1. Introduction
The Victoria mode (VM) is the second dominant pattern of sea surface temperature (SST) variability over the North Pacific, and is independent of the Pacific decadal oscillation (PDO; Zhang et al. 1997; Mantua et al. 1997; Bond et al. 2003; Di Lorenzo et al. 2008). The previous studies have indicated that the VM is an anomalous mode of SST formed mainly by North Pacific Oscillation (NPO)-like atmospheric variability forcing of the ocean (Walker 1925; Rogers 1981). Wind fluctuations associated with NPO variability induce SST anomalies (SSTAs) over the North Pacific by changing surface heat fluxes, giving an anomalously warm SST band extending across the Pacific from off California into the western Bering Sea and an anomalously cool SST band that extends from the central North Pacific to along the Asian coast; this apparent northeast–southwest-tilted dipole structure over the extratropical North Pacific forms the most distinctive spatial characteristic of the VM (Bond et al. 2003; Ding et al. 2015b).
Variability of the VM has been proven to have a significant impact on climate anomalies in many regions of the globe. For instance, the VM has a pronounced effect on low-frequency climate variability and marine ecosystems over the North Pacific (Bond et al. 2003; Di Lorenzo et al. 2008; Ceballos et al. 2009). Additionally, the VM also effectively influences precipitation over the Pacific ITCZ, the intensity of the South China Sea summer monsoon, and the frequency of tropical cyclone activity across the western North Pacific (Ding et al. 2015a, 2018; Li et al. 2020; Pu et al. 2019).
In particular, the boreal spring VM may influence the occurrence of El Niño the following winter. Ding et al. (2015b) used observational data to show that the VM triggers the occurrence of an El Niño event primarily through two processes. The first is the VM-related air–sea surface coupling process over the subtropical/tropical Pacific, which is analogous to the seasonal footprint mechanism (SFM; Vimont et al. 2001, 2003a,b; Alexander et al. 2002, 2010). In this process, the SSTAs related to the spring VM have a Pacific meridional mode (PMM)-like pattern (Chiang and Vimont 2004; Chang et al. 2007; Zhang et al. 2009; Vimont et al. 2014; Thomas and Vimont 2016) over the subtropical northeast Pacific and a significant cool SSTA pattern over the western North Pacific (WNP). Subsequently, the PMM-like anomalously warm SST pattern combined with the WNP negative anomalous SST pattern produces an anomalous zonal SST gradient in the western-central tropical Pacific, followed by surface anomalous westerly winds in this region. The anomalous westerly winds in turn amplify the SSTAs near the equator, causing an initial warming of the SST in the central-eastern equatorial Pacific in the early summer. The second process is the seasonal evolution of VM-associated anomalous subsurface ocean temperature in the tropical Pacific, which is analogous to the trade wind charging (TWC) mechanism (Anderson 2003, 2004, 2007; Anderson and Maloney 2006; Anderson et al. 2013). In this process, the VM-related anomalously warm subsurface ocean temperature gradually propagates upward and eastward from the western-central equatorial Pacific along the thermocline and reaches the eastern equatorial Pacific in summer, further contributing to the warming of SST in the eastern equatorial Pacific. The combination of these two processes eventually generates sufficiently warm SSTs in the central-eastern tropical Pacific to initiate an El Niño event.
Analysis of observations suggests that the occurrence of El Niño is intimately linked to the VM. Recent studies have indicated that including the influence of the extratropical VM can improve ENSO prediction (H.-C. Chen et al. 2020; Shi and Ding 2020). Thus, the VM can be identified as a significant precursor signal to El Niño events. Nevertheless, the performance of the current climate models in simulating the connections between the VM and El Niño is unclear. The aim of this paper is to assess the simulated connections between the boreal spring VM and the subsequent winter El Niño in the CMIP5 and CMIP6 climate models. By analyzing the VM–El Niño relationship and the associated physical dynamic processes between them in the coupled models, this study contributes to improving the understanding of tropical and extratropical interactions and the ability of climate models to predict El Niño.
The remainder of this paper is arranged as follows: section 2 introduces the datasets and methods. Section 3 assesses the ability for the CMIP5 and CMIP6 models to simulate the VM and reproduce the VM–El Niño relationship. Section 4 analyzes the factors that influence model ability to reproduce the VM–El Niño relation. Finally, a summary is presented in section 5.
2. Datasets and methodology
a. Datasets
In this study, we analyze the preindustrial simulations generated by 37 CMIP5 and 42 CMIP6 models (Taylor et al. 2012; Eyring et al. 2016). The key information for the models is presented in Table 1. For all CMIP5 models, we use the ensemble member r1i1p1, and for all CMIP6 models we use r1i1p1f1. We use monthly mean outputs for SST, surface wind, sea level pressure, and seawater potential temperature. We use the 1000-hPa wind in the model as a surrogate for the 10-m wind, following You and Furtado (2018). Since the length of the simulations varies across models, we analyze the output for the last 100 years of each run. For the observations, we use the Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST) version 1.1 (Rayner et al. 2003). Subsurface ocean temperature data are taken from the Simple Ocean Data Assimilation (SODA) version 2.2.4 (Carton and Giese 2008). Surface wind and sea level pressure are from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) Reanalysis 1 dataset (Kalnay et al. 1996). Observational data for the period 1950–2010 are used in this study.
Basic information of the CMIP5 and CMIP6 models used in this study. Expansions for many of the acronyms are available at http://www.ametsoc.org/PubsAcronymList.
b. Methodology
For all variables, monthly anomalies from the observations and model simulations are calculated by removing the respective monthly average climatology. The observed and simulated time series data are then linearly detrended. In addition, all simulations are bilinearly interpolated onto a common 1.0° × 1.0° longitude–latitude grid to allow multimodel ensemble (MME) mean analyses. The results of the MME mean are derived by simply averaging the outcomes of each model with equal weights (Krishnamurti et al. 1999; Peng et al. 2002; Fu et al. 2020).
Following previous studies (Bond et al. 2003; Ding et al. 2015a,b), we apply empirical orthogonal function (EOF) analysis to decompose the monthly SSTAs field in the North Pacific (20.5°–65.5°N, 124.5°E–100.5°W) to obtain the second SST mode (EOF2) known as the VM. In observations and models, the VM index (VMI) is defined by the normalized time series of the second principal component (PC) linked to the EOF. Since the VM tends to reach its maximum strength in the late winter to early spring [February–April (FMA)] (Ding et al. 2015a,b), we apply the FMA-averaged VMI to represent VM variability. The PMM is defined as the dominant mode of coupled variability obtained by applying maximum covariance analysis (MCA; Bretherton et al. 1992) to the tropical Pacific SST and both components of 10-m winds over the region 21°S–32°N, 175°E–95°W), following Chiang and Vimont (2004). The PMM_SST index is calculated by projecting SSTAs onto the spatial structure resulting from the MCA above. The western North Pacific SST index (WNP_SST) is defined by the mean detrended SST anomalies over 18°–28°N, 122°–132°E (Wang et al. 2012, 2013). The TWC index is defined by calculating monthly grid point SLP anomalies relative to climate values for a given month, and then normalizing these grid point anomalies by their interannual standard deviation. The normalized monthly anomalies are then regionally averaged over the boxed region (175°–140°W, 10°–25°N) (Anderson 2004; Anderson et al. 2013). Following Linkin and Nigam (2008) and Yu and Kim (2011) we defined the NPO index using the EOF method. In this study, the second mode obtained from the EOF analysis of SLP anomalies over the North Pacific (20°–60°N, 120°E–120°W) is termed the NPO, while the normalized time series of PC corresponding to EOF2 is defined as the NPO index. The Niño-3.4 index is commonly used to represent the variability of El Niño, which is defined by the SSTAs averaged from 5°S to 5°N and from 170° to 120°W (Anderson 2007; Deser et al. 2012; Chen et al. 2014, 2015). In addition, in order to avoid the possible influence of the El Niño cycle on the VM–El Niño relationship, we applied linear regression to eliminate variability in all monthly averaged variables that are linearly related to the FMA-averaged Niño-3.4 index for the same period.
3. Observed and simulated VM–El Niño relationship
a. VM–El Niño connections in the observations
To describe the observed effects of the boreal spring VM on the subsequent winter El Niño, we correlated the FMA-averaged VMI to the overlapping 3-month averaged SST and surface wind, as well as subsurface ocean temperature anomalies, to analyze the seasonal evolution of VM-related SST, surface wind anomalies [Figs. 1a–f (left panels)] and subsurface ocean temperature anomalies [Figs. 1g–l (right panels)]. Here we use the suffix (0) to refers to the current year and (+1) to indicate the following year. As shown in Fig. 1 (left panels), during FMA(0), the VM-type SSTAs spatial pattern is clearly visible over the Pacific north of 20°N, which also has some similarity with the spring Aleutian low related SSTA pattern over the Barents–Kara Seas (S. Chen et al. 2020). The VM-type SSTA pattern shows a PMM-like pattern in the subtropical northeast Pacific and a distinct anomalously cool SST pattern in the WNP. Meanwhile, VM-related anomalous southwesterlies and anomalous southeasterlies converge over the central tropical Pacific, which contributes to the development of anomalously warm SST there. In addition, positive SST anomalies in the subtropical central North Pacific can also force significant atmospheric heating anomalies, promoting the occurrence of westerly wind anomalies in the central and western tropical Pacific through a Gill-type atmospheric response (Gill 1980; Xie and Philander 1994; Vimont et al. 2003a,b; Chen et al. 2014). During April–June [AMJ(0)], warm anomalous SST in the central tropical Pacific and cool anomalous SST in the WNP together result in an enhanced zonal anomalous SST gradient in the western-central tropical Pacific, thereby strengthening the anomalous westerly winds there. Approximately 11 months after the VM peaks, an El Niño–like warm SST pattern forms over the central-eastern tropical Pacific.
Correlation maps of the FMA(0)-averaged VMI with (left) the 3-month averaged SST (shaded) and surface wind (vectors) anomalies and (right) subsurface ocean temperature anomalies at different depths in meters averaged over 5°S–5°N for (a),(g) FMA(0), (b),(h) AMJ(0), (c),(i) JJA(0), (d),(j) ASO(0), (e),(k) OND(0), and (f),(l) DJF(+1). Positive (red) and negative (blue) SSTAs and subsurface ocean temperature anomalies, with correlation significant at the 95% level are shaded. Only surface wind vectors significant at the 95% level are shown.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
Additionally, the VM can also cause anomalies in subsurface ocean temperatures in the equatorial Pacific. At the peak of the VM, significant warm anomalous subsurface ocean temperatures occur at depths of about 50–250 m in the equatorial Pacific east-central region (180°–120°W) (Fig. 1, right panels), which appears to be caused by anomalous zonal wind stress. The VM-related off-equatorial westerly stress anomalies weaken the prevailing trade winds, which when combined with anomalous equatorial easterly winds produce off-equatorial anticyclonic curl in the central Pacific, causing downward Ekman pumping and equatorward transport of water, leading to deepening of the thermocline and the formation of positive subsurface temperature anomalies in the central equatorial Pacific (Clarke et al. 2007; Anderson 2007; Anderson et al. 2013; Anderson and Perez 2015; Chakravorty et al. 2020). Subsequently, the VM-associated anomalously warm subsurface ocean temperature gradually propagates upward and eastward along the thermocline and reach the surface of the eastern equatorial Pacific during June–August [JJA(0)], further raising SST in the eastern equatorial Pacific. Further enhanced air-sea coupling in the tropical Pacific region generates sufficient equatorial anomalous westerlies to trigger the positive Bjerknes feedback, allowing further eastward warming of subsurface ocean temperature (Figs. 1k,l). Finally, sufficient warm SSTs are generated in the central-eastern tropical Pacific for an El Niño to occur.
The results of the above analysis show that the VM is a basin-scale SST pattern in the North Pacific. The VM combines the effects of surface air–sea interactions linked to the PMM-like and WNP SST anomaly patterns as well as the ocean dynamics associated with the VM, effectively acting like an ocean bridge by which atmospheric variability of extratropical over the North Pacific can impact El Niño. We next investigate the spring VM–El Niño connections under preindustrial forcing in the CMIP5 and CMIP6 climate models to examine whether these climate models can reproduce the results and processes described above.
b. Performance of the CMIP5 and CMIP6 models in reproducing the VM
To evaluate the relationship of the VM with El Niño in the models, we first evaluate the models’ ability to reproduce the VM. Only the models that reproduce the VM well are used to examine further the performance of the models in simulating the VM–El Niño relationship in the next section. Figure 2 shows the EOF2 of SSTAs for the North Pacific from the observations and CMIP5 and CMIP6 preindustrial outputs, which represent the observed and simulated VMs. In the observations, the most notable characteristic of the VM is the northeast–southwest-tilted SSTAs dipole structure, characterized by a warm SSTAs band extending across the Pacific from off California into the western Bering Sea with a cool SSTA band that extends from the central North Pacific to the western North Pacific. This spatial pattern explains 12.2% of the overall variance (Fig. 2a), which is similar to previous results (Bond et al. 2003; Ding et al. 2015b). From the simulation results of the VM pattern with the CMIP5 (models b1–b37, listed in Fig. 2) and CMIP6 models (models c1–c42, listed in Fig. 2), most CMIP5 and CMIP6 climate models can simulate the tilted SSTAs dipole structure over the extratropical North Pacific, reproducing well the main spatial features of the VM. However, the tilt direction of the SSTAs dipole is slightly different from the observations in most models.
Spatial patterns of the second EOF mode of the SST anomalies over the North Pacific (20.5°–65.5°N, 124.5°E–100.5°W) calculated from (a) observation, (b1)–(b37) 37 CMIP5 models, and (c1)–(c42) 42 CMIP6 models (models are referred to in the paper by these numbers). The numbers at the upper right of each panel indicate the percentage of variance explained by each model’s second EOF mode.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
We then use a Taylor diagram (Taylor 2001) to give a quantitative measure of the differences between model simulations and observations and to make intuitive comparisons of the different models. Most of the pattern correlation coefficients between the modeled and observed VM patterns are at least 0.50 (Fig. 3a). The average pattern correlation coefficients for the CMIP5 models and CMIP6 models are 0.76 and 0.78, respectively, indicating that the CMIP5 and CMIP6 models generally give a good representation of the VM pattern. However, there are noticeable biases in the simulated amplitude of the observed VM pattern. Most CMIP5 and CMIP6 models overestimate the amplitude of the VM pattern, with the ratio between the modeled and observed standard deviation of the VM pattern greater than 1.0.
(a) Comparative evaluation of the spatial patterns of VM between the observations and CMIP5 (blue) and CMIP6 (red) model outputs. Each numbered dot in the diagram represents a single model (see Fig. 2 for which model each number represents). Black dashed lines and arcs indicate the correlation coefficient and the ratio of standard deviations between modeled and observed VM patterns, respectively. Black solid arcs show the centered root-mean-square difference. (b) Histograms of 10 000 realizations of a bootstrap method for pattern correlation coefficient between the observed and simulated VM. The black vertical lines indicate the mean values of 10 000 interrealizations for the CMIP5 and CMIP6 models. The gray shaded regions represent the 5% and 95% percentile of the 10 000 interrealizations.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
Combining the results of Figs. 2 and 3, in order to investigate the VM–El Niño relationship in the CMIP5 and CMIP6 models we screened out the models with pattern correlation coefficients between the simulated and the observed VM less than 0.72. [0.72 is the lower limit of the 95% confidence interval for the pattern correlation coefficient between the modeled and observed VM, obtained using the bootstrap method (Hennemuth et al. 2013; Jia et al. 2019; see Fig. 3b). Therefore, five CMIP5 and ten CMIP6 models [(b9) CMCC-CESM, (b11) CNRM-CM5-2, (b21) GISS-E2-R-CC, (b29) MIROC-ESM-CHEM, (b30) MIROC-ESM, (c1) ACCESS-CM2, (c4) AWI-ESM-1-1-LR, (c14) CanESM5, (c15) E3SM-1-0, (c29) INM-CM4-8, (c30) INM-CM5-0, (c38) NESM3, (c39) NorCPM1, (c40) NorESM2-LM, and (c41) NorESM2-MM] are no longer used in the following analyses. Next, we further examined the ability of these selected models to simulate ENSO seasonal phase locking (not shown). The results show that among the 32 CMIP5 modes, except IPSL-CM5A-LR and IPSL-CM5A-MR, the Niño-3.4 index variance peaks correctly in the boreal winter or early spring. In contrast, six of the 32 CMIP6 models (GISS-E2-1-G-CC, GISS-E2-1-G, GISS-E2-1-H, MCM-UA-1-0, MPI-ESM-1-2-HAM, and MPI-ESM1-2-HR) do not correctly simulate the maximum season of Niño-3.4 index variance.
Besides, previous studies have explored the relationship between NPO atmospheric forcing and the VM and quantified the variability of the VM resulting from NPO, suggesting that winter NPO is an important driver of the following spring VM. (Ding et al. 2015b). We therefore also checked the ability of these filtered models to reproduce the NPO spatial pattern (not shown). The results are consistent with Wang et al. (2019); we find that most CMIP5 models, except MRI-CGCM3, can well reproduce the meridional dipole pattern of the NPO. However, the meridional dipole in many models shows a longitudinal displacement bias compared with the observations. In contrast, the simulation results of the CMIP6 models for the spatial pattern and intensity of the NPO show an improvement compared to the CMIP5 models. Also, the spatial correlation coefficients of all 32 CMIP6 model simulations are greater than 0.7. A recent study also demonstrated that the CMIP6 model can reasonably capture the NPO patterns (Chen et al. 2021).
Summarizing the above analysis, due to some models’ poor ability to simulate the ENSO seasonal phase locking (CMIP5: IPSL-CM5A-LR and IPSL-CM5A-MR; CMIP6: GISS-E2-1-G-CC, GISS-E2-1-G, GISS-E2-1-H, MCM-UA-1-0, MPI-ESM-1-2-HAM, MPI-ESM1-2-HR), and one model’s poor performance in simulating NPO (CMIP5: MRI-CGCM3), the outputs of these models will not be used in the following analysis of the VM–El Niño relationship. Therefore, after evaluating the ability of the models to simulate VM, ENSO seasonal phase locking, and NPO, we finally selected 29 CMIP5 models and 26 CMIP6 models for the analysis of the VM–El Niño relationship evaluated next in the article. Figure 4 shows the simulated VM–NPO relationships for the final selection of these models. The results show that the final selection of 29 CMIP5 models and 26 CMIP6 models both successfully capture the relationship when the NPO precedes the VM by one month and the correlation between the two reaches its strongest. This is consistent with the VM reaching its greatest strength in FMA as a delayed response to the NPO forcing as seen in the observations.
Lead–lag correlation coefficients between the monthly VMI and NPO index for (a) observation, (b1)–(b29) CMIP5 models, and (c1)–(c26) CMIP6 models; the horizontal dashed line indicates the 99.9% confidence level.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
c. The simulated VM–El Niño relationship
In this section, we examine the performance of the selected CMIP5 and CMIP6 models for modeling the relationship between the boreal spring VM and the subsequent winter El Niño.
We first examined the time during which VM has a strongest relation with the winter El Niño in the CMIP models. This is because in the observations, the VM in the FMA has the strongest relationship with the next winter El Niño. However, in the CMIP model, VM may produce the strongest relationship between MAM (or AMJ) and winter El Niño. We compute lead–lag correlations between the following winter [DJF(+1)]-averaged Niño-3.4 index and 3-month averaged VMI in the models (Fig. 5). The results shows that most of the CMIP5 and CMIP6 models exhibit the VM at FMA with the closest relationship to the next winter El Niño. Fourteen of the 29 CMIP5 patterns are able to simulate that the VM at FMA was most closely related to the next winter El Niño. This compares to 16 of the 26 CMIP6 models. In addition, when we count the results of the models that exhibit the strongest relationship between VM in other seasons and the following winter El Niño, we find that although the results of these models show that VM is most closely related to the following winter El Niño in other seasons, their correlation coefficients do not differ significantly from the correlation coefficients of the FMA VM with the following winter El Niño. Moreover, there are no model results that show that the correlation coefficient of the FMA-averaged VMI with the DJF(+1)-averaged Niño-3.4 index does not pass the significance test, while the correlation coefficient of VMI in the most closely related season with the DJF(+1)-averaged Niño-3.4 index passes the significance test. These results then ensure that we can use the correlation between the FMA-averaged VMI and the DJF(+1)-averaged Niño-3.4 index uniformly in the model to explore the relationship between VM and El Niño in the model without affecting our conclusions.
Correlation coefficients between the following winter [DJF(+1)]-averaged Niño-3.4 index and 3-month [JFM(0), FMA(0), MAM(0), AMJ(0), MJJ(0), JJA(0)] averaged VMI in the models. Gray (correlation coefficient below 0.20) indicates insignificance at the 95% confidence level according to the two-tailed Student’s t test.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
Figure 6 shows the following winter SSTAs and surface wind anomalies fields regressed onto the normalized FMA-averaged VMI for the observations, and for the CMIP5 and CMIP6 models. MME averages of the CMIP5 (Fig. 6b) and CMIP6 (Fig. 6c) models are also shown in Fig. 6, respectively. To further quantify the VM–El Niño relationship, we calculate the correlation and regression coefficients for the FMA(0)-averaged VMI and the DJF(+1)-averaged Niño-3.4 index for each model and for the MME means as shown in Figs. 7 and 8, respectively.
Regression maps of the FMA-averaged VMI with the following DJF-averaged SST (°C; shaded), and surface wind anomalies (m s−1; vectors) in (a) observation, (b1)–(b29) 29 CMIP5 models, (c1)–(c26) 26 CMIP6 models, and the multimodel ensemble (MME) mean of (b0) CMIP5 and (c0) CMIP6. Only SST and surface wind anomalies significant at the 95% confidence level according to a two-tailed Student’s t test are shown.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
Correlation coefficients of the FMA(0)-averaged VMI with the DJF(+1)-averaged Niño-3.4 index in the 29 CMIP5 models (blue), 26 CMIP6 models (red), and the multimodel ensemble (MME) means of the CMIP5 and CMIP6. The horizontal dashed line shows the 95% confidence level. The error bars in the MME means are calculated as two standard deviations (a 95% confidence interval based on normal distribution) of the 10 000 interrealizations of a bootstrap method.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
(a) Regression coefficients of the FMA(0)-averaged (FMA0) VMI with the DJF(+1)-averaged Niño-3.4 index in the 29 CMIP5 models (blue), 26 CMIP6 models (red), and the MME means of the CMIP5 and CMIP6. The horizontal dashed line shows the 95% confidence level. The error bars in the MME means are calculated as two standard deviations (a 95% confidence interval based on normal distribution) of the 10 000 interrealizations of a bootstrap method. (b) Histograms of 10 000 realizations of a bootstrap method for the Niño-3.4 region SST response. The Niño-3.4 region SST response is measured by the regression coefficients of the FMA(0)-averaged VMI with the DJF(+1)-averaged Niño-3.4 index. The blue and red vertical lines indicate the mean values of 10 000 interrealizations for the CMIP5 and CMIP6 models, respectively. The gray shaded regions represent the 1% and 99% percentiles of the 10 000 interrealizations.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
In the observations (Fig. 6a), during a positive VM phase, significantly warm SSTAs are observed in the central-eastern tropical Pacific, with significantly cool SSTAs in the western tropical Pacific, extending northeastward and southeastward into the subtropical Pacific. At the same time, significant VM-related anomalous westerly and easterly winds converge over the central equatorial Pacific. The anomalous pattern shown in Fig. 6a is very similar to the pattern observed in an El Niño event.
Combining the results of Figs. 6 and 7, 13 of the CMIP5 models [Fig. 6, (b3) BCC-CSM1-1-m, (b5) CCSM4, (b6) CESM1-BGC, (b7) CESM1-CAM5, (b8) CESM1-FASTCHEM, (b9) CMCC-CMS, (b12) CanESM2, (b17) GFDL-ESM2M, (b19) GISS-E2-R, (b24) MIROC5, (b26) MPI-ESM-MR, (b28) NorESM1-ME, and (b29) NorESM1-M] are capable of reproducing to some degree the observed spring VM-associated SSTAs and surface wind anomalies patterns in the subsequent winter over the central-eastern tropical Pacific. The remaining 16 models fail to reproduce the VM–El Niño connections, with weaker spring VM-associated SSTAs and surface wind anomalies over the tropical Pacific, and correlations for the FMA(0)-averaged VMI and the DJF(+1)-averaged Niño-3.4 index that are not statistically significant at the 95% confidence level. In contrast, 22 of the 26 CMIP6 models (Fig. 6) simulate well the observed anomalous SST and surface wind patterns related to the spring VM in the subsequent winter over the tropical Pacific with correlations for the FMA(0)-averaged VMI and the DJF(+1)-averaged Niño-3.4 index that are significant at the 95% confidence level. Only four CMIP6 models are unable to do this (Fig. 6): (c4) BCC-ESM1, (c16) EC-Earth3-Veg, (c18) FGOALS-g3, and (c22) IPSL-CM6A-LR].
When averaged separately for all models in both model groups, the CMIP5 MME mean (Fig. 6b) exhibits weaker spring VM-related SSTAs and surface wind anomalies patterns in the subsequent winter over the tropical Pacific. In contrast, the pattern of the CMIP6 MME mean (Fig. 6c) is similar to the observations. The MME means of the DJF(+1)-averaged Niño-3.4 index regressed onto the FMA(0)-averaged VMI show that the CMIP6 models give a significant improvement on the CMIP5 models. The multimodel average value for the CMIP5 models is 0.197°C, whereas that of the CMIP6 models is 0.403°C, with a 51% increase from the CMIP5 to the CMIP6 models (Fig. 8a). The difference of the two averages is statistically significant above the 99% confidence level, according to the bootstrap method (Fig. 8b). These results show that the ability of the CMIP6 models to simulate the VM–El Niño relationship is significantly improved compared with the CMIP5 models.
4. Possible factors influencing the simulated VM–El Niño relationship
The results of the above analyses show that there are significant differences in the ability of the CMIP5 and CMIP6 models to emulate the VM–El Niño relationship. As described in section 3a, the two main factors affecting the VM–El Niño relationship are the surface air–sea interactions linked to the VM over the subtropical/tropical Pacific and the seasonal evolution of VM-related anomalous subsurface ocean temperature at the equator. Therefore, we consider that the ability of the models to simulate these two processes may be the main reason for intermodel differences in the simulation of the VM–El Niño relationship. We next discuss these two physical processes in the CMIP5 and CMIP6 models.
a. VM-related surface air–sea coupling process in the CMIP5 and CMIP6 models
Since the strongest VM variability occurs in FMA, we use the FMA(0)-averaged PMM_SST and WNP_SST indices to represent the variability of the subtropical northeast Pacific and WNP SSTAs patterns, respectively. We first calculate the correlation coefficients between the FMA(0)-averaged PMM_SST index or the WNP_SST index and the DJF(+1)-averaged Niño-3.4 index separately to represent the relationships between the PMM-like SSTAs pattern and the WNP SSTAs pattern and El Niño. We next link the VM–El Niño correlation to the PMM–El Niño connections and to the WNP–El Niño connections. The results of the CMIP5 and CMIP6 models are presented in Fig. 9. The CMIP5 and CMIP6 models that produce a stronger VM–El Niño correlation tend to have stronger PMM–El Niño connections and stronger WNP–El Niño connections. In the CMIP5 and CMIP6 models, the correlation coefficients between the VM–El Niño correlation and the PMM–El Niño correlation are 0.64 and 0.66 (both statistically significant above the 99% confidence level), respectively (Figs. 9a,d), and the correlation coefficients between the VM–El Niño correlation and the WNP–El Niño correlation are 0.69 and 0.74 (both statistically significant above the 99% confidence level), respectively (Figs. 9b,e). We then measure the contribution of joint PMM_SST and WNP_SST forcing on El Niño to the VM–El Niño correlation in CMIP5 and CMIP6 models. In the CMIP5 models the VM–El Niño correlation based on the joint PMM_SST and WNP_SST forcing on El Niño yields a correlation coefficient of 0.75 (significant at the 99% confidence level; Fig. 9c), while in the CMIP6 models the correlation is 0.76 (significant at the 99% confidence level; Fig. 9f). These results support the previous conclusions from the analysis of observational data, namely that surface air–sea interaction processes linked to the VM are one of the main factors influencing the VM–El Niño relationship. This also demonstrates that most CMIP5 and CMIP6 models are very capable of simulating the role of the coupled surface air–sea process related to the VM linking the VM to El Niño.
(a) Scatterplot of the CMIP5 models between correlation coefficients of FMA(0)-averaged PMM_SST index with the DJF(+1)-averaged Niño-3.4 index and correlation coefficients of FMA(0)-averaged VMI with the DJF(+1)-averaged Niño-3.4 index. The box enclosed by a dashed red line in the upper right corner identifies a region in which the correlation coefficients pass the 95% confidence level test. The best fitting line is represented by the black solid line. (b) As in (a), but for the correlation coefficients of FMA(0)-averaged WNP_SST index with the DJF(+1)-averaged Niño-3.4 index and correlation coefficients of FMA(0)-averaged VMI with the DJF(+1)-averaged Niño-3.4 index. (c) Scatterplot with regression fitting plane of the correlation coefficients of FMA(0)-averaged VM, PMM_SST, and WNP_SST indices with the DJF(+1)-averaged Niño-3.4 index. (d)–(f) As in (a)–(c) respectively, but for the CMIP6 models.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
Furthermore, the correlation of the VM and El Niño and the correlation of the PMM and El Niño are both significant at a 95% confidence level in about 34% (10 out of 29) of the CMIP5 models (the models in the boxes enclosed by red dashed lines in the upper right corner of Fig. 9a), and in about 73% (19 out of 26) of the CMIP6 models (Fig. 9d). At the same time, the correlation of the VM and El Niño and the correlation of the WNP and El Niño are both significant at a 95% confidence level in about 41% (12 out of 29) of the CMIP5 models (the models in the boxes enclosed by red dashed lines in the upper right corner of Fig. 9b) in contrast to about 77% (20 out of 26) of the CMIP6 models (Fig. 9e). This result shows that the CMIP6 models are better able to simulate the coupled surface air–sea processes over the subtropical/tropical Pacific region related to the VM than the CMIP5 models, which partly explains why the CMIP6 models simulate the VM–El Niño relationship better than the CMIP5 models.
The above results showed the role of the PMM-like and WNP SST anomaly patterns in the VM–El Niño relationship, and this further analysis of the CMIP5 and CMIP6 models is presented to show the role of the VM-related surface air–sea interaction process in the VM-induced El Niño process. Based on the correlation coefficients of the FMA(0)-averaged VMI and the DJF(+1)-averaged Niño-3.4 index, we classify models with correlation coefficients passing the 95% significance test as the “good performance” (GP) model group, and the rest as the “poor performance” (PP) model group in the CMIP5 and CMIP6 models. The MME means of the seasonal evolution of the VM-associated anomalous SST and surface wind from the GP and PP group models are then calculated for CMIP5 and CMIP6 separately (Fig. 10). Comparing the PP model results with the GP model results in the CMIP5 and CMIP6 models shows that the models that can simulate the VM–El Niño relationship are the models that successfully simulate the VM-related surface air–sea coupling processes seen in the observational data. However, models that cannot capture the VM–El Niño relationship give results for the surface air–sea coupling processes related to the VM significantly different from those of the observations and the GP model group. In the PP model group in FMA, the SSTA warmth related to the VM over the subtropical northeast Pacific is weak and does not extend to the equatorial Pacific. At the same time, the accompanying anomalous surface winds are also weak and fail to converge over the central tropical Pacific, so that the warm SSTAs in the central tropical Pacific cannot develop. Alternatively, the intensity and area of cool SSTAs in the WNP region are also smaller than in the GP models. Thus, the anomalous zonal SST gradient over the western-central tropical Pacific is weaker than in the GP models. As a result the warm SSTs and strong anomalous westerly winds required to initiate positive Bjerknes feedback (Bjerknes 1969) do not develop in the tropical Pacific during summer. Consequently, El Niño events are not induced in the subsequent winter. This result further shows that the surface air–sea thermodynamic coupling processes related to the VM play a significant role in the VM–El Niño relationship. It also shows that the model simulation of these processes is one of the main reasons for the differences between models in simulating the VM–El Niño connections.
Regression maps of the FMA(0)-averaged VMI with the 3-month averaged SST (shading; °C) and surface wind (vectors; m s−1) anomalies for FMA(0), JJA(0), SON(0), and DJF(+1). Only areas exceeding the 95% confidence level are displayed. The two left columns show the ensemble mean of the (a)–(d) “poor performance” (PP) and (e)–(h) “good performance” (GP) group models in the CMIP5 models, respectively. The two right columns show the ensemble mean of the (a)–(d) PP and (e)–(h) GP group models in the CMIP6 models, respectively.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
b. Equatorial subsurface ocean temperature evolution related to the VM in the CMIP5 and CMIP6 models
In addition to the surface air–sea coupling process discussed above, the seasonal evolution of the VM-related equatorial subsurface ocean temperature anomalies is also an essential factor in the VM–El Niño relationship. In this section, we explore the ability of the CMIP5 and CMIP6 models to emulate the seasonal evolution of the VM-related equatorial subsurface ocean temperature anomalies and their role in linking the VM to El Niño. In the analysis in section 3a, it is shown that VM can force subsurface ocean temperature anomalies in the equatorial Pacific, producing anomalously high upper ocean heat content and increasing SST in the eastern equatorial Pacific, with the TWC mechanism playing an important role in this process. We first calculate the correlation coefficients between the FMA(0)-averaged TWC index and the DJF(+1)-averaged Niño-3.4 index to represent the relationships between the TWC mechanism and El Niño. We next link the VM–El Niño correlation to the TWC–El Niño connections. The results of the CMIP5 and CMIP6 models are presented in Figs. 11a and 11b. It can be seen that the CMIP5 and CMIP6 models that produce a stronger VM–El Niño correlation tend to have stronger TWC–El Niño connections. In the CMIP5 and CMIP6 models, the correlation coefficients between the VM–El Niño correlation and the TWC–El Niño correlation are 0.61 and 0.68 (both statistically significant above the 99% confidence level), respectively. These results support the conclusion from previous analysis of the observations that the process of the TWC mechanism is one of the key factors affecting the VM–El Niño relationship. This also indicates that most CMIP5 and CMIP6 models are able to simulate the role of TWC mechanisms linking the VM to El Niño. Further analysis shows that the VM–El Niño correlation and the TWC–El Niño correlation are both significant at 95% confidence level for about 41% (12 out of 29 models) in the CMIP5 model and about 81% (21 out of 26 models) in the CMIP6 models. This result indicates that the CMIP6 models has improved the ability to simulate the process of VM induced El Niño through the TWC mechanism compared to the CMIP5 models.
Scatterplot of the CMIP5 models between their correlation coefficients of FMA(0)-averaged TWC index with the DJF(+1)-averaged Niño-3.4 index and correlation coefficients of FMA(0)-averaged VMI with the DJF(+1)-averaged Niño-3.4 index. The box enclosed by a dashed red line in the upper right corner identifies a region in which the correlation coefficients pass the 95% confidence level test. The best fitting line is represented by the black solid line. (b) As in (a), but for the CMIP6 models. (c) Scatterplot of the CMIP5 models between their correlation coefficients of FMA(0)-averaged VMI with the DJF(+1)-averaged E-WWV (eastern equatorial Pacific warm water volume) index and correlation coefficients of FMA(0)-averaged VMI with the DJF(+1)-averaged Niño-3.4 index. The box enclosed by a dashed red line in the upper right corner identifies a region in which the correlation coefficients pass the 95% confidence level test. The best fitting line is represented by the black solid line. (d) As in (c), but for the CMIP6 models.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
Subsequently we calculated the correlation between the FMA(0)-averaged VMI and DJF(+1)-averaged equatorial eastern Pacific warm water volume (EWWV; 5°N–5°S, 155°–80°W) and then linked it to the VM–El Niño relationship (Figs. 11c,d). The results show that the VM–El Niño relationship simulated by the CMIP5 and CMIP6 models is very closely related to the VM–EWWV correlation. In addition, the correlation of the VM and El Niño as well as the correlation of the VMI and EWWV are both significant at the 95% confidence level in about 38% (11 out of 29) of the CMIP5 models, and in about 77% (20 out of 26) of the CMIP6 models (the models in the boxes enclosed by red dashed lines in the upper right corner of Figs. 11c and 11d, respectively). The analysis of the results in Fig. 11 above shows that the CMIP6 models outperform the CMIP5 models results for the simulation of VM affecting El Niño through the VM-associated ocean dynamics processes (TWC as well as eastward propagation of Kelvin waves).
Next, the role of the seasonal evolution of VM-associated anomalous subsurface ocean temperature in the occurrence of VM-induced El Niño is further examined in the CMIP5 and CMIP6 models, as well as the composite differences between the CMIP5 and CMIP6 models for the impact of the VM on El Niño through ocean dynamics processes. We again classify the models with correlation coefficients exceeding 95% significance for the VMI and Niño-3.4 index as the “good performance” (GP) model group, and the rest as the “poor performance” (PP) model group for the CMIP5 and the CMIP6 models. The MME means of the seasonal evolution of VM-related subsurface ocean temperature anomalies are then computed separately for the GP and PP group models. A comparison of the results in Fig. 12 shows that the CMIP5 PP models cannot simulate the seasonal evolution of VM-related subsurface ocean temperature anomalies. However, the GP models showed similar results to the observed data analysis for the seasonal evolution of these VM-related anomalies (see Fig. 1, right panel). The same results are also found for the CMIP6 models. This suggests that the VM-related seasonal evolution of subsurface ocean temperature anomalies plays an influential role in the VM-induced occurrence of El Niño. Also, comparison of the results of the CMIP5 GP and CMIP6 GP models shows that the average value of the VM-associated positive anomalous subsurface ocean temperature simulated by the CMIP6 GP models is larger than for the CMIP5 GP models, although there are more GP models in CMIP6. Thus the CMIP6 models have an improved ability to simulate the evolution of VM-related ocean subsurface temperature anomalies compared to the CMIP5 models. This result also supports the conclusion of our previous analysis that the CMIP6 models simulate the VM–El Niño relationship better than the CMIP5 models.
Regression maps of the FMA(0)-averaged VMI with the 3-month averaged subsurface ocean temperature anomalies (shaded; °C) at different depths in meters averaged over 5°S–5°N for FMA(0), JJA(0), SON(0), and DJF(+1). Only areas exceeding the 95% confidence level are displayed. Thick black contour indicates the climatological position of the 23°C isotherm. The two left columns show the ensemble mean of the (a)–(d) “poor performance” (PP) and (e)–(h) “good performance” (GP) group models in the CMIP5 models, respectively. The two right columns represent the ensemble mean of the (a)–(d) PP and (e)–(h) GP group models in the CMIP6 models, respectively.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
Finally, we explore the relative importance of the two processes discussed above (the VM-related surface air–sea interactions and the VM-related ocean dynamics) for the VM-induced El Niño occurrence. We have made the comparisons of the composite (similar to Figs. 10 and 12) of model simulate only thermodynamic coupling [FMA(0)-averaged PMM_SST and DJF(+1)-averaged Niño-3.4 index with correlation coefficients above 95% significance], only dynamical coupling [FMA(0)-averaged VMI and DJF(+1)-averaged EWWV with correlation coefficients above 95% significance], and both in the CMIP5 and CMIP6 models, respectively. The results (Figs. 13 and 14) show that the ocean dynamical processes associated with the evolution of the VM-related subsurface ocean temperature anomalies along the equator play a more pronounced role in the VM-induced generation of El Niño than the role of VM-related surface air–sea coupling. The model composites simulating only the thermodynamic coupling show that the magnitude of the response to anomalously warm SSTs in the central-eastern equatorial Pacific during the next winter appears small compared to the models simulating only the dynamical coupling and both, and it seems to occur mainly in the central-western equatorial region; furthermore, the corresponding subsurface ocean temperature anomalies appear only in the equatorial central Pacific, with no trend of eastward propagation.
Regression maps of the FMA(0)-averaged VMI with the 3-month averaged SST (shading; °C) and surface wind (vectors; m s–1) anomalies for FMA(0), JJA(0), SON(0), and DJF(+1). Only areas exceeding the 95% confidence level are displayed. The three left columns represent the ensemble mean of the CMIP5 models of (a)–(d) only “dynamical coupling” [FMA(0)-averaged VMI and DJF(+1)-averaged EWWV with correlation coefficients above 95% significance] group models and (e)–(h) the “only thermodynamic coupling” [FMA(0)-averaged PMM_SST and DJF(+1)-averaged Niño-3.4 index with correlation coefficients above 95% significance] group models, as well as (i)–(l) both of the above 95% significance group models. The three right columns are the same as the three left columns, but for the CMIP6 models.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
Regression maps of the FMA(0)-averaged VMI with the 3-month averaged subsurface ocean temperature anomalies (shaded; °C) at different depths in meters averaged over 5°S–5°N for FMA(0), JJA(0), SON(0), and DJF(+1). Only areas exceeding the 95% confidence level are displayed. The thick black contour indicates the climatological position of the 23°C isotherm. The three left columns show the ensemble mean of the CMIP5 models of (a)–(d) only “dynamical coupling” [FMA(0)-averaged VMI and DJF(+1)-averaged EWWV with correlation coefficients above 95% significance] group models and (e)–(h) the “only thermodynamic coupling” [FMA(0)-averaged PMM_SST and DJF(+1)-averaged Niño-3.4 index with correlation coefficients above 95% significance] group models, as well as (i)–(l) both of the above 95% significance group models. The three right columns are the same as the three left columns, but for the CMIP6 models.
Citation: Journal of Climate 34, 18; 10.1175/JCLI-D-20-0927.1
5. Summary
Previous studies have shown that the boreal spring VM can effectively induce the occurrence of El Niño in the subsequent winter (Ding et al. 2015a,b). However, the capacity of current climate models to simulate the VM–El Niño relationship is unclear. This study evaluates the ability of 37 CMIP5 and 42 CMIP6 models to simulate the effect of the spring VM in inducing the occurrence of El Niño in the following winter, and analyzes the reasons for the differences in simulation performance between the CMIP5 and CMIP6 models.
Before evaluating the simulated the VM–El Niño relationship in the CMIP5 and CMIP6 models, we first examined the performance of each model in simulating the VM. We evaluated the simulations of the CMIP5 and CMIP6 models for the spatial pattern as well as the strength of the VM (Figs. 2 and 3). Thirty-two of the 37 CMIP5 models reproduce well the VM’s spatial pattern [the exceptions are (b9) CMCC-CESM, (b11) CNRM-CM5-2, (b21) GISS-E2-R-CC, (b29) MIROC-ESM-CHEM, and (b30) MIROC-ESM]. Thirty-two of the 42 CMIP6 models also simulate the VM with a relatively good spatial pattern. In the other 10 models [(c1) ACCESS-CM2, (c4) AWI-ESM-1-1-LR, (c14) CanESM5, (c15) E3SM-1-0, (c29) INM-CM4-8, (c30) INM-CM5-0, (c38) NESM3, (c39) NorCPM1, (c40) NorESM2-LM, and (c41) NorESM2-MM] the spatial correlation coefficients between the simulated and observed VMs is less than 0.72. However, there are some biases in the model results with respect to the spatial structure of the VM in the observations. According to the bootstrap method, we select the CMIP5 and CMIP6 models with spatial correlation coefficients greater than 0.72 between the simulated and observed VMs for the following analysis. We then examined the ability of these models, which are well simulated for VM, to simulate the ENSO seasonal phase locking as well as the NPO spatial patterns. It turns out that the output of these modes will not be used in the following analysis of the VM–El Niño relationship because of the poor simulation of the El Niño seasonal phase locking (CMIP5: IPSL-CM5A-LR and IPSL-CM5A-MR; CMIP6: GISS-E2-1-G-CC, GISS-E2-1-G, GISS-E2-1-H, MCM-UA-1-0, MPI-ESM-1-2-HAM, and MPI-ESM1-2-HR) and the NPO (CMIP5: MRI-CGCM3). Therefore, after evaluating the ability of the models to simulate VM, ENSO seasonal phase locking, and NPO, we finally selected 29 CMIP5 models and 26 CMIP6 models for the analysis of the VM–El Niño relationship. Our results indicate that the ability of the CMIP6 models to simulate the VM–El Niño relationship is superior to the CMIP5 models. The combined results of Figs. 6 and 7 show that about 45% (13 out of 29) of the CMIP5 models can reproduce the significant positive correlation between the positive phase VM and El Niño, compared with about 85% (22 out of 26) of the CMIP6 models. The MME means of the DJF(+1)-averaged Niño-3.4 index regressed onto the FMA(0)-averaged VMI shows an 51% increase from 0.197°C in the CMIP5 models to 0.403°C in the CMIP6 models. This improvement in CMIP6 is statistically significant above the 99% confidence level, according to the bootstrap method (Fig. 8b).
Finally, we investigate the possible reasons for the improved ability of the CMIP6 models to simulate the VM–El Niño relationship as compared to the CMIP5 models (Figs. 9, 10, 11, and 12). Overall, the CMIP6 models are significantly better than the CMIP5 models in simulating the VM-related surface air–sea thermodynamic coupling over the subtropical/tropical Pacific as well as the seasonal evolution of VM-related anomalous subsurface ocean temperature in the equatorial Pacific, which are two key physical processes that determine the VM–El Niño relationship. The improved simulation of these two key processes in the CMIP6 models is responsible for the improved performance of CMIP6 models in simulating the VM–El Niño relationship compared with the CMIP5 models.
Acknowledgments
This research was jointly supported by the National Natural Science Foundation of China (41975070), and the State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences (Project No. LTO1901).
REFERENCES
Alexander, M. A., I. Bladé, M. Newman, J. R. Lanzante, N.-C. Lau, and J. D. Scott, 2002: The atmospheric bridge: The influence of ENSO teleconnections on air–sea interaction over the global oceans. J. Climate, 15, 2205–2231, https://doi.org/10.1175/1520-0442(2002)015<2205:TABTIO>2.0.CO;2.
Alexander, M. A., D. J. Vimont, P. Chang, and J. D. Scott, 2010: The impact of extratropical atmospheric variability on ENSO: Testing the seasonal footprinting mechanism using coupled model experiments. J. Climate, 23, 2885–2901, https://doi.org/10.1175/2010JCLI3205.1.
Anderson, B. T., 2003: Tropical Pacific sea-surface temperatures and preceding sea level pressure anomalies in the subtropical North Pacific. J. Geophys. Res., 108, 4732, https://doi.org/10.1029/2003JD003805.
Anderson, B. T., 2004: Investigation of a large-scale mode of ocean–atmosphere variability and its relation to tropical Pacific sea surface temperature anomalies. J. Climate, 17, 4089–4098, https://doi.org/10.1175/1520-0442(2004)017<4089:IOALMO>2.0.CO;2.
Anderson, B. T., 2007: On the joint role of subtropical atmospheric variability and equatorial subsurface heat content anomalies in initiating the onset of ENSO events. J. Climate, 20, 1593–1599, https://doi.org/10.1175/JCLI4075.1.
Anderson, B. T., and E. Maloney, 2006: Interannual tropical Pacific sea surface temperatures and their relation to preceding sea level pressures in the NCAR CCSM2. J. Climate, 19, 998–1012, https://doi.org/10.1175/JCLI3674.1.
Anderson, B. T., and R. C. Perez, 2015: ENSO and non-ENSO induced charging and discharging of the equatorial Pacific. Climate Dyn., 45, 2309–2327, https://doi.org/10.1007/s00382-015-2472-x.
Anderson, B. T., R. C. Perez, and A. Karspeck, 2013: Triggering of El Niño onset through trade wind-induced charging of the equatorial Pacific. Geophys. Res. Lett., 40, 1212–1216, https://doi.org/10.1002/grl.50200.
Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97, 163–172, https://doi.org/10.1175/1520-0493(1969)097<0163:ATFTEP>2.3.CO;2.
Bond, N., J. Overland, M. Spillane, and P. Stabeno, 2003: Recent shifts in the state of the North Pacific. Geophys. Res. Lett., 30, 2183, https://doi.org/10.1029/2003GL018597.
Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate, 5, 541–560, https://doi.org/10.1175/1520-0442(1992)005<0541:AIOMFF>2.0.CO;2.
Bretherton, C. S., M. Widmann, V. P. Dymnikov, J. M. Wallace, and I. Bladé, 1999: The effective number of spatial degrees of freedom of a time-varying field. J. Climate, 12, 1990–2009, https://doi.org/10.1175/1520-0442(1999)012<1990:TENOSD>2.0.CO;2.
Carton, J. A., and B. S. Giese, 2008: A reanalysis of ocean climate using Simple Ocean Data Assimilation (SODA). Mon. Wea. Rev., 136, 2999–3017, https://doi.org/10.1175/2007MWR1978.1.
Ceballos, L. I., E. Di Lorenzo, C. D. Hoyos, N. Schneider, and B. Taguchi, 2009: North Pacific Gyre Oscillation synchronizes climate fluctuations in the eastern and western boundary systems. J. Climate, 22, 5163–5174, https://doi.org/10.1175/2009JCLI2848.1.
Chakravorty, S., R. C. Perez, B. T. Anderson, B. S. Giese, S. M. Larson, and V. Pivotti, 2020: Testing the trade wind charging mechanism and its influence on ENSO variability. J. Climate, 33, 7391–7411, https://doi.org/10.1175/JCLI-D-19-0727.1.
Chang, P., and Coauthors, 2007: Pacific Meridional Mode and El Niño–Southern Oscillation. Geophys. Res. Lett., 34, L16608, https://doi.org/10.1029/2007GL030302.
Chen, H.-C., Y.-H. Tseng, Z.-Z. Hu, and R. Ding, 2020: Enhancing the ENSO predictability beyond the spring barrier. Sci. Rep., 10, 984, https://doi.org/10.1038/s41598-020-57853-7.
Chen, S., B. Yu, and W. Chen, 2014: An analysis on the physical process of the influence of AO on ENSO. Climate Dyn., 42, 973–989, https://doi.org/10.1007/s00382-012-1654-z.
Chen, S., R. Wu, W. Chen, and B. Yu, 2015: Influence of the November Arctic Oscillation on the subsequent tropical Pacific sea surface temperature. Int. J. Climatol., 35, 4307–4317, https://doi.org/10.1002/joc.4288.
Chen, S., and Coauthors, 2020: Potential impact of preceding Aleutian low variation on El Niño–Southern Oscillation during the following winter. J. Climate, 33, 3061–3077, https://doi.org/10.1175/JCLI-D-19-0717.1.
Chen, S., B. Yu, R. Wu, W. Chen, and L. Song, 2021: The dominant North Pacific atmospheric circulation patterns and their relations to Pacific SSTs: Historical simulations and future projections in the IPCC AR6 models. Climate Dyn., 56, 701–725, https://doi.org/10.1007/s00382-020-05501-1.
Chiang, J. C., and D. J. Vimont, 2004: Analogous Pacific and Atlantic meridional modes of tropical atmosphere–ocean variability. J. Climate, 17, 4143–4158, https://doi.org/10.1175/JCLI4953.1.
Clarke, A. J., S. V. Gorder, and G. Colantuono, 2007: Wind stress curl and ENSO discharge/recharge in the equatorial Pacific. J. Phys. Oceanogr., 37, 1077–1091, https://doi.org/10.1175/JPO3035.1.
Deser, C., and Coauthors, 2012: ENSO and Pacific decadal variability in the Community Climate System Model version 4. J. Climate, 25, 2622–2651, https://doi.org/10.1175/JCLI-D-11-00301.1.
Di Lorenzo, E., and Coauthors, 2008: North Pacific Gyre Oscillation links ocean climate and ecosystem change. Geophys. Res. Lett., 35, L08607, https://doi.org/10.1029/2007GL032838.
Ding, R., J. Li, Y. Tseng, and C. Ruan, 2015a: Influence of the North Pacific Victoria mode on the Pacific ITCZ summer precipitation. J. Geophys. Res., 120, 964–979, https://doi.org/10.1002/2014JD022364.
Ding, R., J. Li, Y. Tseng, C. Sun, and Y. Guo, 2015b: The Victoria mode in the North Pacific linking extratropical sea level pressure variations to ENSO. J. Geophys. Res., 120, 27–45, https://doi.org/10.1002/2014JD022221.
Ding, R., J. Li, Y. Tseng, L. Li, C. Sun, and F. Xie, 2018: Influences of the North Pacific Victoria mode on the South China Sea summer monsoon. Atmosphere, 9, 229, https://doi.org/10.3390/atmos9060229.
Eyring, V., S. Bony, G. A. Meehl, C. A. Senior, B. Stevens, R. J. Stouffer, and K. E. Taylor, 2016: Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev., 9, 1937–1958, https://doi.org/10.5194/gmd-9-1937-2016.
Fu, Y., Z. Lin, and D. Guo, 2020: Improvement of the simulation of the summer East Asian westerly jet from CMIP5 to CMIP6. Atmos. Oceanic Sci. Lett., 13, 550–558, https://doi.org/10.1080/16742834.2020.1746175.
Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447–462, https://doi.org/10.1002/qj.49710644905.
Hennemuth, B., and Coauthors, 2013: Statistical methods for the analysis of simulated and observed climate data: Applied in projects and institutions dealing with climate change impact and adaptation. Climate Service Center Rep. 13, 135 pp., https://www.climate-service-center.de/products_and_publications/publications/detail/062667/index.php.en.
Jia, F., W. Cai, L. Wu, B. Gan, and N. Keenlyside, 2019: Weakening Atlantic Niño–Pacific connection under greenhouse warming. Sci. Adv., 5, eaax4111, https://doi.org/10.1126/sciadv.aax4111.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–472, https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Krishnamurti, T., and Coauthors, 1999: Improved weather and seasonal climate forecasts from multimodel superensemble. Science, 285, 1548–1550, https://doi.org/10.1126/science.285.5433.1548.
Li, X., W. Zhang, R. Ding, and L. Shi, 2020: Joint impact of North Pacific Victoria mode and South Pacific Quadrapole mode on Pacific ITCZ summer precipitation. Climate Dyn., 54, 4545–4561, https://doi.org/10.1007/s00382-020-05243-0.
Linkin, M. E., and S. Nigam, 2008: The North Pacific Oscillation–west Pacific teleconnection pattern: Mature-phase structure and winter impacts. J. Climate, 21, 1979–1997, https://doi.org/10.1175/2007JCLI2048.1.
Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78, 1069–1080, https://doi.org/10.1175/1520-0477(1997)078<1069:APICOW>2.0.CO;2.
Peng, P., A. Kumar, H. van den Dool, and A. G. Barnston, 2002: An analysis of multimodel ensemble predictions for seasonal climate anomalies. J. Geophys. Res., 107, 4710, https://doi.org/10.1029/2002JD002712.
Pu, X., Q. Chen, Q. Zhong, R. Ding, and T. Liu, 2019: Influence of the North Pacific Victoria mode on western North Pacific tropical cyclone genesis. Climate Dyn., 52, 245–256, https://doi.org/10.1007/s00382-018-4129-z.
Rayner, N., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, https://doi.org/10.1029/2002JD002670.
Rogers, J. C., 1981: The North Pacific Oscillation. J. Climatol., 1, 39–57, https://doi.org/10.1002/joc.3370010106.
Shi, L., and R. Ding, 2020: Contributions of tropical–extratropical oceans to the prediction skill of ENSO after 2000. Atmos. Oceanic Sci. Lett., 13, 338–345, https://doi.org/10.1080/16742834.2020.1755600.
Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106, 7183–7192, https://doi.org/10.1029/2000JD900719.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, https://doi.org/10.1175/BAMS-D-11-00094.1.
Thomas, E. E., and D. J. Vimont, 2016: Modeling the mechanisms of linear and nonlinear ENSO responses to the Pacific meridional mode. J. Climate, 29, 8745–8761, https://doi.org/10.1175/JCLI-D-16-0090.1.
Vimont, D. J., D. S. Battisti, and A. C. Hirst, 2001: Footprinting: A seasonal connection between the tropics and mid-latitudes. Geophys. Res. Lett., 28, 3923–3926, https://doi.org/10.1029/2001GL013435.
Vimont, D. J., D. S. Battisti, and A. C. Hirst, 2003a: The seasonal footprinting mechanism in the CSIRO general circulation models. J. Climate, 16, 2653–2667, https://doi.org/10.1175/1520-0442(2003)016<2653:TSFMIT>2.0.CO;2.
Vimont, D. J., J. M. Wallace, and D. S. Battisti, 2003b: The seasonal footprinting mechanism in the Pacific: Implications for ENSO. J. Climate, 16, 2668–2675, https://doi.org/10.1175/1520-0442(2003)016<2668:TSFMIT>2.0.CO;2.
Vimont, D. J., M. A. Alexander, and M. Newman, 2014: Optimal growth of central and east Pacific ENSO events. Geophys. Res. Lett., 41, 4027–4034, https://doi.org/10.1002/2014GL059997.
Walker, G. T., 1925: Correlation in seasonal variations of weather—A further study of world weather. Mon. Wea. Rev., 53, 252–254, https://doi.org/10.1175/1520-0493(1925)53<252:CISVOW>2.0.CO;2.
Wang, S.-Y., M. L’Heureux, and H. H. Chia, 2012: ENSO prediction one year in advance using western North Pacific sea surface temperatures. Geophys. Res. Lett., 39, L05702, https://doi.org/10.1029/2012GL050909.
Wang, S.-Y., M. L’Heureux, and J.-H. Yoon, 2013: Are greenhouse gases changing ENSO precursors in the western North Pacific? J. Climate, 26, 6309–6322, https://doi.org/10.1175/JCLI-D-12-00360.1.
Wang, X., M. Chen, C. Wang, S.-W. Yeh, and W. Tan, 2019: Evaluation of performance of CMIP5 models in simulating the North Pacific Oscillation and El Niño Modoki. Climate Dyn., 52, 1383–1394, https://doi.org/10.1007/s00382-018-4196-1.
Xie, S.-P., and S. G. H. Philander, 1994: A coupled ocean–atmosphere model of relevance to the ITCZ in the eastern Pacific. Tellus, 46A, 340–350, https://doi.org/10.3402/tellusa.v46i4.15484.
You, Y., and J. C. Furtado, 2018: The South Pacific meridional mode and its role in tropical Pacific climate variability. J. Climate, 31, 10 141–10 163, https://doi.org/10.1175/JCLI-D-17-0860.1.
Yu, J.-Y., and S. T. Kim, 2011: Relationships between extratropical sea level pressure variations and the central Pacific and eastern Pacific types of ENSO. J. Climate, 24, 708–720, https://doi.org/10.1175/2010JCLI3688.1.
Zhang, L., P. Chang, and L. Ji, 2009: Linking the Pacific meridional mode to ENSO: Coupled model analysis. J. Climate, 22, 3488–3505, https://doi.org/10.1175/2008JCLI2473.1.
Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal variability: 1900–93. J. Climate, 10, 1004–1020, https://doi.org/10.1175/1520-0442(1997)010<1004:ELIV>2.0.CO;2.
