The 30-60-day Intraseasonal Oscillations over the Subtropical Western North Pacific during the Summer of 1998

The 30-60-day range represents a major band of periods for intraseasonal oscillations (ISOs), the variance of which is strong during summer over the South China Sea (SCS), tropical western North Pacific (WNP), and Indian Ocean (Knutson et al., 1986; Lau and Chan, 1986, 1988; Ren and Huang, 2003). At that time of year, 30-60-day oscillations exhibit a prominent northward and northwestward propagation over the SCS and WNP (Lau and Chan, 1986; Hsu and Weng, 2001; Mao and Chan, 2005; Tsou et al., 2005).

Although relatively weaker than in the tropics, 30-60-day oscillations are also pronounced over the subtropical WNP during summer (Lau and Chan, 1988; Ren and Huang, 2003). However, ISOs exhibit different features of propagation over the subtropical and tropical WNP. For instance, the northward or northwestward propagation of convection ISOs is basically confined to the tropics over the WNP (Lau and Chan, 1986; Hsu and Weng, 2001; Mao and Chan, 2005; Tsou et al., 2005). Furthermore, (Han et al., 2006) noticed that 30-60-day oscillations propagate westwards over the subtropical WNP in some summers and have an impact on rainfall over eastern China. However, subtropical ISOs have been much less studied in comparison with tropical ISOs, and are thus far from well understood.

Although the northward propagation of tropical convection ISOs cannot reach the extratropics over the WNP, there is evidence for tropical-extratropical interaction over the WNP and East Asia on intraseasonal timescales (Chen and Murakami, 1988; Kawamura and Murakami, 1995; Fukutomi and Yasunari, 2002). Recently, Mao et al. (2010) revealed that there is a seesaw pattern in 20-50-day convection oscillations between the Yangtze basin in China and the SCS/WNP. They suggested that the ISOs propagate northward and northwestward over the tropical WNP, and result in Yangtze rainfall fluctuations through modulating the shift of the subtropical high over the WNP. Gao et al. (2012) also suggested that 30-60-day tropical heating oscillations can affect June rainfall anomalies over South China.

In this paper we show that 30-60-day oscillations were extremely dominant over both the subtropical and tropical WNP in the summer of 1998, which provided an opportunity to investigate the features of 30-60-day oscillations over the subtropical WNP and their linkage to tropical oscillations. In fact, eastern China experienced a series of severe floods during that summer, which Zhu et al. (2003) suggested were related to the propagation and activity of 30-60-day oscillations over the WNP. They also indicated that 30-60-day oscillations dominate the SCS during the summer.

In the present study we use daily outgoing longwave radiation (OLR) data from the National Oceanic and Atmospheric Administration (NOAA) for the 31 years from 1979 to 2009 to obtain the 30-60-day component for each year by applying a band-pass Lanczos filter to the annual cycle-removed daily anomalies. The annual cycle for each year is defined by the sum of the annual mean and the first three harmonics at each grid point. The results for summer (June-July-August; JJA) are the focus, though some results are also given for May and September.

Figure 1 shows the variance of 30-60-day oscillations in summer 1998 and the ratio of variance of the oscillations to total variance. The climatological results are also given to facilitate comparisons. For climatology, 30-60-day oscillations were strongest over the SCS and the Philippine Sea and became weaker with an increase in latitude (Fig. 1b). The variance of 30-60-day oscillations was 200-300 W² m^-4 over the SCS and tropical WNP, and was 100-200 W² m^-4 over the subtropical WNP. However, the ratios of 30-60-day oscillation variance to total variance were similar between the tropics and subtropics (Fig. 1d). The ratio was about 0.10 over the subtropical WNP, and was only slightly lower than that over the tropical WNP.

Figure 1.  Variance of 30-60-day OLR oscillations in (a) summer (JJA) 1998 and (b) the climatological mean averaged over summers during the period 1979-2009 (units: W² m^-4), and the ratio of the 30-60-day band variance to total variance in (c) summer 1998 and (d) the climatology.

The most striking feature of summer 1998 was that 30-60-day oscillations were extremely strong over the subtropical WNP and a part of South China (Fig. 1a). The variance ranged from 300 to 1000 W² m^-4 over these regions, much greater than the climatology and comparable even to the variance in the tropics. Actually, the variance of oscillations was also very strong in the tropics. The variance ranged from 300 to 1000 W² m^-4 over the SCS and tropical WNP, which was also much greater than the climatology. There were two clear bands of strong variance of 30-60-day oscillations in summer 1998: one over the subtropics and the other over the tropics, in contrast to the climatological distribution. The variance was relatively weaker along 20°-22.5°N, making the subtropical and tropical strong variance bands distinct.

The ratios of 30-60-day oscillation variance to total variance were also much greater over the WNP in summer 1998 (Fig. 1c). The variance of 30-60-day oscillations was almost half the total variance along 27.5°N over the WNP. The ratios also exhibited two bands of strong values: subtropical and tropical bands, with relatively weaker values between these bands.

Figure 2 shows the year-to-year variation of 30-60-day band variance averaged over the regions (25°-30°N, 110°-140°E) and (10°-17.5°N, 105°-130°E). These two regions represent the subtropical and tropical bands of maximum variance, respectively. Figure 2a clearly indicates that the 30-60-day oscillations over the subtropical North Pacific were extremely strong during summer 1998. While all other years had a variance ranging from 50 to 200 W² m^-4, the variance in summer 1998 was as great as 500 W² m^-4, about four times greater than the climatological variance (133.5 W² m^-4).

Figure 2.  The 30-60-day band variance during summer (JJA) averaged over (a) the subtropical region (25°-30°N, 110°-140°E) and (b) the tropical region (10°-17.5°N, 105°-130°E) (units: W² m^-4).

The variance averaged over the tropical band was also very strong in summer 1998, only second to summer 1979 (Fig. 2b). During summer 1979, the variance of 30-60-day oscillations was very strong in the central eastern extent of the SCS (not shown), making that summer have the highest variance in the tropical band.

Figure 3 shows the seasonal evolution of 30-60-day oscillations along the specified latitudes over the WNP in summer 1998. During that summer, the oscillations propagated westward over the subtropical WNP, i.e., along the latitudes of 25°N, 27.5°N and 30°N, from 150°E to 110°E, roughly at a speed of 3-4 m s^-1. The oscillations were in phase between these latitudes, suggesting that 30-60-day oscillations were consistent over the subtropical WNP. The subtropical oscillations seem to have propagated more slowly in early summer and faster in late summer along these latitudes. The clearest propagation and strongest oscillations appeared during the period of summer, i.e., during JJA. The oscillations became much weaker below and above these latitudes, consistent with low variance shown in Fig. 1a.

Figure 3.  Seasonal evolution of 30-60-day oscillations along (a) 25°N, (b) 27.5°N, (c) 30°N, and (d) 32.5°N. The contour interval is 10 W m^-2, and the zero contour line is omitted. The positive and negative values are shown as light and heavy shading, respectively.

Figure 4 shows the seasonal evolution of 30-60-day oscillations along the specified longitudes. The oscillations propagated northwards in the tropics, which is consistent with results reported in previous studies (Lau and Chan, 1986; Hsu and Weng, 2001; Mao and Chan, 2005; Tsou et al., 2005). The northward propagation of convection oscillations was confined in the tropics along the longitudes of 110°E, 120°E, and 140°E during the entire summer, and along 130°E during late summer. The oscillations, however, propagated from the tropics into the subtropics during early summer along 130°E, and tended to become stronger with an increase in latitude.

Figure 4.  The same as Fig. 3, but along (a) 110°E, (b) 120°E, (c) 130°E, and (d) 140°E.

Along 130°E and 140°E, the northward propagation of convection oscillation could be seen in the subtropics during early summer. The speed of northward propagation was approximately 1 m s^-1, much lower than that of westward propagation in the subtropics. The oscillations exhibited a similar feature of propagation along 125°E and along 135°E (not shown). The direct propagation of oscillations along these longitudes may have contributed to the relatively larger ratio of variance along 125°-135°E between the subtropical and tropical bands shown in Fig. 1c. These results indicate that the oscillations over the subtropical WNP can be affected by tropical oscillations through the direct northward propagation of the oscillations from the tropics into the subtropics.

Another feature of tropical-subtropical linkage shown in Fig. 4 is the seesaw patterns between the tropics and subtropics. These patterns can be seen clearly during the entire summer along the longitudes of 110°E, 120°E, and 140°E, and during late summer along 130°E. The tropical oscillations were dominant between 10°N and 15°N, and the subtropical ones were dominant between 25°N and 30°N, which was in good agreement with the locations of subtropical and tropical bands of larger variance shown in Fig. 1a. The seesaw patterns are also noticeable even along the longitude of 130°E during early summer, when the northward propagation of the oscillations from the tropics into the subtropics was dominant (Fig. 4c).

Figure 5.  Leading modes of EOF analysis (EOF1) and corresponding principle components (PC1) for (a) 25°N, (b) 27.5°N, (c) 30°N, and (d) 32.5°N. The ratios of variance explained by the leading modes are given in the upper right corner of each panel.

Figure 6.  The same as Fig. 5, but for the results of the EOF analysis after removing summer 1998.

In each seesaw pattern, the tropical oscillations appeared prior to their subtropical counterparts, leading by 6-7 days. These tropical-subtropical seesaw patterns, as well as the earlier appearance of tropical anomalies, have also been shown to exist in lead-lag anomalies regressed onto rainfall oscillation over the Yangtze basin using data from 1979 to 2003 (Mao2010, Fig. 7). The lead of tropical oscillations suggests that subtropical oscillations are affected by tropical oscillations through the seesaw patterns.

We have shown that, during summer 1998, 30-60-day convection oscillations propagated westwards over the subtropical WNP, and that there was a seesaw pattern between the tropical and subtropical oscillations. But are these characteristics unique for that particular summer, or are they common phenomena for all summers? This question will now be examined through the results of an Empirical Orthogonal Function (EOF) analysis of OLR for the period 1979-2009.

Figure 5 shows the leading modes of EOF analysis (EOF1) and the corresponding principle components (PC1) for the specified latitudes in the WNP. The domains of the EOF analysis were the same as the figure shows, i.e., 100°E-180° and 1 June to 31 August. As can be seen, the patterns and phases of the leading modes are very similar to those of the oscillations during summer 1998 shown in Fig. 3 along the latitudes of 25°N, 27.5°N and 30°N. The oscillations propagate westwards from 150°E to 110°E at the same speed as the subtropical oscillations during summer 1998, and are stagnant over 150°E-180° along these latitudes. Actually, PC1 is strongest in 1998 for these latitudes, suggesting a unique feature of westward propagation of oscillations over the subtropical WNP during that particular summer. In comparison with the oscillations during summer 1998, the leading modes show relatively stronger anomalies over 150°E-180°. The leading mode along 32.5°N also exhibits a westward propagation over the subtropical WNP and eastern China, but explains the lowest variance, and PC1 is not very strong in 1998.

The time series of PC1 indicate that the oscillations in 1998 may have a crucial influence on the leading modes. Therefore, we repeated the EOF analysis after removing 1998 to investigate whether the westward propagation of oscillations over the subtropical WNP is a common phenomenon, or only induced by the oscillations in summer 1998. Figure 6 shows the leading modes and corresponding principle components after removing the summer of 1998 from the EOF analysis. Without 1998, the oscillations still show a westward propagation over the subtropical WNP and eastern China, but the anomalies become much weaker over these regions. In addition, without 1998, the starting points of westward-propagating oscillations are located obviously westwards along 25°N and 27.5°N. The interannual variations of the corresponding principle components are also similar to those when 1998 is included. The correlation coefficients between the PC1s with and without 1998 are 0.95, 0.76, 0.76 and 0.97 for 25°N, 27.5°N, 30°N and 32.5°N, respectively. The relatively weaker correlation coefficients for 27.5°N and 30°N are likely due to the oscillations along these latitudes being the strongest in comparison with the climatology (Figs. 1a and b). These results suggest that the westward propagation of subtropical oscillations could be a common phenomenon, although the feature is less prominent when excluding data for summer 1998.

This study investigated the features of 30-60-day convection oscillations over the subtropical WNP and tropical-subtropical linkage of the oscillations over the WNP during summer 1998. The results indicated that 30-60-day oscillations were extremely strong in summer 1998 over both the subtropical and tropical WNP, with two separate zonal bands of maximum variance over the two regions, and the variance was roughly two to four times stronger than the climatology. The oscillations were particularly strong over the subtropical WNP and even comparable to their tropical counterparts.

Different to over the tropical WNP, where 30-60-day oscillations propagated northwards or northwestwards, over the subtropical WNP and eastern China they propagated westwards. The westward propagation of subtropical oscillations seems to be a common phenomenon, although it was particularly pronounced during summer 1998—the feature becomes vaguer when excluding data for that year.

It was also found that the tropical 30-60-day oscillations contributed to subtropical ones during summer 1998, and that this contribution was achieved through two mechanisms: (1) direct propagation from the tropics into the subtropics; and (2) a seesaw pattern between the tropics and subtropics. The latter of these two mechanisms was the more dominant in 1998, but as to which is the most influential at any given moment seems to depend on longitude and subseasonality.

Following this work, many questions remain. For instance, what kinds of circulation oscillations are associated with the 30-60-day convection oscillations of summer 1998? What kinds of roles do these circulation oscillations play in linking tropical and subtropical oscillations? Do midlatitude circulation oscillations have any impacts upon subtropical convection oscillations, as suggested by (Lu et al., 2007) and (Wu and Chou, 2012)? Why do 30-60-day convection oscillations exhibit such a unique feature over the WNP, particularly in the subtropics? Are there any especially anomalous mean flows or atmosphere-ocean interactions responsible for this unique feature? These questions require further research, and relevant analyses are ongoing.