This study investigates the features of atmospheric circulation and moisture transport associated with two modes of decadal variability in the North Pacific: the Pacific decadal oscillation (PDO) and the North Pacific Gyre Oscillation (NPGO), with emphasis on the multiscale water vapor transport and atmospheric river (AR) over the North Pacific region. During the positive phase of PDO, the geopotential height anomaly at 500-hPa exhibits a Pacific–North American-like pattern. During the positive phase of NPGO, the geopotential height anomaly at 500 hPa features a dipole pattern with a negative anomaly north of 40°N and a positive anomaly south of 40°N over the North Pacific. Associated with the positive PDO phase, the ocean-to-land moisture transport is enhanced between 25° and 35°N and reduced over the northeastern Pacific (25°–62°N, 180°–110°W) for the time-mean integrated vapor transport (IVT). The synoptic poleward transport is suppressed north of 40°N and enhanced south of 40°N. In the positive NPGO phase, the zonal moisture transport is intensified south of 20°N and between 40° and 50°N for the time-mean IVT and weakened over the west coast of North America for the low-frequency (10–100 days) IVT. The synoptic poleward transport is suppressed south of 30°N. The eastern part of the North Pacific AR belt moves southward during positive PDO as the entire North Pacific AR belt shifts slightly northward during positive NPGO. An investigation of AR anomalies during a period over which the PDO and NPGO coexist demonstrates that the AR frequency over the North American western coastal regions is significantly influenced by the conjunction of the PDO and NPGO modes.
The natural variability of sea surface temperature (SST) over the North Pacific basin operates over multiple time scales (Chao et al. 2000; Mantua and Hare 2002; Deser et al. 2004; Messié and Chavez 2011; Liu 2012; Hong et al. 2014). The two dominant modes of decadal variability are the Pacific decadal oscillation (PDO; Mantua et al. 1997) and the North Pacific Gyre Oscillation (NPGO; Di Lorenzo et al. 2008). The PDO and NPGO signals can affect the weather and climate over the North Pacific, East Asia, and North America on decadal time scales. In addition, they can influence the local weather and climate indirectly through modulating interannual variability such as El Niño–Southern Oscillation (ENSO; Alexander et al. 2002; Mantua and Hare 2002; Zhu and Yang 2003a,b; Yang et al. 2004; Chan and Zhou 2005; Zhou et al. 2006; Linkin and Nigam 2008; Ceballos et al. 2009; Di Lorenzo et al. 2010; Furtado et al. 2012; Mills and Walsh 2013; Zhu et al. 2014). Moreover, natural decadal variability is of critical importance to climate prediction for the next few decades (Meehl et al. 2009). Therefore, the PDO and NPGO modes have been the focus of extensive research.
The PDO pattern is defined as the first mode of empirical orthogonal function (EOF) analysis of SST anomalies over the North Pacific region. The positive phase of the PDO is characterized by negative SST anomalies in the central North Pacific region and positive SST anomalies along Alaska, the west coast of North America, and the southern to central and eastern areas of the tropical Pacific (Mantua et al. 1997; Bond et al. 2003). Associated with the positive phase of the PDO, anomalously dry periods are observed in northeastern Asia, the interior of Alaska, an elongated belt from the west coast region of the United States near 45°N to the Great Lakes, the Ohio Valley, and much of central North America, while anomalously wet periods occur in the coastal Gulf of Alaska, the southwestern United States, and Mexico (Mantua and Hare 2002). An interdecadal change of East Asian monsoon rainfall associated with the PDO was demonstrated in the literature (Zhu and Yang 2003b; Mao et al. 2011).
The NPGO pattern represents the second EOF mode of sea surface height (SSH) anomalies and SST anomalies over the northeastern Pacific (25°–62°N, 180°–110°W; Di Lorenzo et al. 2008). The positive phase of NPGO is characterized by a north negative–south positive dipolar structure of SST or SSH anomalies straddling 40°N latitude over the northeastern Pacific region. The large-scale atmospheric circulation associated with the NPGO is the North Pacific Oscillation (NPO), which is the second dominant mode of sea level pressure variability over the North Pacific region (Di Lorenzo et al. 2008; Ceballos et al. 2009). The NPO is associated with continent-wide warming and increased precipitation along the Alaskan–Canadian coast and over the south-central Great Plains in the winter (Linkin and Nigam 2008).
The PDO and NPGO influence the precipitation and temperature over the pan-Pacific region through atmospheric circulation and hydrological processes. Among these processes, water vapor transport is a key one, combining atmospheric circulation with moisture conditions in the air. The water vapor that plays a role in generating precipitation is furnished by the low-level convergence of moisture flux (Chen et al. 2004). The convergence of water vapor transport indicates an increase in local atmospheric moisture and is favorable to precipitation. However, our understanding of the water vapor transport anomalies associated with the PDO and NPGO is incomplete. In previous studies, the North Pacific region was found to be a key area of ocean-to-land and poleward water vapor transport (Sohn et al. 2004; Newman et al. 2012; Zhang et al. 2013; Liu and Barnes 2015). Water vapor transport is conducted by time-mean atmospheric circulation, low-frequency circulation (10–100 days), and synoptic circulation (<10 days). The time-mean circulation primarily moves moisture zonally within ocean basins while the low-frequency and synoptic transport processes operate both from ocean to land and toward the poles (Newman et al. 2012). From a synoptic perspective, storm tracks are regions over which midlatitude cyclones are prevalent (Trenberth and Hurrell 1994; Straus and Shukla 1997; Chang et al. 2002). The North Pacific storm track exhibits pronounced interdecadal variability, which is partially related to the PDO mode (Chang and Fu 2002). The NPGO may also induce a distinct storm-track anomaly. The storm track as well as large-scale circulation anomalies associated with the PDO and NPGO may finally result in moisture transport anomalies on different time scales. This makes up the core subject of this study.
Newell et al. (1992) revealed from twice-daily data that water vapor transport is concentrated into filaments in the midlatitudes—this phenomenon was later termed as atmospheric river (AR) by Zhu and Newell (1994, 1998). The AR plays an important role in shaping the precipitation over the pan-Pacific region because of the large magnitude of moisture flux carried by the AR (Ralph et al. 2006; Neiman et al. 2008a,b; Smith et al. 2010; Jiang and Deng 2011; Kim et al. 2013; Dettinger 2013; Rutz et al. 2014; Warner et al. 2015). The means by which the PDO and NPGO influence ARs is not clear. This issue is also addressed in this study.
The study is organized as follows: section 2 describes the data and methods used, section 3 shows the large-scale atmospheric circulation and storm-track anomalies associated with PDO and NPGO, section 4 analyzes multiscale moisture transport anomalies associated with the two decadal modes, section 5 investigates the features of AR anomalies, and section 6 contains a discussion and conclusions.
2. Data and methods
In this study, the monthly mean SST used is the Hadley Centre Sea Ice and Sea Surface Temperature dataset on a 1° latitude–longitude grid (HadISST1; Rayner et al. 2003). The atmospheric variables are from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalyses (Kalnay et al. 1996) with a resolution of 2.5° × 2.5°, including daily and monthly data for wind fields (zonal u, meridional υ), geopotential height z, and specific humidity q. This study was conducted for the winter season from December 1950 through February 2013. The winter mean of 1950 is defined as the average of December 1950, January 1951, and February 1951; the winter means are defined in the same way for the subsequent years. The filter used for this study is the Butterworth filter (Murakami 1979). All the filtered data are obtained by using the whole-year monthly (or daily) data from which the winter-mean values are calculated.
The monthly SST data are filtered using an 8–50-yr bandpass filter to obtain the decadal signal. The EOF and regression analyses are then performed to reveal the decadal fluctuations of SST anomalies and the associated anomalies of atmospheric elements. The 8-yr, high-pass window ensures the removal of interannual variations such as ENSO. The 50-yr, low-pass window can retain decadal time scales of approximately 10–20 years and remove multidecadal time scales (Minobe 1997; Chao et al. 2000; Ware and Thomson 2000; Mantua and Hare 2002; Zhu and Yang 2003a).
The multiple scales used in this study are the time-mean, the low-frequency (10–100 days), and the synoptic (2–8 days) time scales. The low-frequency signal is obtained by using a 10–100-day Butterworth filter. The North Pacific synoptic activity (storm track) is isolated from the daily data using a 2.5–8 day filter. The strength of the storm track is represented by the variance of the 2.5–8 day filtered meridional wind at 250 hPa. The spatial–temporal variations of storm track at other levels are similar to those at 250 hPa (results not shown). The storm track is closely related to the Eady growth rate maximum (Lindzen and Farrell 1980) defined as . In this equation, is the Coriolis parameter, is the Brunt–Väisälä frequency, is the time-mean horizontal wind, and z is the vertical height. The pressure levels 850 and 700 hPa are used for computing .
The integrated vapor transport (IVT) is calculated by using the formula . In this equation, g is the acceleration due to gravity, denotes surface air pressure (1000 hPa), is set to 300 hPa, p is pressure level, q is the specific humidity, and is horizontal wind. Zhu and Newell (1998) provided a criterion for judging AR days and non-AR days: , where Q is the integrated vapor transport, which is the same as IVT. The variable denotes the zonal-mean value of the magnitude of Q in all directions along a given latitude, and denotes the magnitude of the maximum flux (Q) along a given latitude. A grid point is declared as an AR grid when the magnitude of IVT at this grid satisfies the above formula. When the grid is identified as an AR grid, the corresponding date is grouped as an AR day or a non-AR day. The AR frequency (the ratio of the number of AR days to the number of days in a whole winter) is calculated for every winter during 1950–2012.
The application of the bandpass filter may reduce the effective number of degrees of freedom. Therefore, a Monte Carlo method is adopted herein to test the statistical significance of the regression and composite results (Hegyi and Deng 2011).
3. Large-scale circulation and storm-track anomalies associated with the PDO and NPGO
An EOF analysis is first performed on the SST anomalies over the region with the coordinates 20°–60°N and 120°E–120°W for the winters of 1950–2012. The variances of the first and the second EOF modes (EOF1 and EOF2) are 35.73% and 17.95%, respectively. The normalized first two principal components are denoted PC1 and PC2, respectively. Figure 1 displays the regression fields of the North Pacific SST anomalies on PC1 and PC2. Figure 1a shows a typical pattern of PDO positive phase. The correlation coefficient between the PC1 and the filtered PDO index (http://jisao.washington.edu/pdo/PDO.latest) is 0.87, indicating that the anomalous pattern in Fig. 1a is strongly related to the PDO. Figure 1b is similar to the NPGO positive phase. The correlation coefficient between the PC2 and the filtered NPGO index (http://www.o3d.org/npgo/npgo.php, using satellite SSH anomalies) is 0.54. The moderate correlation between the PC2 and the NPGO index is due to using the SST data rather than the SSH data. Previous studies have demonstrated that the second EOF mode of SST anomalies can also reflect the NPGO (Furtado et al. 2011; Ding et al. 2013; Chhak et al. 2009). Therefore, the regression field in Fig. 1b can represent the NPGO mode. In this study, PC1 and PC2 in Figs. 1c and 1d are used as the time series of the PDO and NPGO modes, respectively.
The PDO mode was more robust prior to the 1990s (Fig. 1c) with an obvious phase shift from negative to positive in 1976/77 (Graham 1994; Deser et al. 2004; Schneider and Cornuelle 2005; Jo et al. 2013). The value of PC1 has weakened since the 1990s and has been in a negative phase during the past 10 years. Figure 1d shows that the NPGO has been significant during the past 30 years. Previous studies have argued that the NPGO pattern explains more wintertime North Pacific SST anomaly variance than the PDO pattern after 1988 (Bond et al. 2003; Di Lorenzo et al. 2008; Yeh et al. 2011). The NPGO experienced a shift from a positive to a negative phase around 1988/89 and a shift from a negative to a positive phase at about 1996/97. During the past 10 years, the NPGO has been in a negative phase.
The anomalies in large-scale atmospheric circulation associated with the PDO and NPGO are illustrated in Fig. 2. The anomalous field of geopotential height at 500 hPa related to the positive PDO exhibits a pattern similar to the Pacific–North American teleconnection pattern. In addition, the positive center over the Asian region and the negative center over the North Pacific basin indicate a deepening of the East Asian trough (Wang and Chen 2014). The positive center over North America suggests strengthening of the North American Ridge. The zonal wind anomalies at 250 hPa associated with the positive PDO exhibit a wave train pattern, stretching from the tropics to the mid- to high latitudes over the North Pacific region. It is noted that the easterly flow over the tropical Pacific is enhanced, the subtropical westerly jet along 30°N is intensified and extends eastward, and the westerly flow over the Aleutian chain region is weakened. Over the Eurasian region, the westerly flow is greater than normal along 60°N where the East Asia polar front jet is dominant (Ren et al. 2010) but is weakened over the region of Mongolia and the Tibetan Plateau.
The signal associated with the NPGO positive phase is mainly located over oceanic regions. The geopotential height anomaly at 500 hPa displays a north negative–south positive dipolar structure bordering about 40°N, with the southern lobe centered at about 30°N near the date line and the northern lobe centered over the Aleutian chain and the Alaskan region (Fig. 2b). Thus, the North American Ridge is substantially weakened during the positive phase of the NPGO. Over the North Pacific region, the westerly wind at 250 hPa is weakened along 20°N and strengthened along 45°N, indicating a poleward shift of the subtropical westerly jet. Meanwhile, a positive center is present over the tropical Pacific east of the date line.
Figure 3 shows the regression fields of and storm-track anomalies. During the positive PDO phase, the enhanced baroclinicity almost overlaps with the strengthened upper-level zonal wind around 30°N in the central North Pacific region (Fig. 3a). In Fig. 3c, the storm track is invigorated in the southern and southeastern parts of the regions with increased baroclinicity and zonal wind. To the north of 40°N over the North Pacific, negative anomalous centers of baroclinicity and storm track are present (Figs. 3a,c). Simultaneously, there is a positive storm-track center over the region 45°–65°N, 60°–90°E, where the Siberian storm track prevails (Hoskins and Hodges 2002). The strengthened Siberian storm track is attributed to enhancement of the East Asia polar front jet and of baroclinicity over the central Eurasian region during positive PDO. During the positive NPGO phase, the baroclinicity is intensified in its climatological position and in a region northeastward of its climatological position. Meanwhile, an area with negative anomaly of baroclinicity is seen to the south of 25°N. This pattern of a poleward shift of the baroclinicity (Fig. 3b) is due to the poleward shift of the subtropical westerly jet (Fig. 2d). Therefore, the North Pacific winter storm track also shows a small poleward movement during the NPGO positive phase (Fig. 3d).
These results of a coupled spatial pattern between wind field (baroclinicity) and storm-track anomalies are not unexpected, because the storm track and the mean flow have a mutually dependent relationship in the midlatitudes (Cai and Mak 1990; Chang et al. 2002). However, the magnitudes and maximum centers of anomaly for baroclinicity and storm track for the PDO and NPGO possess slightly different features. For both the PDO and NPGO modes, the baroclinicity anomaly is 0.01–0.02 day−1, while the magnitude of the storm-track anomaly associated with the PDO mode (~8 m2 s−2) is larger than that of the NPGO mode (~4 m2 s−2). In addition, the maximum anomalous center of baroclinicity emerges in the central and eastern North Pacific for the PDO mode and in the central and western North Pacific for the NPGO mode. In contrast, for the storm-track anomalies, the maximum centers occur over the eastern North Pacific for both the PDO and NPGO modes. A plausible explanation for this nonlinear behavior may be nonlinear interactions among atmospheric external forcing (e.g., SST anomalies over the Pacific), the storm track, and the midlatitude stationary wave, or the inherently nonlinear processes related to the midwinter suppression of the North Pacific storm track (Nakamura 1992; Nakamura and Sampe 2002; Chang et al. 2002; Chang and Fu 2002; Penny et al. 2010; Lee et al. 2013). Further discussion of this issue is beyond the scope of this study.
4. Multiscale moisture transport anomalies associated with the PDO and NPGO
Figure 4 contains a plot of the wintertime climatological fields of the time-mean IVT , the low-frequency IVT , and the synoptic-scale IVT and their corresponding divergence fields over the North Pacific. The spatial patterns of the three terms in Fig. 4 are similar to the results of previous studies (Sohn et al. 2004; Newman et al. 2012). Because of the dominant westerly wind, over the extratropics is mostly zonal and only slightly poleward over the eastern margins of the North Pacific basin. The divergence field of shows broad areas of moisture sources within the subtropics and western North Pacific and sinks over the northeastern Pacific–North American continent, the warm pool region, and the eastern tropical Pacific along 10°N.
The winter-mean is controlled by the low-frequency moisture and wind field signals in the mid- to lower levels. Strong low-frequency variability in lower-level moisture is observed to the south of the Kuroshio Extension over the North Pacific (Grodsky et al. 2009). The low-frequency fluctuation of low-level atmospheric circulation over the North Pacific forms an anomalous cyclonic pattern at the basin scale (Hsu 1996; Anderson 2007; Newman et al. 2012). Therefore, the low-frequency moisture transport builds up moisture sources over the region south of 30°N and the eastern parts of the North Pacific basin (Fig. 4b). The low-frequency moisture fluxes from these source regions travel northwestward to eastern China, Japan, and Alaska and northeastward to the North American continent. The northeastward transport from the northeastern Pacific to North America is stronger than the poleward and northwestward transport over the western and central North Pacific.
The pattern of climatological (Fig. 4c) is to some extent similar to that of over the western and central North Pacific. The climatological shows greater organization than does the . The ocean-to-land transport of the climatological along the west coast of North America is much weaker than that of the . The climatological diverges over the region south of 35°N and converges north of 35°N.
Figure 5 shows the regression fields of multiscale IVT and the corresponding divergence anomalies. The regression field of anomaly associated with the positive PDO shows anomalous cyclonic transport between 15° and 60°N over the North Pacific region. The amplitude is particularly large over the northeastern Pacific and along the subtropical jet stream region (Fig. 5a). The anomalous cyclonic pattern of over the North Pacific is closely linked with the intensified Aleutian low (Miller et al. 1994; Hare and Mantua 2000) and the jet stream wind during the positive PDO. Anomalous moisture sources are located over the area between the South China Sea and the Philippine Sea and over the central North Pacific centered at 15°N and southwestern North America. Anomalous sink regions are observed over the western North Pacific to the east of the Philippines and over the northeastern Pacific. These results indicate that the anomaly associated with the positive PDO transports an increased amount of moisture from the central North Pacific to the northeastern Pacific and the west coast of North America as far as Alaska. Thus, the poleward transport near the Gulf of Alaska and the zonal ocean-to-land transport between 25° and 35°N are enhanced during positive PDO episodes.
During the positive NPGO, the anomaly shows anomalous anticyclonic transport approximately between 5° and 50°N (Fig. 5b). This transport is associated with the strengthened westerly wind along 45°N at 250 hPa and the anomalous easterly wind in the lower latitudes (Fig. 2). The anomaly associated with the positive NPGO forms anomalous moisture source regions over the eastern part of the North Pacific and the central tropical Pacific and an anomalous sink over the area between the northwestern Pacific and the central basin region. The zonal moisture transport south of 20°N is dramatically enhanced, and the zonal ocean-to-land transport between 40° and 50°N is also increased.
The poleward over the North Pacific is markedly reduced for both the PDO and NPGO positive phases, but in different places. The poleward transport is suppressed north of 30°N around the date line during the positive PDO (Fig. 5c) and mainly suppressed over the northeastern Pacific during the positive NPGO (Fig. 5d). The ocean-to-land transport is reduced over the northwestern coast of North America and enhanced over a small region near the west coast between 30° and 40°N during the positive PDO phase (Fig. 5c). It is reduced along the west coast of North America between 20° and 45°N in association with the positive NPGO (Fig. 5d).
The regressed anomaly displays anomalous equatorward transport north of about 40°N and poleward transport south of about 40°N over the central North Pacific basin associated with the positive PDO (Fig. 5e). Thus, the poleward in mid- to high latitudes is lower than normal during the positive PDO. This pattern is in good agreement with the regressed pattern of storm-track anomaly with the invigorated storm track south of 40°N and the suppressed storm track north of 40°N illustrated in Fig. 3c. Correspondingly, anomalous moisture convergence into the central basin takes place during the positive PDO. During positive NPGO episodes, the anomaly shows equatorward transport in an elongated area south of 30°N (Fig. 5f), coincident with the suppressed local storm-track activity shown in Fig. 3d. The anomalous moisture convergence center is located south of approximately 25°N, where the reduced storm-track activity is dominant. The anomalous divergence center is located north of about 25°N, where the enhanced storm-track activity is pronounced. In the mid- to high latitudes, the influence of the NPGO mode on poleward is negligible.
5. Atmospheric river anomalies associated with the PDO and NPGO
a. Regression fields
The winter-mean AR (Fig. 6) occurs mostly in two belt-shaped regions: one is over the North Pacific along 25°–35°N and curving to the northeastern Pacific (hereinafter termed the North Pacific AR belt), and the other stretches from the warm pool to the eastern tropical Pacific (hereinafter termed the warm pool–tropical Pacific AR belt).
The AR frequency is enhanced over 160°E–170°W in the North Pacific AR belt during the positive PDO phase (Fig. 6a); simultaneously, the eastern part of the North Pacific AR belt moves equatorward. The AR frequency is greater over the Gulf of Alaska and its coastal region. The AR frequency in the warm pool–tropical Pacific AR belt increases in a narrow region between 10° and 15°N near the date line. At the same time, the warm pool–tropical Pacific AR belt extends farther eastward and southeastward over the eastern tropical Pacific.
Figure 6b shows that during the positive NPGO, the AR frequency is significantly enhanced over the western part of the North Pacific AR belt. The whole North Pacific AR belt is displaced slightly poleward. The AR frequency in the entire warm pool–tropical Pacific AR belt is increased, particularly over the western Pacific warm pool region.
The regression fields of IVT and the divergence anomalies for AR day and non-AR day groups are illustrated in Fig. 7. The anomalous patterns of the IVT and their divergent fields for the group of AR days associated with the positive PDO (Fig. 7a) and NPGO modes (Fig. 7b) closely resemble those of the individual patterns of anomalies (Figs. 5a,b). The anomalous patterns for the group of non-AR days (Figs. 7c,d) also resemble Figs. 5a and 5b to some extent. The AR definition does not distinguish between transient and steady moisture transport and usually includes transport by mean, low-frequency, and synoptic components (Newman et al. 2012). Therefore, the regression results in Fig. 7 suggest that the IVT and divergence anomalies for the AR days and non-AR days associated with the positive PDO and NPGO mostly reflect the corresponding anomalies of large-scale atmospheric circulation.
The magnitude of the IVT anomaly of the AR day group is stronger than that of the anomaly and is markedly stronger than that of the non-AR day group. For example, the value of the anomalous IVT flux, averaged over the anomalous eastward moisture conveyor belt along 30°N for the AR day group (Fig. 7a), is approximately 3 times larger than that of the non-AR day group (Fig. 7c) and approximately twice that of the (Fig. 5a). Similarly, the value of the anomalous IVT flux averaged over the anomalous westward moisture conveyor belt along 20°N for the AR day group (Fig. 7b) is approximately 4.5 times larger than that of the non-AR day group (Fig. 7d) and approximately 3.5 times that of (Fig. 5b). Over the North Pacific AR belt region, the total number of AR days is smaller than the number of non-AR days. The weighted mean of the IVT anomalies on AR days and non-AR days contributes to the moderate magnitude of the anomaly over the North Pacific region.
b. Joint effect of PDO and NPGO on AR anomalies
As previously stated, the PDO and NPGO modes are the two major decadal signals over the North Pacific region. In some years, these modes coexist and influence the AR anomaly. To identify those specific periods, the values of PC1 plus PC2 and PC1 minus PC2 are calculated (figure not shown). The years with high positive values of PC1 plus PC2 reveal that the PDO and NPGO are both in a positive phase. The years with a high positive value of PC1 minus PC2 indicate the periods over which the relatively strong positive PDO and negative NPGO phases coexist. The opposite conditions are used to identify the years that the PDO and NPGO signals are both in a negative phase as well as the years that relatively strong negative PDO and positive NPGO phases coexist.
Figure 8 shows plots of the composite AR frequency anomalies during specific periods selected by using the above-mentioned method. Figure 8a illustrates the results for the winters of 1983–87 when the PDO and NPGO signals are both positive. The AR frequency in Fig. 8a is enhanced between 150°E and 160°W in the North Pacific AR belt and decreased over the North Pacific basin between 40° and 60°N. This is influenced by the AR anomalies having the same sign over these areas during the positive PDO and positive NPGO phases. In Fig. 8a, the eastern part of the North Pacific AR belt maintains its climatological location, possibly because the anomalous southward shift of the eastern part of the North Pacific AR belt during positive PDO phases is offset by the anomalous poleward movement during positive NPGO phases. More AR occurs over the regions between the Gulf of Alaska and northwestern North America, mostly due to the positive PDO and partly due to the positive NPGO phases. Increased AR over the warm pool region is mostly the result of positive NPGO.
Figure 8b shows the results for the winters of 2008–12 when the PDO and NPGO signals are both negative. Figure 8b is a mirror image of Fig. 8a to some extent. For example, the AR frequency in Fig. 8b is decreased between 150°E and 160°W in the North Pacific AR belt and increased over the North Pacific basin between 40° and 60°N because the AR anomalies have the same sign over these regions during the negative PDO and negative NPGO. Less AR occurs over the warm pool region and tropical eastern Pacific. However, there are some regions where AR anomalies have the same sign in Figs. 8a and 8b; for example, more AR is present around the Bering Strait and on the southern side of the eastern part of the North Pacific AR belt, around 30°N, 180°–150°W. We speculate that this is due to nonlinear interactions between the PDO and NPGO or other non-PDO or non-NPGO related factors.
Figure 8c contains the results for the winters of 1978–82 when relatively strong positive PDO and negative NPGO phases coexist. Enhanced AR frequency over the eastern North Pacific region is visible in Fig. 8c, centered around 30°N, 140°W, indicating a southward shift of the eastern part of the North Pacific AR belt. Meanwhile, the AR frequency is noticeably lower in the region northeastward from the western North Pacific to the northeastern North Pacific as well as the region along 15°N and between 150°E and 180°. The reason for these anomalies is the combination of the positive PDO and negative NPGO phases.
Figure 8d shows the results for the winters of 1971–75 when relatively strong negative PDO and positive NPGO phases coexist. Between Figs. 8c and 8d, the AR anomalies have opposite signs over the eastern North Pacific region, the western North Pacific region, and the narrow region along 15°N. However, this mirror-image pattern is not prominent in the western part of the North Pacific AR belt and over the South Pacific region.
To illustrate the general features of the observed decadal time-scale AR anomaly during the periods over which both modes coexist, Fig. 9 contains plots of the sum of EOF1 and EOF2 anomalies of AR frequency (which were shown in Figs. 6a,b) and the difference between the two. The pattern of the sum demonstrates the general characteristics of the AR frequency anomaly when the PDO and NPGO signals are both positive. The AR frequency in the North Pacific AR region and the warm pool–tropical Pacific AR region is intensified in Fig. 9a. The former region extends northeastward to the Gulf of Alaska and northwestern North America, while the latter area extends farther eastward to the tropical coastal line. The pattern of the difference between Figs. 6a and 6b can indicate the general characteristics of the AR frequency anomaly during the period over which the PDO signal is in relatively strong positive episodes and the NPGO signal is in negative. The southward shift of the North Pacific AR belt in its eastern part is clearly visible in Fig. 9b. Thus, more AR is present along the west coast of North America. Simultaneously, the AR frequency in the western and central parts of the warm pool–tropical Pacific AR region is significantly decreased. A comparison of Fig. 9a with Fig. 8a shows that there are similarities between the two maps, such as the persistence of the North Pacific AR belt and the strong anomalies around the Gulf of Alaska and in the warm pool region. Similarly, Figs. 8c and 9b show common features of a southward shift in the eastern part of the North Pacific AR belt and a region with decreased AR along 15°N and between 150°E and 180°. These results may suggest that the AR anomalies associated with the PDO and NPGO modes are additive.
6. Discussion and conclusions
This study investigates the anomalous features of large-scale circulation, storm track, water vapor transport, and the AR over the North Pacific region during the winter associated with the PDO and NPGO. The notable features of large-scale circulation anomalies associated with the positive PDO include the deepening of the East Asian trough, strengthening of the North American Ridge, and intensification and downstream extension of the subtropical jet over the North Pacific region. Large-scale circulation anomalies associated with positive NPGO are the weakening of the North American Ridge and the poleward shift of the subtropical jet over the North Pacific region.
The features of multiscale moisture transport associated with the PDO and NPGO modes exhibit some differences. During the positive phase of PDO, the anomalous reveals a cyclonic moisture transport over the North Pacific basin between 15° and 60°N. More moisture is transported by the anomaly from the central North Pacific region to the northeastern Pacific region and the west coast of North America and to Alaska. The anomalies display equatorward transport over the central North Pacific in the area poleward of 30°N and offshore transport over the northwestern coast of North America. The anomaly associated with positive PDO displays equatorward moisture transport over mid- to high latitudes and poleward moisture transport over lower latitudes.
During the positive phase of NPGO, the anomalous pattern shows a significant anticyclonic pattern over the North Pacific basin between 5° and 50°N. The anomalies exhibit southeastward transport over the central North Pacific region and southwestward transport over the west coast of North America. The poleward transport is reduced south of 30°N.
The winter-mean AR is located in two belt-shaped regions: the North Pacific AR belt and the warm pool–tropical Pacific AR belt. During the PDO positive phase, the eastern part of the North Pacific AR belt moves equatorward anomalously and during the NPGO positive phase, the entire North Pacific AR belt moves slightly poleward. The warm pool–tropical Pacific AR belt extends farther eastward during the positive PDO and is intensified during the positive NPGO. The IVT anomalies for the AR day group and the non-AR day group associated with the positive PDO (or NPGO) are dependent on the corresponding anomalies of large-scale atmospheric circulation.
When the PDO and NPGO are both in their positive phases, the winter AR frequency is intensified in both the North Pacific AR belt region and the warm pool–tropical Pacific AR belt, and both AR regions extend to their individual downstream regions. During winters in which the PDO is positive and the NPGO is negative, more ARs occur along the west coast of North America. Meanwhile, the AR frequency decreases in the western and central parts of the warm pool–tropical Pacific AR belt region.
Future studies should examine other related factors such as integrated water vapor and evaporation as well as their relationship with the PDO and NPGO modes. Our results suggest that the AR anomaly (or perhaps other moisture variables) has a linear response to the PDO and NPGO signals to some extent, meaning that the AR anomalies associated with the PDO and NPGO are additive. In addition, it is difficult to determine whether the PDO or the NPGO solely affect AR during a specific period on the basis of the present study. However, it is clear that the AR frequency over the North American western coastal regions is significantly influenced by the combination of the PDO and NPGO modes. These results imply that the PDO and NPGO are equally important for the AR (or other moisture variables). Thus, both the PDO and NPGO must be considered when investigating the observed pan-Pacific hydroclimate variability as previously suggested by Di Lorenzo et al. (2009).
This work was jointly supported by the National Natural Science Foundation of China under Grants 41275068 and 41330420, the Special Fund for Meteorology Research in the Public Interest (Grant GYHY201106017), and the Jiangsu Collaborative Innovation Center for Climate Change. We thank the anonymous reviewers for their insightful comments that greatly improved the quality of this manuscript.