-
In this study, winter refers to the months from December of a year to February of the following year (DJF); e.g., the winter of 1950 refers to the period from December 1949 to February 1950. The SST data are obtained from the Hadley Centre Sea Ice and Sea Surface Temperature Data Set Version 1 (HadISST; Rayner et al., 2003) product at a horizontal resolution of 1° latitude × 1° longitude. Geopotential height (GH), temperature, and zonal and meridional wind data with a horizontal resolution of 2.5° latitude × 2.5° longitude are obtained from the National Centers for Environmental Prediction and the National Center for Atmospheric Research (NCEP–NCAR) Reanalysis products (Kalnay et al., 1996). The data used in this study cover the period from December 1949 to February 2016.
-
Because the first leading mode of North Pacific SST has a close relationship with ENSO (e.g., Alexander and Scott, 2008; Newman et al., 2016; Rao et al., 2019a), to focus on the climate impacts of North Pacific SST, ENSO signals in the various factors (SST, geopotential height, etc.) are removed by a linear regression method before analysis (Ren et al., 2017; Rao et al., 2019a). Then, following Mantua et al. (1997) and Ding et al. (2015), the first leading mode and corresponding PC1 of North Pacific SST (20.5°–65.5°N, 124.5°E–100.5°W) in winter (DJF) are obtained by EOF decomposition, and the EOF1 mode is shown in Fig. 1a. Note that for simplicity, we use the positive phase of PC1 to describe the anomalies composited by PC1 index hereafter. The PC1 events are selected based on the threshold of 1.0 standard deviation of PC1 indices, and the results are not sensitive to the choice of the threshold. The composite SST anomalies associated with PC1 events are shown in Fig. 1b. We can see that the negative SST anomaly center has a well-defined elliptical shape, as denoted by the yellow contour in Fig. 1b. Thus, the central latitude of the negative center is used to denote the position of a PC1 event. Because the SST anomalies less than –0.6 K are evident and significant, the –0.6-K contour is regarded as the border of the negative center. Note that Fig. 1b denotes the difference between +PC1 and –PC1 events, and thus the border is –/+0.3 K for +/–PC1 events. The central latitude of the negative center is defined as the averaged latitude of the enclosed negative SST anomalies weighted by the SST anomalies and is calculated by the following formula (Wang et al., 2022b):
Figure 1. (a) Spatial pattern of the EOF1 mode of SST anomaly fields in the North Pacific (20.5°–65.5°N, 124.5°E–100.5°W) in winter, and the corresponding principal component 1 (PC1) is also obtained. (b) Differences in SST between +PC1 and –PC1 events. The yellow contour denotes the –0.6-K contour of the SST anomaly fields. (c) As in (a) but for differences between North +PC1 and North –PC1 events. (d) Differences in SST between South +PC1 and South –PC1 events. The green contour in (d) denotes the –0.6-K contour of the SST anomaly fields, and the yellow contour in (c) is superimposed on (d) to facilitate direct comparison. The SST data are from HadISST. The dotted regions are statistically significant at the 90% confidence level according to Student’s t-test.
where n is the total number of the grids within the border (i.e., less/greater than –/+0.3 K for +/–PC1 events). SST(i) and Lat(i) are the SST anomaly and latitude of the ith grid cell. Latc denotes the central latitude of the negative center.
Based on Eq. (1), we obtained the central latitude of the negative SST anomalies of each PC1 event (Table 1). As shown in Table 1, the averaged central latitude of all PC1 events during 1950–2016 is 36.84°N. Then, the PC1 events are categorized into North +PC1 events, North –PC1 events, South +PC1 events, and South –PC1 events based on the phases and central latitudes of PC1 events. North +PC1/–PC1 events represent the +PC1/–PC1 events with central latitudes north of 36.84°N, while South +PC1/–PC1 events are those with central latitudes south of 36.84°N. These four-type PC1 events are listed in Table 1. “North +PC1/–PC1 events” and “South +PC1/–PC1 events” are used in this study for simplicity, and this study does not intend to define new PC1 events. The composite SST anomalies associated with North and South PC1 events are shown in Figs. 1c–d. We can see that North PC1-related negative SST anomalies (yellow contour) are located relatively northward compared to South PC1 events (green contour). Note that our results are not sensitive to the criterion of the –0.3-K contour versus –0.2-K contour to derive the central latitude of a PC1 event.
Year PC1 index Central latitude Year PC1 index Central latitude (a) North +PC1 events North –PC1 events 1960 1.19 41.7°N 1956 –1.23 39.2°N 1961 1.22 39.3°N 1973 –1.34 38.0°N 1970 1.28 36.9°N 1991 –1.52 41.3°N 1977 1.51 38.1°N 1995 –1.75 37.5°N 1985 1.47 37.4°N 1987 1.42 39.2°N (b) South +PC1 events South –PC1 events 1959 1.47 36.3°N 1952 –1.68 35.7°N 1981 2.03 34.7°N 1966 –1.06 34.1°N 1984 2.22 36.8°N 1969 –1.11 36.4°N 1986 1.70 35.5°N 1972 –2.25 36.8°N 1996 1.49 33.7°N 2009 –1.03 36.4°N 2015 1.72 29.8°N 2012 –1.46 35.7°N Table 1. North +PC1 events, North –PC1 events, South +PC1 events, and South –PC1 events during 1950–2016 based on HadISST. The averaged central latitude of all PC1 events is 36.84°N during 1950–2016. North/South PC1 events denote the PC1 events with central latitudes north/south of 36.84°N.
-
The model used for the numerical analyses in this study is the WACCM version 4 (WACCM4) (Marsh et al., 2013), and WACCM4 is part of the Community Earth System Model (CESM) framework developed by the NCAR (Vertenstein et al., 2012). This study used CESM version 1.2.0, and the case used is “FWSC”. WACCM4 has a well-resolved stratosphere with 66 vertical levels extending from the surface to approximately 140 km. The vertical resolution of this model is approximately 1 km in the tropical tropopause layer and the lower stratosphere. The model has a horizontal resolution of 1.9° × 2.5°, and the simulations performed for this study do not include interactive chemistry.
To analyze the effects of PC1 events with different central latitudes on the Arctic SPV, we conducted three time-slice experiments, including R0, R1, and R2. R0 is a control experiment that is forced by the climatologically (1950–2016) monthly mean SST. R1 used the same SST as R0 plus the SST anomalies composited by the North PC1 events over the North Pacific regions of 20°–65°N, 120°E–110°W (Fig. 1c). R2 used the same SST as R0 plus the SST anomalies composited by the South PC1 events over the same North Pacific regions (Fig. 1d). To alleviate the discontinuities due to added SST anomaly forcing at the boundary of the North Pacific region considered, SST anomalies at the boundary are added to the three grid points closest to the boundary with weights of 0.75, 0.50, and 0.25, respectively, moving away from the boundary (Zhou et al., 2018). To alleviate the discontinuities about time, annual cycle monthly SST anomalies composited from wintertime PC1 events were used. All the experiments were run for 34 years. The first four years are excluded for the model spin-up time, and the last 30 years are analyzed. Note that the SST forcing in the experiments is fixed, and thus the SST forcing in each model year is held constant.
-
To verify the results obtained from the reanalysis data, the model outputs from CMIP5 are analyzed by using the last 100 years (1905/06–2004/05) of the first historical run. The simulations of the 100 years from CMIP5 models examine the robustness of the results by taking into account a large variety of model simulated internal variabilities in the ocean and atmosphere. Because the low-top models simulate a worse state of the polar vortex than high-top models (Rao et al., 2022), the high-top models are used in this study. We analyzed the outputs from 20 high-top models available, i.e., CanESM2, CMCC-CESM, CMCC-CM, CMCC-CMS, GFDL-CM3, GISS-E2-H, GISS-E2-H-CC, GISS-E2-R, GISS-E2-R-CC, HadGEM2-CC, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR, MIROC-ESM, MIROC-ESM-CHEM, MPI-ESM-LR, MPI-ESM-MR, MPI-ESM-P, MRI-CGCM3, and MRI-ESM1. The tops of these models are equal or higher than 1 hPa. Similar to Rao et al. (2019c), PC1 events in CMIP5 model outputs are picked as follows: Firstly, the linear trends, seasonal climatology, and ENSO signals of the data are removed in each model to produce anomalies. Then, the data from the 20 high-top models are concatenated in a fixed order (i.e., from the first model, CanESM2, to the last model, MRI-ESM1) to construct a 2000-yr dataset. Finally, PC1 indices from the model data are obtained by performing EOF analysis for the 2000-yr dataset.
Year | PC1 index | Central latitude | Year | PC1 index | Central latitude |
(a) North +PC1 events | North –PC1 events | ||||
1960 | 1.19 | 41.7°N | 1956 | –1.23 | 39.2°N |
1961 | 1.22 | 39.3°N | 1973 | –1.34 | 38.0°N |
1970 | 1.28 | 36.9°N | 1991 | –1.52 | 41.3°N |
1977 | 1.51 | 38.1°N | 1995 | –1.75 | 37.5°N |
1985 | 1.47 | 37.4°N | |||
1987 | 1.42 | 39.2°N | |||
(b) South +PC1 events | South –PC1 events | ||||
1959 | 1.47 | 36.3°N | 1952 | –1.68 | 35.7°N |
1981 | 2.03 | 34.7°N | 1966 | –1.06 | 34.1°N |
1984 | 2.22 | 36.8°N | 1969 | –1.11 | 36.4°N |
1986 | 1.70 | 35.5°N | 1972 | –2.25 | 36.8°N |
1996 | 1.49 | 33.7°N | 2009 | –1.03 | 36.4°N |
2015 | 1.72 | 29.8°N | 2012 | –1.46 | 35.7°N |