The Relationships Between Meridional Position of North Pacific Sea Surface Temperature Anomalies and North American Surface Temperatures Revealed by CMIP6 Models

In this study, we obtained the first leading mode and principal component 1 (PC1) of North Pacific sea surface temperature (SST). The PC1‐related SST anomalies, located relatively north/south, are referred to as North/South PC1 events. Model outputs, observations, reanalysis datasets and sensitivity experiments show that during the North PC1 events, an enhanced Aleutian low occurs and is located relatively north, favoring anomalous southerlies over western North America. The anomalous southerlies induce a strong warming anomaly over North America via warm advection of the anomalous southerlies and temperature advection of the climatological westerlies over North America. However, the Aleutian low anomaly and corresponding southerly anomaly associated with South PC1 events shift southward, favoring weakened effects of South PC1 events on atmospheric circulations and surface temperatures over North America. The meridional position of PC1 events deserves to be considered in the studies of the PC1 and its related climate changes.

fisher data, Mantua et al. (1997) reported that the positive/negative phase of the first leading mode induces a increase/decrease in Alaskan salmon catches.
The first leading mode phase change not only modulates ecosystem variations, but also broadly affects climate changes (e.g., Hurwitz et al., 2012;Tao et al., 2020). Its phase change significantly influences climate variations over North America, including surface temperature, precipitation, streamflow and water resources (Barlow et al., 2001;Cayan et al., 1998;Dettinger et al., 1998;Mantua & Hare, 2002). Over Asia, variations in surface temperature, precipitation and Asian monsoon are closely connected to the first leading mode phase change (Krishnan & Sugi, 2003;Mantua & Hare, 2002;Simon et al., 2022). Additionally, Rao, Garfinkel, and Ren (2019) reported that the first leading mode can constructively (destructively) intertwine with El Niño-Southern Oscillation (ENSO) forcing to induce Pacific-North American teleconnections when ENSO and the first leading mode are in (out of) phase. Recent studies even found that the positive phase of the first leading mode weakens the Arctic stratospheric polar vortex and induces more stratospheric sudden warming in winter (Hurwitz et al., 2012;Kren et al., 2016;Woo et al., 2015).
As mentioned above, the effects of first leading mode phase change on climate variations have got much attentions. However, recent studies reported that the meridional positions of SST anomalies associated with the first leading mode are not stationary, and they oscillate between the north and the south (Wang, Fu, et al., 2022;. But the impacts of meridional position changes of the first leading mode-related SST anomalies on surface climate have been unclear. This study examines the effects of meridional positions of the first leading mode-related SST anomalies on surface temperatures over North America (NA) and underlying mechanisms.

Data and Methods
This study mainly uses the model outputs from the first historical run (1850/51-2013/14) of the Phase 6 of the Coupled Model Intercomparison (CMIP6) to perform analysis because the CMIP6 outputs provide large sample sizes, which examine the robustness of the results via considering various simulated internal variability in the atmosphere and ocean. The variables we used include SST, surface temperature, geopotential height (GH) and horizontal winds. We analyzed 36 CMIP6 models available, which are listed in Table S1 in Supporting Information S1.
The National Centers for Environmental Prediction and the National Center for Atmospheric Research (NCEP-NCAR) Reanalysis data set (Kalnay et al., 1996) for the period of 1948/49-2019/2020 and the fifth generation European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis (ERA5; Hersbach et al., 2020) for the period of 1950/51-2019/2020 are also analyzed. SST data from the Hadley Centre Sea Ice and Sea Surface Temperature Data Set Version 1 (HadISST) product (Rayner et al., 2003) and the Extended Reconstructed Sea Surface Temperature version 5 (ERSST; Huang et al., 2017) are analyzed.
In this study, winter denotes the months from December of a year to February of the following year (DJF); for instance, the winter of 2010 represents the period from December 2009 to February 2010. The first leading mode and corresponding principal component 1 (PC1) of North Pacific SST are obtained by Empirical Orthogonal Function (EOF) decomposition. Notably, the PC1 has a close relationship with ENSO (Alexander & Scott, 2008;Newman et al., 2016). Thus, to exclude ENSO's interference on the effects of PC1-related SST anomalies, ENSO signals in variables (SST, GH, etc.) are removed in each model through a linear regression method (Rao, Ren, et al., 2019). Similar to Rao, Garfinkel, and Ren (2019), the PC1 index is obtained as follows: The linear trends and seasonal climatology of the data are removed in each model to produce anomalies. Then, the 36 model data are concatenated in a fixed order (from the first model, ACCESS-CM2, to the last model, TaiESM1) to construct a 5904-year data set. Finally, the PC1 index of North Pacific (20.5°-65.5°N, 124.5°E-100.5°W) SST is obtained by performing EOF decomposition for the 5904-year data set. The spatial pattern of the first leading mode is shown in Figure 1a.
PC1 events are picked according to a threshold of 1.0 standard deviation of PC1 index, and we get a total of 1,072 PC1 events. Note that if there are several consecutive winters with PC1 indices higher than 1.0 standard deviation, these winters are defined as a single PC1 event in CMIP6, however, if they are defined as multiple PC1 events, the results are similar (not shown). Figure 1b shows SST anomalies composited by PC1 events. The negative SST anomalies have a well-defined elliptical shape (the green contour in Figure 1b). Therefore, the central latitude of this negative center is used to represent the meridional position of a PC1 event. Since the SST anomalies less than −0.4 K (i.e., the green contour) are apparent and statistically significant, the −0.4-K contour is denoted as the negative center border. Notably, Figure 1b shows SST differences between positive (+) and negative (−) PC1 events, and hence the border is −0.2 K and +0.2 K for +PC1 and -PC1 events, respectively. The central latitude of the negative center is defined as the averaged latitude of enclosed negative SST anomalies, with considering the weights of SST anomalies : where n is the total number of the grids enclosed by the border. Lat (i) and SST (i) are the latitude and SST anomaly of the ith grid cell, respectively. Latc represents the central latitude of the negative center. According to Equation 1, the central latitude of negative SST anomalies during each PC1 event is obtained. The averaged central latitude of the 1072 PC1 events is 36.83°N. Based on the central latitudes and phases of PC1 events, the PC1 events are categorized into North +PC1, North -PC1, South +PC1 and South -PC1 events. South +PC1/-PC1 events denote the +PC1/-PC1 events with central latitudes south of 36.83°N while North +PC1/-PC1 events are those north of 36.83°N.

Surface Temperature Anomalies During North and South PC1 Events
Based on CMIP6 data, we get 276 North +PC1, 271 North -PC1, 261 South +PC1 and 264 South -PC1 events, and Figures 1c and 1d show surface temperature anomalies over NA and East Asia in winter associated with North and South PC1 events (+PC1 minus -PC1 events). Interestingly, surface temperature anomalies during North PC1 events (referred to as NPE hereafter for simplicity) are different from those associated with South PC1 events (referred to as SPE hereafter) over NA (50°-70°N, 170°-90°W), northeast Asia (55°-70°N, 120°E-170°W; NEA) and midlatitude East Asia (30°-45°N, 80°-120°E; MEA). Over NA/NEA, a significant anomalous warming/cooling occurs during NPE ( Figure 1c) while during SPE, those are not prominent ( Figure 1d). However, over the MEA region, NPE-related anomalous cooling is weak, but the MEA cooling is apparent during SPE. Over the regions of 45°-55°N, 100°-130°E, NPE-related surface temperature anomaly is similar to that during SPE, and both them show anomalous cooling (Figures 1c and 1d). Note that the above descriptions correspond to the positive phase of the PC1, and anomalous cooling will occur over NA during the negative phase of the PC1. Hence, Figures 1c and 1d indicate that the anomalous cooling is more serious over NA during North -PC1 events compared to South -PC1 events. However, it should be noted that the magnitudes of NEA and MEA cooling anomalies are apparently smaller than the magnitude of NA warming anomaly.
The PC1-related surface temperature anomalies in NCEP-NCAR data set are also analyzed (Figures 1e and 1f). During NPE, a significant anomalous warming occurs over NA (50°-70°N, 170°-90°W; Figure 1e), consistent with CMIP6 data. Although the NEA (45°-65°N, 110°-160°E)/MEA (20°-40°N, 80°-120°E) cooling anomaly is also weaker/stronger during SPE compared to NPE in reanalysis data (Figures 1e and 1f), the NEA and MEA cooling anomalies are less significant and much weaker compared to NA warming anomaly in reanalysis data. Thus, we mainly focus on the NA temperature variations associated with NPE and SPE hereafter, and we also briefly discuss the NEA and MEA surface temperature anomalies during NPE and SPE. Figure 2a displays the scatter plots of PC1 event latitude and surface temperature anomalies averaged over NA in CMIP6. When PC1 events shift northward (i.e., with larger PC1 event latitude), the anomalous warming over NA becomes stronger (Figure 2a), consistent with Figures 1c and 1d. A significant linear relationship, with a slope of 0.12 K per degree, between the PC1 event latitude and surface temperature anomalies over NA is established, suggesting that the magnitude of surface temperature anomaly over NA will increase by 0.12 K for each degree of increase in PC1 event latitude. The results from NCEP-NCAR data set (Figure 2d) are consistent with those from CMIP6 (Figure 2a). The slope of linear fitted line in NCEP-NCAR data set is also close to that in CMIP6 over NA (0.13 vs. 0.12). In addition, for NEA and MEA, there are also linear relationships between PC1 event latitude and surface temperature anomalies over NEA and MEA (Figures 2b-2c and 2e-2f). These results indicate that the NA surface temperature anomalies during NPE quite differ from those during SPE. Now a key question arises as to why the NPE-related NA surface temperature anomalies show different characteristics compared to SPE.

Mechanisms for Different Surface Temperature Anomalies During North and South PC1 Events
Figure 3 displays SST and circulation anomalies associated with North and South PC1 events. Indeed, NPE-related negative SST anomalies are located further north of those during SPE (Figures 3a and 3b). The formation of the SST anomalies located relatively north or south may be related to the positive air-sea interactions in the North Pacific (e.g., Fang & Yang, 2016;Okajima et al., 2014), that is, the PC1-related SST anomalies help the Aleutian low anomaly via influencing atmospheric baroclinicity and synoptic-scale eddy (e.g., Chen et al., 2020;Fang & Yang, 2016), and in turn, the Aleutian low anomaly reinforces the PC1-like SST anomalies by affecting surface heat flux and oceanic temperature advection (e.g., Alexander & Scott, 2008;Newman et al., 2016). Thus, the initial perturbation in the North Pacific SST or atmospheric circulation located relatively north/south could be developed and maintained by the positive air-sea interactions, and thereby North Pacific SST anomalies and corresponding atmospheric circulation anomalies form over the north/south. Notably, the spatial patterns of SST anomalies during NPE and SPE (Figures 3a and 3b) resemble those of subarctic and subtropical oceanic fronts in Nakamura et al. (1997), respectively. Therefore, the NPE and SPE could represent relative importance of the contributions between the two oceanic fronts in the North Pacific (Nakamura et al., 1997).
There are negative GH anomalies over high-latitude North Pacific during NPE (Figure 3c), which are in phase with the climatological trough (dashed contours), indicating a strengthened Aleutian low. This is consistent with previous studies that the positive phase of the PC1 potentially helps an enhanced Aleutian low (Mantua & Hare, 2002;Jadin et al., 2010;Mills & Walsh, 2013;Kren et al., 2016;Rao, Garfinkel & Ren, 2019;Rao, Ren, et al., 2019). According to geostrophic wind relations, anomalous northerlies appear over western NA (Figure 3g), and anomalous easterlies and westerlies occur over northern and middle North Pacific, respectively (Figure 3e). On the climatology, the high latitudes are generally colder than middle latitudes. Therefore, the southerly anomalies over western NA (Figure 3g) favor local warming (Figure 1c) via warm advection. The climatological westerly over NA ( Figure S1 in Supporting Information S1) further favors eastward advection of the warming anomaly (induced by the southerly anomalies) over western NA, contributing to the NA warming anomaly (Figure 1c). Additionally, the NEA northerly anomaly with relatively small magnitudes during NPE (Figure 3g) may contribute to local cooling anomaly (Figure 1c), while the MEA circulation anomalies are weak during NPE, leading to MEA weak temperature anomalies (Figure 1c).
However, compared to NPE, the Aleutian low anomaly and wind anomaly associated with SPE shift southward (Figures 3d, 3f and 3h). Especially, SPE-related meridional wind anomalies (Figure 3h) apparently shift southward The red lines are corresponding linear fitted lines, and slopes (k, units: K/degree) and significance (p) of the red lines are added to each panel. A slope is statistically significant at the 95% confidence level if p is less than 0.05. Notably, to measure the relationship between PC1 event latitude and surface temperature, surface temperature anomalies and PC1 indices during -PC1 events are multiplied by −1 in this figure.
Figure 3. Differences in wintertime SST between (a) North +PC1 and North -PC1 events, (b) South +PC1 and South -PC1 events based on CMIP6 outputs. The dotted regions are statistically significant at the 95% confidence level according to Student's t test. (c-d) As in (a-b) but for 500-hPa geopotential height (colors). The contours denote climatological geopotential height with zonal mean removed at 500 hPa. The solid and dashed contours represent positive and negative values, respectively, and the contour interval is 50 m. (e-f) As in (a-b) but for 500-hPa zonal wind. (g-h) As in (a-b) but for 500-hPa meridional wind. compared to NPE (Figure 3g). This southward shift leads to weak meridional wind anomalies over NA, inducing relatively weak temperature anomalies over NA during SPE (Figure 1d). The results from NCEP-NCAR data set also show that SPE-related meridional wind anomalies shift from high-latitude North Pacific to midlatitude North Pacific ( Figure S2h in Supporting Information S1) compared to NPE ( Figure S2g in Supporting Information S1). Therefore, the SPE-related surface temperature anomalies over NA are weak (Figure 1f). The southward shift of meridional wind anomalies also favors the weak surface temperature anomalies over NEA during SPE (Figures 1d and 1f) but enhances northerly anomalies over MEA and western North Pacific (Figure 3h, Figure  S2h in Supporting Information S1), which may help the cooling anomaly over MEA during SPE.
We check whether our results are sensitive to datasets and methods. In Figures S3 and S4 in Supporting Information S1, the PC1 index is obtained based on SST with removing global mean value instead of removing linear trends, and ENSO signals are kept. In Figures S5 and S6 in Supporting Information S1, the PC1 events are categorized into three types (North, Classic and South PC1 events) and four types (Extreme North, Moderate North, Moderate South and Extreme South PC1 events), respectively. Figure S7 in Supporting Information S1 shows the results from ERA5 and ERSST data. The results based on different methods and datasets still indicate that compared to NPE, the SPE-related Aleutian low anomaly shifts southward, favoring weak NA surface temperature anomalies during SPE (Figures S3-S7 in Supporting Information S1), supporting the robustness of the results.

Summary and Discussion
This study defines a meridional position index of PC1 events, and its role in surface temperature variations over North America (NA) is examined using CMIP6 outputs, NCEP-NCAR data set, ERA5 data set and sensitivity experiments. The PC1 events are categorized into North PC1 events (NPE) and South PC1 events (SPE) according to the meridional position index.
We found that during NPE (positive-minus-negative events), a strengthened Aleutian low anomaly occurs over high-latitude North Pacific, which is accompanied by anomalous southerlies over western NA. The southerly anomalies favor NA warming via warm advection due to the southerly anomalies and NA climatological westerly. However, the SPE-related NA surface temperature anomalies are relatively weak due to a southward shift of Aleutian low anomaly from high-latitude North Pacific to midlatitude North Pacific during SPE. The results from NCEP-NCAR data, ERA5 data and sensitivity experiments resemble those in CMIP6. The above results suggest that the meridional position property of PC1 events deserves to be considered in the studies of the PC1 and its related weather and climate changes.
In addition, we also conduct three sensitivity experiments to analyze the NA temperature variations during NPE and SPE, including experiments R0, R1 and R2. The model we used is the Whole Atmosphere Community Climate Model version 4 (WACCM4; Marsh et al., 2013), which has 66 vertical levels and a horizontal resolution of 1.9° × 2.5° (latitude × longitude). The R0 is a control experiment forced by the climatological (1950-2020) monthly mean SST. R1 and R2 used the same SST as R0 plus the SST anomalies composited by NPE ( Figure 4a) and SPE (Figure 4b) in the North Pacific (20°-65°N, 120°E-110°W), respectively. To alleviate the discontinuities about time, annual cycle monthly SST anomalies composited from wintertime PC1 events were used. All the experiments were run for 50 years. The first 5 years are excluded for model spin-up time, and the last 45 years are analyzed.
The circulation anomalies associated with North and South PC1-related SST anomalies are shown in Figures 4c  and 4d, respectively. The NPE-related SST anomalies are accompanied by negative GH (cyclonic circulation) anomalies, which are located over the northern North Pacific (30°-60°N, 150°E-140°W; Figure 4c). Figures S8a and S8c in Supporting Information S1 also show that North PC1-related GH anomaly over northern North Pacific (blue lines) is significantly (at the 98% confidence level) smaller than that in control run (red lines). The cyclonic circulation anomaly favors NA warming anomaly (Figure 4e). By contrast, the circulation anomaly (25°-45°N, 140°E-120°W in Figure 4d and blue line in Figure S9 in Supporting Information S1) associated with South PC1 shifts southward compared to that associated with North PC1 (Figure 4c and red line in Figure S9 in Supporting Information S1), and thus the NA warming anomaly is not prominent (Figure 4f). The sensitivity experiments support the mechanism proposed by this study, that is, the SPE-related NA warming anomaly is weaker than that during NPE by a southward shift of Aleutian low anomaly. Additionally, the spatial correlation coefficient between North Pacific (20°-70°N, 120°E-130°W) GH anomalies in sensitivity experiments (Figures 4c and 4d) and those in NCEP-NCAR data (Figures S2c and S2d in Supporting Information S1) is 0.77, and the consistency in the ratio of sign between North Pacific GH anomalies from sensitivity experiments and those from NCEP-NCAR data is 81%, suggesting that the spatial patterns of simulated GH anomalies associated with the PC1 overall resemble those in NCEP-NCAR data.
Although the NEA cooling anomaly, with relatively small magnitudes compared to NA warming anomaly, is significant during NPE in CMIP6 (Figure 1c), it is less significant in NCEP-NCAR data (Figure 1e), ERA5 data ( Figure S7e in Supporting Information S1) and sensitivity experiments (box regions in Figure 4e). This may be due to that the SST anomalies ( Figure S10a in Supporting Information S1) and Aleutian Low anomaly ( Figure 3c) associated with NPE in CMIP6 are located further north compared to observed data (Figures S10c and S10d, S2c in Supporting Information S1). And the sensitivity experiments also used observed SST as forcing  (20°-65°N, 120°E-110°W). This SST anomaly in the North Pacific is the same as that in Figure S2a (Supporting Information S1), which denotes SST anomalies associated with North PC1 events. The SST forcing is constructed from HadISST. (b) As in (a) but for R2, which is the same as that in Figure S2b in Supporting Information S1 and denotes SST anomalies related to South PC1 events. (c-d) Differences in wintertime geopotential height (colors) and horizontal wind (vectors) at 500 hPa between experiments (c) R1 and R0, (d) R2 and R0. Dotted and hatched regions denote statistically significant anomalies at the 90% and 95% confidence level according to Student's t test, respectively. (e-f) As in (c-d) but for surface temperature. The box in (e) denotes the regions of 62°-70°N, 165°E-170°W.