Spatiotemporal differences in the interannual variability of Baiu frontal activity in June

Authors


Abstract

The Baiu frontal activity (BFA) clearly shows spatiotemporal differences in its interannual variability. This work examines the physical mechanisms behind these differences. On interannual time scales, the Baiu front can be divided into three subregions: (1) the western Baiu (WB), (2) the central Baiu (CB), and (3) the eastern Baiu (EB). Time series analysis revealed that the dominant periods in these three subregions are long eastward periods of approximately 2 years in the WB, 4 years in the CB, and 6 years in the EB.

The biennial oscillation of the Asian monsoon controls the interannual variation in the WB through specific meridional circulation in the western North Pacific, whereas the El Niño/Southern Oscillation forces the interannual variation in the CB through the Pacific–East Asian teleconnection. The interannual variation in the EB is controlled by mid-latitude atmospheric circulations, not by effects from the Tropics. The summertime North Atlantic Oscillation (SNAO) with a 6-year period excites the stationary Rossby waves, the energies of which reach Japan through the strong upper tropospheric westerlies over Eurasia. Geopotential height anomalies then appear around Japan with an equivalent barotropic structure that modifies the precipitation in the EB. Copyright © 2009 Royal Meteorological Society

1. Introduction

The early summer climate in East Asia is characterized by the Baiu (in Japanese; also called Mei-yu in Chinese and Changma in Korean) phenomenon, which appears in the northwestern periphery of the North Pacific subtropical high (NPSH) (Figure 1). Baiu precipitation is crucial for summertime water resources in East Asia, but heavy rainfalls sometimes cause natural disasters. Therefore, many researchers have investigated the physical mechanisms of the interannual variations in Baiu precipitation. This work examines the spatiotemporal coherence in the interannual variations of Baiu frontal activity (BFA) and the physical mechanisms behind those variations.

Figure 1.

Monthly mean precipitation (dotted contours) and the standard deviation (shading) in June. The boxes indicate the three subregions of the Baiu front. The letters W, C, and E denote the respective subregions. The contour interval is 2.0 mm day−1, and the light (dark) shading indicates the regions with a standard deviation larger than 2.0 (3.0) mm day−1

The Baiu front, which appears from southeastern China to the central North Pacific, is identified by a large gradient of equivalent potential temperature and characterized by slow northward migration from May to July (Yoshino, 1965, 1966). In addition, although appearing like a large zonal band, the Baiu front has complex structures in its circulation system. As an example, Baiu precipitation in southwestern Japan is usually greater than that in northeastern Japan due to differences in the strength of the baroclinicity and in the atmospheric vertical instability (Ninomiya and Mizuno, 1987). As another example, to the east of 130°E, a large meridional temperature gradient is formed between a maritime tropical air mass of the NPSH and a counterpart maritime polar air mass in the Okhotsk High, leading to large baroclinicity and resultant westerly thermal winds in the upper troposphere; in contrast, to the west of 130°E, the meridional gradient of specific humidity, which is formed between a dry air mass over the Asian continent and the wet flow of the East Asian summer monsoon, is greater than that of temperature. Among such circulation systems, the precipitation in the Baiu front is maintained by the water vapour carried through the western edge of the NPSH (Ninomiya and Akiyama, 1992; Kawamura and Murakami, 1998; Ding and Chan, 2005). Thus, the Baiu front, appearing in East Asia in early summer, is regarded as a complex circulation system that is modulated by the NPSH, the Okhotsk High, the monsoon trough, and the heat low over the Asian continent.

Many studies have examined the interannual variations of BFA in relation to teleconnections from the Tropics. For example, Nitta (1987) and Tsuyuki and Kurihara (1989) suggested that when convective activities are intensified in early summer around the eastern Philippine Sea, the anomalous high appears near Japan through the so-called Pacific–Japan (PJ) pattern, which brings effects of the Tropics to the mid-latitudes. Nitta and Hu (1996) and Lu and Dong (2001) also noted that the reverse phase of the PJ pattern increases Baiu precipitation with the anomalous high around the eastern Philippine Sea. Kosaka and Nakamura (2006) physically demonstrated that the PJ pattern is a dynamical mode with a self-sustaining mechanism in the zonally asymmetric baroclinic flow associated with the Asian summer monsoon. The PJ pattern is a major factor controlling Baiu precipitation in the Asian summer monsoon.

Nitta (1987) also pointed out that the PJ pattern is influenced by the El Niño/Southern Oscillation (ENSO) on interannual time scales. In addition, Wang et al. (2000) showed that wintertime positive sea surface temperature (SST) anomalies in the equatorial central Pacific associated with the ENSO excited the anomalous high in and around the eastern Philippine Sea through the so-called Pacific–East Asian teleconnection, and that the persistence of the anomalous high changed the course of water vapour to East Asia and modified the Baiu precipitation anomalies in the following early summer. Tanaka (1997) and Chang et al. (2000a, 2000b) confirmed that the BFA in early summer has a positive lag correlation with the interannual variation of SST in the tropical eastern Pacific in winter.

The BFA also has a biennial tendency related to the biennial oscillation of the Asian monsoon, which is linked through the meridional circulation in the western North Pacific (Tomita et al., 2004). In general, the tropospheric biennial oscillation (TBO) is dominant in the variations of SST and zonal winds in the equatorial regions from the Indian Ocean to the central Pacific (Shen and Lau, 1995; Chang and Li, 2000). The TBO forms the anomalous anticyclonic (cyclonic) circulation in the eastern part of the Philippine Sea that is associated with the anomalous Hadley circulation excited by positive (negative) SST anomalies in the tropical western Pacific; like the PJ pattern, the effects of the anomalous anticyclonic (cyclonic) circulation modify the course and amount of water vapour available for Baiu precipitation.

Not only effects from the Tropics, but also mid-latitude variations can modulate the BFA. Enomoto et al. (2003) pointed out that the stationary Rossby waves in the Asian jet along 40°N established the anomalous Bonin High to the south of Japan, which modified the flows of water vapour and controlled the northward shift of the Baiu front. Krishnan and Sugi (2001) further confirmed the existence of a wave train in June and July over Eurasia. Hsu and Lin (2007) reported that stationary Rossby waves controlling the BFA appeared from the eastern Tibetan Plateau to Japan. Propagations of Rossby waves in the Asian jet modify the pressure anomalies in and around East Asia, change the flows of water vapour, and control the interannual variations of Baiu precipitation.

The Okhotsk High, which appears to the north of the Baiu front, is another controlling factor of the interannual variability of BFA, forcing an anomalous condition of cooler and wetter northeasterlies near Japan and strengthened baroclinicity around the Baiu front. When this condition is strong, cool and wet, northeasterlies called ‘Yamase’ appear in northeastern Japan. The Okhotsk High in the lower troposphere tends to be formed under a blocking high in the upper troposphere around the Sea of Okhotsk. Nakamura and Fukamachi (2004) discussed the physical relationship between the upper blocking high and the counterpart lower anticyclonic circulation. They found that the upper blocking high was developed through barotropic forcing associated with transient eddies in May, whereas in July, the development of the blocking high was caused by the propagation of Rossby waves from the west. Ogi et al. (2003) and Tachibana et al. (2004) have also reported that the upper blocking high is forced by stationary Rossby waves in the north Eurasian jet along 70°N. As such, the propagating Rossby waves in the north Eurasian jet influence the lower pressure field around the Sea of Okhotsk, leading to changes in flow that affect the interannual variations of the Baiu front.

Although many researchers have revealed the characteristics of the interannual variations of BFA and their relationships to effects from other regions, many questions still remain. For example, do interannual variations in the Tropics, such as the ENSO or TBO, uniformly modify the Baiu front? Does the North Atlantic Oscillation (NAO), which is representative of mid-latitude atmospheric circulations in the Northern Hemisphere, have an impact on the interannual variations of BFA? If the NAO affects the BFA, do the effects appear uniformly across the entire region of the Baiu front? To approach these questions, this work will first examine whether the interannual variability of BFA appears uniformly in the Baiu region. If the appearance is spatially varied, we will then define the subregions of the Baiu front and investigate the physical mechanisms behind the interannual variation in each subregion. In particular, we will discuss how the effects of the interannual variations in mid-latitude atmospheric circulations are reflected in the BFA.

The rest of this paper is organized as follows. Section 2 describes the data sets and methodologies used in this work. Section 3 confirms that the Baiu region can be divided into subregions. For each determined subregion, we then describe the interannual variability, spatial pattern, and associated atmospheric circulations in detail. Section 4 diagnoses the effects of the tropical controls and of the mid-latitude atmospheric circulations on the interannual variability in the Baiu subregions. Finally, we give a summary of this work in Section 5, along with some additional discussion and comments.

2. Data and methodology

For precipitation, we used data produced by the Global Precipitation Climatology Project (GPCP) (Adler et al., 2003). We also used data sets from the National Centre for Environment Prediction (NCEP)–Department of Energy (DOE) Atmospheric Model Intercomparison Project II (AMIP-II) reanalysis (Kalnay et al., 1996; Kanamitsu et al., 2002) for atmospheric parameters. The precipitation and atmospheric parameter data sets are both gridded globally at intervals of 2.5° in the longitudinal and latitudinal directions. Data for SST are extracted from the NCEP–DOE AMIP-II reanalysis data sets, which are gridded at 1.875° intervals of longitude, with a Gaussian distribution in the latitudinal direction. All the data sets were compiled monthly for the period of 1979–2007 (29 years).

This work focuses on the BFA in June. In western to central Japan, the Baiu season starts the beginning of June and ends in the mid- to late July. Precipitation in June is the strongest among the 3 months of May, June, and July in the area around Japan and represents BFA well. These features suggest that, for monthly data, June is the month most representative of the characteristics of the Baiu. To focus on the strong interannual variability in BFA around Japan, the Baiu front was defined as areas in which standard deviation of the precipitation rate was larger than 2 mm day−1 (Figure 1). The region meeting this criterion covers the area of 25°–40°N, 110°–170°E, with 92 grid points. Generally, the areas with large standard deviations corresponded to those with large mean precipitation. To divide the Baiu front into subregions, we employed cluster analysis based on the Ward method using square Euclidean distances (Ward, 1963). In this work, we intentionally avoided employing the popular empirical orthogonal function (EOF) analysis because the so-called EOF modes are strongly constrained by their orthogonality.

The cluster analysis used in this work is as follows. First, we assume that a cluster C is formed by two clusters, A and B. When the spatiotemporal data (precipitation in this work) are defined as xij at time i and a grid point j, the spatial mean of cluster A is estimated as

equation image(1)

where mA is the total grid number of cluster A. The sum of squared differences in cluster A (SA), which is regarded as a variable related to the variance in the cluster, is then estimated as follows:

equation image(2)

where n denotes the record length in time (29 in this work); SA in Equation (2) is definitely non-negative. When mA equals 2, SA indicates half of the squared Euclidean distance between the two grid points. SB is obtained in a similar way. The sum of these squared differences forms the new cluster C, for which the sum of squared differences (SC) is

equation image(3)

where ΔSC indicates the increment of SC for the two clusters, A and B, and is expressed as follows:

equation image(4)

Note that the sum of mA and mB becomes mC. Equation (4) indicates that, when ΔSC is small, the difference between xiA and xiB tends to be small. That is, the two clusters have similar temporal variations. In the case that ΔSC is the smallest of the increments of the sum of squared differences in other clusters, we connect the two clusters A and B to make cluster C. Such connecting of clusters is recursively conducted to obtain a final single cluster by finding the smallest increment for arbitrary clusters.

When ΔSC is large, the correlation is small between the variations in clusters A and B. In the case of EOF analysis, the correlation between the score time series for any two EOF modes is zero. That is, the regional division using the EOF analysis is constrained by a stronger condition than that for the cluster analysis.

We also examined correlation coefficients between the interannual variations of mean precipitation in the subregions determined here and of the nine indices listed in Table I. Hereafter, we use the abbreviations in Table I for the nine indices. The SST in the eastern equatorial Pacific in January [SSTE(Jan)] and in June (SSTE) are representative of the ENSO. Because the ENSO often reaches its mature phase in winter, we confirmed both the lagged and simultaneous correlations. The SST in the western equatorial Pacific (SSTW) and zonal wind around there (UW) represent the TBO of the Asian monsoon. When the strong Asian summer monsoon appears in the TBO, the SST anomalies tend to be positive in the tropical western Pacific, and the zonal winds at 850 hPa are enhanced in the equatorial western Pacific. Concurrently, anomalous easterlies tend to appear in the lower troposphere to the east of the positive SST anomalies (Shen and Lau, 1995; Chang and Li, 2000). The NAO index in June (summertime NAO; SNAO) is defined by subtracting the sea level pressure (SLP) in the area 80°–85°N, 10°–15°E from that in 55°–65°N, 5°–0°W, because the centres of the NAO migrate seasonally (Hurrell and van Loon, 1997). The precipitation seesaw index (PSI) is derived from Fukutomi et al. (2007) and represents the east–west variation of precipitation in northern Eurasia. When the PSI is positive, precipitation increases in the eastern part of northern Eurasia and decreases in the western part of northern Eurasia. The all-India rainfall (AIR) index is a good representative of the South Asian monsoon. The representative time series of the lower-frequency (LF) mode corresponding to the ENSO and the quasi-biennial oscillation (BO) mode with respect to the TBO of the Asian monsoon, obtained by the EOF analysis of the 130°–140°E mean precipitation in the western North Pacific was derived from Tomita et al. (2004). Note that the time period for AIR is 28 years (1979–2006), whereas the other variables cover 29 years (1979–2007).

Table I. Details of the indices used for the correlation analysis
AbbreviationParameterRegionPeriod
SSTE(Jan)SST(5°S–5°N, 100°–80°W)January, 1979–2007
SSTESST(5°S–5°N, 100°–80°W)June, 1979–2007
SSTWSST(5°S–5°N, 120°–140°E)June, 1979–2007
UWZonal Wind at 850 hPa(5°S–5°N, 140°–160°E)June, 1979–2007
SNAOSLP(55°–65°N, 5°–0°W)–(80°–85°N, 10°–15°E)June, 1979–2007
PSIPrecipitation((50°–70°N, 110°–135°E)–(50°–70°N, 60°–85°E))/2June, 1979–2007
AIRPrecipitationAll IndiaJune, 1979–2006
LF modePrecipitation, First EOF(20°–45°N, 130°–140°E)June + July, 1979–2007
BO modePrecipitation, Second EOF(20°–45°N, 130°–140°E)June + July, 1979–2007

3. Interannual variations in the three Baiu subregions

Using the cluster analysis, we first divided the Baiu front into three subregions (Figure 2): the western, central, and eastern regions (Figure 1). Hereafter, the three subregions are referred to as the western Baiu (WB), central Baiu (CB), and eastern Baiu (EB), respectively. The WB occupies the western part of the Baiu front, including eastern China, South Korea, and western Japan. The CB extends northeastward in the middle of the Baiu front, and the EB is located in the easternmost part of the Baiu front.

Figure 2.

Dendrogram of cluster analysis, applied to data in a region where the standard deviations of the Baiu precipitation are larger than 2.0 mm day−1 in (25°–40°N, 110°–170°E) (Figure 1)

The temporal variations and their power spectral densities (PSDs), which were estimated from the spatially averaged precipitation in the three subregions, clearly demonstrate the different temporal structures in these regions (Figure 3). Here, we abbreviate the interannual variabilities of spatially averaged precipitation in the WB, CB, and EB as IVWB, IVCB, and IVEB, respectively. Comparison of the three fluctuations demonstrates that the dominant periodicity tends to be long eastward. The quasi-biennial variability is predominant in the IVWB (Figure 3(a)), which seems to correspond to the BO mode of Tomita et al. (2004) reflecting the TBO of the Asian monsoon. On the other hand, the IVCB is characterized by a 3- to 4-year oscillation associated with the ENSO connecting to the LF mode of Tomita et al. (2004) (Figure 3(b)). In the IVEB, a 6-year periodicity is dominant (Figure 3(c)) and certainly longer than those of the IVWB and IVCB.

Figure 3.

The time series of precipitation in the (a) WB, (b) CB, and (c) EB (left) and the respective PSDs (right). For the left panels, the ordinate indicates the precipitation anomaly (mm day−1), and the abscissa denotes years. In the right panels, the ordinate shows PSDs (mm2 day−2), and the abscissa denotes the frequency expressed by cycles per year. Dotted lines in the right panels exhibit significance at the 95% level based on the red noise spectrum estimated from a first-order autoregression

The 6-year periodicity has been observed in the summertime precipitation in northern Eurasia (Fukutomi et al., 2003) and in the summertime temperature in northern Japan (Kurihara, 2003; Kanno, 2004). The NAO index in June also features the dominant quasi-biennial and 5- to 6-year periodicities (Figure 4), suggesting a connection to the IVWB or the IVEB. In boreal summer, the anomalous atmospheric circulations in the North Atlantic may have the potential to modify the circulations in and around East Asia through the stationary Rossby waves over northern Eurasia (Nakamura and Fukamachi, 2004).

Figure 4.

As in Figure 3 but for the SNAO index (Table I; the unit is hPa)

All correlations between the IVWB, IVCB, and IVEB are nonsignificant, and each of the time series has a different dominant period, suggesting that they are influenced by different physical mechanisms. To confirm whether the three interannual variations with periods of 2, 4, and 6 years have specific teleconnections, correlation coefficients of the nine indices in Table I were examined for the IVWB, IVCB, and IVEB (Table II). The boldfaced figures in Table II denote significant correlations at the 95% level ( ± 0.39). The IVWB has a significant positive correlation (+0.43) with the SSTW and a significant negative correlation (−0.42) with the UW. As these indices well represent the variation of the Asian monsoon TBO, these correlations suggest that the IVWB reflects part of this large-scale TBO. However, the correlation between the IVWB and the BO mode (+0.20) is somewhat weak, which may be caused by differences in the chosen regions and months. Tomita et al. (2004) defined the BO mode from the EOF analysis using precipitation in June and July. Note that the relationship between the IVWB and the variations of SSTE in January (+0.15) and in June (+0.08) is nonsignificant, indicating that the IVWB is independent of the ENSO. Chang and Li (2000) revealed that the TBO can theoretically occur without the SST anomalies in the equatorial eastern Pacific.

Table II. Correlation coefficients of the IVWB, IVCB, and IVEB with the nine indices in Table I
 SSTE (Jan)SSTESSTWUWSNAOPSIAIRLF modeBO mode
  1. Bold type indicates significant correlations at the 95% level ( ± 0.39).

IVWB0.150.080.430.420.11−0.110.170.050.20
IVCB0.410.220.30−0.240.190.160.470.570.01
IVEB0.33−0.05−0.01−0.210.410.44−0.370.210.16

The IVCB has significant correlations with the LF mode (+0.57), AIR (−0.47), and SSTE(Jan) (+0.41), although it shows a somewhat weak correlation with the SSTE (+0.22), which indicates a strong link between the IVCB and ENSO; that is, the Baiu precipitation has an interannual tendency to be larger than normal after warm ENSO events (Tanaka, 1997; Chang et al., 2000a, 2000b).

The interannual variability with a period of 5–6 years has not been well reported for the BFA. To reveal the spatiotemporal structure of this interannual variability, we first consider that the IVEB is significantly correlated with the interannual variations of the SNAO (+0.41) and the PSI (+0.44), which implies that the IVEB is linked to the summertime NAO through the variation in precipitation in northern Eurasia. The IVEB definitely has nonsignificant correlations with the SSTE(Jan) (+0.33), SSTE (−0.05), SSTW (−0.01), and UW (−0.21), indicating its independence from the ENSO and TBO of the Asian monsoon.

Variations with periods longer than 6 years are also confirmed in the time series in Figure 3(a) and (c). An upward trend is dominant from the 1980s to 1990s in Figure 3(a), whereas a downward trend is shown in Figure 3(c). Tomita et al. (2007) reported that precipitation in the western part of the Baiu front increased in the 1990s on a decadal time scale, but precipitation to the east decreased. They concluded that the interdecadal variations of BFA were caused by a change of surface heat fluxes due to SST variations in a region of the Kuroshio. The longer-term tendencies in the IVWB and IVEB are consistent with the results of Tomita et al. (2007).

To reveal the spatial structures associated with the IVWB, IVCB, and IVEB, we performed one-point correlation analysis based on the time series in Figure 3 (Figure 5). Significant positive correlations relevant to the IVWB clearly extend around the WB including South Korea and the northern East China Sea, while negative correlations broaden to the southeast (Figure 5(a)). In association with the IVCB, significant positive correlations elongate northeastward from the southeastern part of China and are broadly surrounded by negative correlations (Figure 5(b)). When the IVEB predominates, significant positive correlations tend to appear to the southeast of Japan and extend farther to the southeast of the CB, whereas significant negative correlations exist over the Sea of Okhotsk and the eastern Philippine Sea (Figure 5(c)).

Figure 5.

Correlation coefficients for precipitation associated with the (a) IVWB, (b) IVCB, and (c) IVEB. The contour interval is 0.2, and dotted contours denote negative correlations. Dark (light) shading indicates significant positive (negative) correlations at the 95% level ( ± 0.39). Zero lines are omitted for clarity

In the three panels, the regions with significant positive correlations are all consistent with the three subregions of the Baiu front estimated from the cluster analysis (Figure 1). Note here that the three areas of clusters are smaller than the regions with significant positive correlations, since the areas for cluster analysis are confined by the large standard deviation of the Baiu precipitation. In effect, anomalous circulations associated with these three areas affect the precipitation in the larger regions as shown in Figure 5 of the one-point correlation analysis. The regions with significant correlations are much wider and larger than the regions of cluster analysis.

To identify the spatial patterns associated with the IVWB, IVCB, and IVEB in the lower troposphere, we then estimated anomalous geopotential heights at 850 hPa using linear regression analysis based on the time series in Figure 3 (Figure 6). Corresponding to the IVWB, a significant anomalous high is observed to the south of Japan, which is explained as anomalous westward extension of the NPSH; in contrast, a significant anomalous low appears near the Yellow Sea (Figure 6(a)). The geostrophic balance suggests that southwesterlies and their convergence are enhanced over the East China Sea between the anomalous high and low. On the other hand, when the IVCB is predominant, an anomalous high appears in the western part of NPSH, and a significant anomalous low extends zonally east–northeastward from southeastern China (Figure 6(b)). These anomalous high and low indicate that southwesterlies are stronger than normal to the southeast of Japan. In Figure 6(c), with regard to the IVEB, we can identify a significant anomalous low to the southeast of Japan and the strengthened Okhotsk High in the western part of the Sea of Okhotsk. Furthermore, a small but significant anomalous high is located farther to the southeast of Japan near 25°N, 170°E. In the geostrophic approximation, the anomalous southwesterlies extend around 25°N, 150°E, whereas the counterpart northeasterlies are anomalously dominant in and around northern Japan.

Figure 6.

Regression coefficients of geopotential height at 850 hPa for the (a) IVWB, (b) IVCB, (c) IVEB, estimated based on the time series in Figure 3. The contour interval is 1.0 gpm, and zero lines are omitted for clarity. Solid contours denote regions with an anomalous high, and dotted contours indicate regions with an anomalous low. Dark (light) shading represents regions with significant positive (negative) correlations at the 95% level ( ± 0.39)

In anomalous patterns of geopotential height associated with both the IVWB and IVCB, the anomalies are distributed in the Tropics (Figure 6(a) and (b)), whereas the anomalous patterns associated with the IVEB can be found to the north of 20°N (Figure 6(c)). This difference implies that the IVWB and IVCB are largely controlled by the atmospheric circulations in the Tropics, whereas the IVEB is connected to the variations in the mid-latitudes.

The anomalous winds in the lower troposphere modify the horizontal water vapour flux into the Baiu front, and its divergence/convergence anomalies control the BFA. To demonstrate these effects, the vectors of the vertically integrated (1000-300 hPa) water vapour flux (Q) and its divergences were diagnosed using a similar regression technique (Figure 7). Here, Q is defined as follows:

equation image(5)

where g is the acceleration of gravity, u is a horizontal wind vector, and q is specific humidity that is very small and negligible above 300 hPa. In the divergence/convergence field, a positive value represents divergence. Figure 7(a) shows that anomalous Q significantly converges around the East China Sea to central Japan when the IVWB is predominant, and the axis of large east–northeastward Q appears over western Japan around 30°N, 130°E, which corresponds to the WB. Concurrently, anomalous divergences appear to the south. The spatial pattern associated with the IVCB is represented by the significant convergence to the southeast of Japan, whereas the significant divergence extends farther southeast (Figure 7(b)). Small but significant divergences are also found in and around the Yellow Sea. Strong southwesterlies are located in the anomalous convergence, i.e. the CB. When the IVEB is active, the significant convergences tend to extend east–northeastward from 25°N, 125°E, and the significant divergences appear in the regions surrounding Japan and to the southeast of the anomalous convergence (Figure 7(c)). The axis of southwesterlies is clearly located in the EB. In general, the regions of anomalous convergence of Q are consistent with those that show significant positive correlations in precipitation (Figure 5) and with regions between the anomalous high and low in the geopotential height at 850 hPa (Figure 6). The anomalous divergence/convergence of Q reflects the anomalous pressure systems in the lower troposphere associated with the interannual variations in the three Baiu subregions.

Figure 7.

Regression coefficients of Q (vectors), and the divergences (contours). The scale of Q is shown at the bottom. Zero lines and vectors smaller than 10 kg m−1 s−1 are omitted for clarity. The contour interval is 5.0 × 10−6 kg m−1 s−1, and the negative values are depicted by dotted contours. Dark (light) shading represents regions with significant positive (negative) correlations at the 95% level ( ± 0.39)

4. Factors affecting the BFA

4.1. Effects from the Tropics

The Baiu front was divided into three subregions with different interannual periodicities of approximately 2, 4 and 6 years. Tomita et al. (2004) have discussed the spatiotemporal structures of the shorter 2- and 4-year periods of interannual variation. Their findings revealed that the two variations are forced by effects from the Tropics. This subsection examines the differences and similarities with Tomita et al.'s (2004) results.

To clearly demonstrate the effects from the Tropics, we examined the associated anomalies of SST and large-scale circulations at 850 hPa (Figure 8). In the IVWB, an anomalous anticyclonic circulation appears to the south of Japan around 25°N, 140°E, and a wave train extends northeastward from the anomalous anticyclonic circulation (Figure 8(a)). The anomalous convergence of Q centred at 30°N, 130°E (Figure 7(a)) occurs between the anomalous anticyclonic and cyclonic circulations near the Japan Sea, and the anomalous southerlies near 25°N, 120°E come from the easterlies of the equatorial zonal circulation in the tropical Pacific. Positive SST anomalies in the tropical western Pacific around 0°, 125°E are likely to force anomalous ascents there, whereas negative SST anomalies extending to the east are favourable for the anomalous descents. Shen and Lau (1995) pointed out that the precipitation of the East Asian summer monsoon is positively correlated with SST in the tropical western Pacific and negatively correlated with the surface zonal winds in the tropical central Pacific from spring to the following winter. Our study confirms that such a significant correlation appears among the IVWB, SST in the tropical western Pacific, and the variation of zonal winds at the 850 hPa in the tropical central Pacific (Table II). The expected anomalous ascents with respect to the positive SST anomalies in the tropical western Pacific may excite the anomalous anticyclonic circulation to the north through the anomalous local Hadley circulation, which further forces the anomalous water vapour transport and leads to the increased precipitation in the WB. The anomalous SST pattern is slightly different from that of the BO mode reported by Tomita et al. (2004), which possibly contributed to the weak correlation between the IVWB and BO mode (Table II).

Figure 8.

Regression coefficients of SST and horizontal wind at 850 hPa for the (a) IVWB, (b) IVCB, and (c) IVEB, estimated based on the time series in Figure 3. The contour interval is 0.1 K, and dotted contours denote negative values of SST. Dark (light) shading indicates regions with significant positive (negative) correlations at the 95% level ( ± 0.39) for SST. The scale of wind vectors (unit is m s−1) is shown at the bottom right

When the IVCB is predominant, an anomalous anticyclonic circulation, which is recognized as the westward extension of the NPSH, appears at around 25°N, 160°E and strengthens the anomalous convergence of Q to the northwest with a small anomalous cyclonic circulation near Japan (Figure 8(b)). Another large anomalous cyclonic circulation appears to the northeast of the anomalous anticyclonic circulation. The weak negative SST anomalies appear in the southern part of the anomalous anticyclonic circulation, and weak positive SST anomalies are located in the northern part of that circulation (Figure 8(b)), suggesting that the anomalous anticyclonic circulation could be part of the Pacific–East Asian teleconnection from the winter of a warm ENSO event to early summer (Wang et al., 2000). The distribution of anomalous SST in January is similar to the pattern when a typical El Niño event occurs (not shown), but different from the warm ENSO pattern in June. In fact, the correlation between the IVCB and the SSTE(Jan) is significantly positive (+0.41; Table II). These results are consistent with the findings of Wang et al. (2000) and Chang et al. (2000a, 2000b). In addition, the strong northeasterlies appearing in the western Indian Ocean indicate a weak Indian summer monsoon, which is consistent with the significant negative correlation between the IVCB and AIR (−0.47; Table II). This negative correlation may be substantial, because Rossby waves, which emanate from the anomalous cyclonic circulation in the upper troposphere with the weak Indian summer monsoon, possibly propagate and reinforce the anomalous anticyclonic circulation to the south of Japan (Krishnan and Sugi, 2001).

The IVEB has no significant correlations in the Tropics (Figure 8(c)). Significant correlations only appear around Japan with the negative anomaly in the SST field. In early summer, the anomalous northeasterlies in the lower troposphere around Japan cause cold advection (Nakamura and Fukamachi, 2004). This cold advection seems to induce the significant negative SST anomalies because mid-latitudinal ocean has small heat capacity due to the shallow upper ocean mixed layer. The IVEB does not seem to be controlled by the tropical effects.

The atmospheric circulation anomalies associated with the IVWB are linked to the TBO through specific anomalous meridional circulation between the Tropics and mid-latitudes. The IVCB coincides with ENSO events through the Pacific–East Asian teleconnection in the tropical Pacific. On the other hand, the IVEB shows significant correlations with atmospheric circulations in the mid-latitudes (Figure 8(c); Table II). In the next subsection, we discuss the effects of mid-latitude circulations on the IVEB.

4.2. Effects of mid-latitude circulations

This subsection confirms the effects of mid-latitude circulations on the IVEB and examines the underlying physical mechanisms. Because the western part of the Okhotsk High in the lower troposphere is strengthened when the amplitudes of the IVEB are positively large (Figure 6(c)), the IVEB is probably linked to the variations of the Okhotsk High, which are further forced by the blocking high in the upper troposphere around the Sea of Okhotsk. Some studies have suggested that this upper blocking high is modified by the propagation of Rossby waves from the west in early summer (Ogi et al., 2003; Tachibana et al., 2004).

The IVEB has the 6-year period variation (Figure 3(c)) that is found in mid-latitude circulations such as the NAO in June (Figure 4), the summertime precipitation in northern Eurasia (Fukutomi et al., 2003), and the summertime temperature in northern Japan (Kurihara, 2003; Kanno, 2004). The regions associated with the 6-year period variation are located from northern Europe to northeastern Asia, which are consistent with the route of Rossby wave propagation in the upper troposphere. The 6-year variation of the IVEB may be forced by Rossby wave propagation from the west.

To confirm the routes of Rossby wave propagation in the upper troposphere in June, we first examined the long-term mean of zonal wind in the upper troposphere, because the axes of upper tropospheric westerlies act as a waveguide for Rossby wave propagation (Hoskins and Ambrizzi, 1993). Figure 9 shows the distributions of the long-term mean of zonal winds at 300 hPa and the total wave number (Ks) in local Cartesian coordinates. Here, Ks is estimated from

equation image(6)

where

equation image(7)

where β is the beta parameter, and U denotes the zonal wind. Figure 9(a) illustrates that the strong westerlies in the North Atlantic separate into two parts over Europe; one is the so-called Asian jet along 30°N, and the other is the North Eurasian jet along 70°N. Local maximums of Ks also appear on the two axes, although the maximum on the northern jet is smaller than that on the southern one (Figure 9(b)). Not only the Asian jet along 30°N but also the westerly axis along 70°N over northern Eurasia seems to be preferable for Rossby wave propagation in early summer. On the other hand, the westerlies are locally weak to the east of Lake Baikal around 50°N, 110°E (Figure 9(a)), which also leads a small Ks there (Figure 9(b)). This deceleration of the westerlies may be attributable to the thermal wind caused by the differential heating between the Sea of Okhotsk and the land to the north in early summer, and would prevent Rossby waves from propagating eastward. The Rossby waves could propagate eastward on the zonally elongated Asian jet along 30°N and to the west of 110°E on the North Eurasian jet along 70°N.

Figure 9.

Long-term (1979–2007; 29 years) means of (a) zonal wind and (b) total wave number (Ks) at 300 hPa. The contour intervals are (a) 2.0 m s−1 and (b) 1.0 (no unit), and the regions with negative values are indicated by dotted contours in (a). Shading represents regions of more than 8.0 m s−1 in (a) and 2.0 in (b), respectively

To underscore the characteristics of the 6-year variation appearing in the IVEB, we applied the band-pass filter of Murakami (1979) to the IVEB time series (Figure 3(c)) and the SNAO time series (Figure 4), in which 0.5 of the frequency response function was set at 5 and 8 years (Figure 10). It is interesting that the two 6-year oscillations are almost in phase, except for differences around 2000, when the peak of the IVEB is small compared to that of the SNAO.

Figure 10.

Band-pass filtered time series of the normalized IVEB (solid line; Figure 3(c)) and the normalized SNAO index (dotted line; Figure 4), where the period range is set from 5 to 8 years. The ordinate indicates the standard deviation (σ), and the abscissa denotes years

To reveal the significant signals of the 6-year variability associated with the IVEB, we performed two composite analyses defined by half of the difference between the means of positive and negative years (positive minus negative). First, we will display the result of the non-filtered composite based on the time series in Figure 3(c). However, because noise other than periodicity was somewhat large, we could not confirm well whether this result truly shows the 6-year period variation. Therefore, we additionally applied the band-pass filter described above for the atmospheric parameters and then estimated the composite anomalies based on the time series in Figure 10 (solid line) to emphasize the 6-year period variation more clearly.

To confirm the Rossby wave propagation in the upper troposphere, we diagnosed the anomalies of the geopotential height at 300 hPa and of the wave activity flux (WAF) vectors for stationary Rossby waves at this height (Figure 11). The WAF vectors are parallel to the group velocity and show the direction and strength of the energy propagation of stationary Rossby waves (Takaya and Nakamura, 2001). In the non-filtered result (Figure 11(a)), it is notable that the meridional tripole (i.e. SNAO) appears in the North Atlantic, and a significant anomalous low is located around Japan. The upper tropospheric Rossby waves that are amplified by the SNAO propagate on the North Eurasian jet along 60°N to the east of Lake Baikal and change its direction to south-southeastward because of the deceleration of westerlies there (Figure 9(a)). Then, the Rossby waves arrive in Japan and form the negative geopotential height anomalies there with an equivalent barotropic structure (Figure 6(c)). In addition, the Rossby waves seem to propagate on the Asian jet along 30°N, which may be part of the circumglobal teleconnection in boreal summer found by Ding and Wang (2005). Iwao and Takahashi (2008) pointed out that Rossby waves are amplified in and around the anomalous blocking high in the North Atlantic and propagate eastward to northeastern Asia through both northern and southern Eurasia.

Figure 11.

Composite anomalies of geopotential height (contours) and WAF (vectors) at 300 hPa, estimated from the composite analysis based on the time series in (a) Figure 3(c) and (b) Figure 10 (solid line). The contour interval is 5 gpm, and regions with negative values are indicated by dotted contours. Dark (light) shading represents the region with significant positive (negative), estimated by Welch's t-test at the 95% level. The scale of the vector of WAF (unit is m2 s−2) is shown at the bottom. Vectors smaller than 0.2 m2 s−2 are omitted for clarity

With the band-pass filter (Figure 11(b)), the noise due to the variation, except for the IVEB, seems to decrease; however, the anomalous pattern is still like that in Figure 11(a). The significant tripole signal in the North Atlantic and the significant anomalous low around Japan are also identified. Additionally, the zonal dipole, with centres at 60°N, 65°E, and 60°N, 105°E, is developed in central northern Eurasia and reflects the anomalous seesaw variation in precipitation (Fukutomi et al., 2003). In the upper troposphere, the positive geopotential height anomalies over the Sea of Okhotsk are small, but the anomalous high appears in the lower troposphere (Figure 6(c)). This anomaly pattern suggests that the anomalous high over the Sea of Okhotsk has nearly baroclinic structure, possibly caused by the anomalous cold advections around the Sea of Okhotsk in the lower troposphere. The propagation of Rossby waves along the North Eurasian jet is also similar to the results from the non-filtered estimations, but the southern linkage seems to be weak (Figure 11(b)). Thus, the 6-year variation associated with the IVEB is linked to the SNAO through the Rossby wave propagation over northern Eurasia.

The phase difference and small amplitude around 2000 suggest that other external factors modified the 6-year variation of the IVEB. Around 2000, the position of the strong westerly region acting as the waveguide might have changed for some reason; Rossby waves amplified by the SNAO might not have reached Japan or might not have formed the anomalous circulation suitable for precipitation in the EB. More detailed analysis is needed to investigate these possibilities.

5. Summary

The Baiu front is a quasi-stationary front appearing in the northwestern periphery of the NPSH. This phenomenon, unique to East Asia, is affected by the large-scale circulations around it, including the NPSH, the Okhotsk High, the heat low over the Asian continent, and the southwesterly monsoon. To grasp how such large-scale circulations modify the interannual variability of BFA, we diagnosed the detailed spatiotemporal structures of BFA after dividing the Baiu region into three subregions; we then examined the physical mechanisms of the interannual variations in each of the three subregions. The findings of this work are summarized in Table III.

Table III. Summary of the results of this study
RegionPeriodAssociated oscillation
Western Baiu2 yearsTBO
Central Baiu4 yearsENSO
Eastern Baiu6 yearsNAO

In June, the strongest Baiu precipitation occurs around Japan, and the structure of the Baiu front is exhibited more clearly than in other months. Therefore, the BFA in June tends to correspond with the typical case around Japan during the Baiu season. Using cluster analysis with the Ward method, the spatiotemporal coherence of interannual variability was first examined for the BFA in June. By this approach, we could divide the Baiu front into three subregions: (1) the WB, (2) the CB and (3) the EB. Power spectral analysis revealed that the dominant periods in their temporal variations are long eastward, i.e. 2–3 years in the WB, 3–4 years in the CB, and 6–7 years in the EB. The interannual variations of spatially averaged precipitation in the WB, CB, and EB (IVWB, IVCB, and IVEB) were then correlated with various indices, such as those associated with the tropospheric biennial oscillation (TBO), El Niño/Southern Oscillation (ENSO), and North Atlantic Oscillation (NAO). The IVWB has a significant positive correlation with SST in the tropical western Pacific and a negative correlation with the zonal wind at 850 hPa to the east, indicating the linkage with the TBO of the Asian monsoon. The IVCB is positively correlated with the lower-frequency (LF) mode of Tomita et al. (2004), which is associated with ENSO. The IVEB, however, is not correlated to the tropical variations. Instead, the IVEB shows significant correlations with the mid-latitude circulations, particularly the summertime NAO (SNAO) and the zonal dipole in anomalous precipitation in central Eurasia.

The characteristics of the shorter two variations (i.e. the IVWB and IVCB) are summarized as follows. When the positive IVWB is predominant, positive and negative anomalies appear to the south of Japan and around the Yellow Sea in the geopotential height field at 850 hPa (Figure 6(a)). The anomalous low-level winds are coherent with flows of vertically integrated (1000–300 hPa) water vapour flux (Q), and the anomalous convergences of Q enhance the precipitation in the WB (Figure 7(a)). The detailed spatial pattern of SST anomalies is characterized by the positive SST anomalies found in the tropical western Pacific and the negative ones in the tropical central Pacific. These characteristics are consistent with the interannual variation of the TBO in the Asian monsoon but somewhat different from that in the quasi-biennial (BO) mode described by Tomita et al. (2004) (Figure 8(a)). However, the quasi-biennial tendency in the IVWB is concluded to link to the TBO of the Asian summer monsoon by the correlations with other indices associated with this TBO.

As the predominant characteristics of positive years in the IVCB, negative geopotential height anomalies appear near Japan and positive ones occur in the western part of the NPSH at 850 hPa (Figure 6(b)). Because anomalous convergence of southwesterlies in Q is dominant to the southeast of Japan, the precipitation in the CB is increased (Figure 7(b)). The anomalous anticyclonic circulation at 850 hPa associated with the IVCB appears around 25°N, 160°E, which is coherent with the Pacific–East Asian teleconnection (Wang et al., 2000) from winter to early summer with the 3- to 4-year variation of the ENSO cycle.

This work has emphasized the characteristics of the IVEB with a specific 6-year periodicity that shows nonsignificant correlations with variations in the Tropics but significant correlations with mid-latitude circulations. Figure 12 presents a schematic diagram of the IVEB. When the amplitude of the IVEB is positively large, negative geopotential height anomalies at 850hPa appear to the southeast of Japan, whereas positive ones are strengthened in the western part of the Sea of Okhotsk (Figure 6(c)). The anomalous circulation around Japan causes the anomalous convergence of Q in the EB (Figure 7(c)). In the upper tropospheric geopotential height field in early summer, the significant anomalies in the North Atlantic are clearly identified as the SNAO with the 6-year oscillation (Figure 11). The Rossby waves amplified in the North Atlantic propagate eastward along the waveguides (Figure 9). The northern axis of westerlies is locally weak to the east of Lake Baikal, and the weak westerlies prevent the Rossby waves from propagating eastward. The Rossby waves then propagate southeastward to Japan and form the negative geopotential height anomalies with an equivalent barotropic structure there (Figure 11). As such, it is confirmed that the IVEB is linked to the SNAO by the propagation of Rossby waves via northern Eurasia.

Figure 12.

Schematic diagram illustrating the anomaly fields in June associated with the IVEB. The circles with a plus (minus) sign indicate anomalies of positive (negative) geopotential height in the upper troposphere. The vectors depict the propagating Rossby waves, in which large, grey vectors also correspond to the axes of upper tropospheric westerlies that are regarded as typical waveguides of Rossby waves

We examined the detailed spatiotemporal structures associated with the interannual variations in the three subregions of the Baiu front and the physical mechanisms. However, some questions still remain. As an example, do the spatiotemporal structures in other months (i.e. May or July) have different characteristics? Because a few heavy rainfall events can easily modify the monthly mean precipitation in the Baiu season, the amount of precipitation would be uneven in each month (Ninomiya and Mizuno, 1987). The BFA may also be different in those months, leading to differences in the interannual variation. As another example, the correlation of Rossby wave activity associated with the IVEB should be examined with the circumglobal teleconnection in boreal summer found by Ding and Wang (2005). As such, further studies are needed to clarify the mechanisms of the interannual variations of BFA in more detail.

Acknowledgements

The authors thank two anonymous reviewers for their careful reading and constructive suggestions that led to a much-improved manuscript. This work was supported by Grants-in-Aid for Scientific Research (17540413, 20240075) from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan.

Ancillary