropics and extratropics in the preceding spring. However, because of the simplicity of the SPEEDY model, we could not explore the effects of ocean temperature advection on MHW generation, which may explain the relatively small magnitude of simulated circulation and SST anomalies in our analyses. In the future, a more realistic climate model may be used for further clarification of the current results. On the other hand, the intensity of NEP MHWs is also related to the local initial anomalies (Xu et al 2021), but the role of the extratropical internal variability in the NEP SST variability is not investigated in this study. The impact of the Arctic forcing as well as the influence of the Atlantic multidecadal oscillation are also not considered in the current study. We also intend to address these issues in the future study.
Although the increasing frequency of the CP ENSO was potentially responsible for the changing Pacific decadal variability over the past decades (Sun et al 2022), this study revealed the role of internal variability of the CP SST in amplifying the interannual variability of the NEP. Our results highlighted a stronger tropical-extratropical coupling in the North Pacific over the recent decades and implied potential subseasonal predictability for summertime NEP MHWs from the tropics.

Data availability statement
The data that support the findings of this study are available upon reasonable request to the authors.