The summer western Pacific subtropical high (WPSH) has intensified during the past three decades. However, the underlying mechanism is not yet well understood. Here, it is shown that baiu rainband activity in midsummer, which is part of the East Asian summer monsoon, plays an important role in recent intensification in the WPSH along the baiu rainband. In contrast with the WPSH, the summer Okhotsk high, which is located to the north of the baiu rainband, has weakened during the past three decades. The north–south contrasting changes between the two highs reflect a response to northward-moved and enhanced baiu heating, which intensifies the upper-tropospheric ridge, resulting in the baroclinic intensification of the WPSH. Regional climate model experiments also support the observational analysis. Therefore, baiu convective activity in midsummer can act as a major driver for the WPSH intensification. The results here suggest that the mechanism intensifying the summer North Pacific subtropical high clearly differs between the western and eastern Pacific.
Over recent summers, East Asia has frequently experienced previously unprecedented heat waves. During the summer, East Asia is under the influence of the western Pacific subtropical high (WPSH), which plays a fundamental role in abnormally hot summers. Indeed, the western ridge of the WPSH has been observed to extend significantly westward (Lu and Dong 2001; Sui et al. 2007; Zhou et al. 2009). In addition to abnormally hot summers, the WPSH strongly affects tropical cyclone tracks (Stowasser et al. 2007; Du et al. 2011). However, it is difficult to predict the development of the WPSH (Kosaka et al. 2012) because there are many factors that may drive its interannual variability, such as the Pacific–Japan (PJ) pattern (Nitta 1987), the Silk Road pattern (Enomoto et al. 2003), the Indian Ocean capacitor effect (Xie et al. 2009), and local sea surface temperature (SST) (Xiang et al. 2013).
During early summer (June and July), East Asia is characterized by a quasi-stationary baiu rainband, which is part of the East Asian summer monsoon and extends northeastward over the western Pacific (Fig. 1a). The baiu rainfall is supplied by moisture transported by the lower-tropospheric southwesterlies along the western ridge of the WPSH (e.g., Akiyama 1973). Simultaneously, the lower-tropospheric circulation is also forced by condensational heating in the baiu rainband (Lu and Lin 2009; Sampe and Xie 2010). These two processes both link the lower-tropospheric circulation with the baiu convection. The end of the baiu season is characterized by the abrupt northward migration of the upper-tropospheric jet and the seasonal northward advance of the WPSH (e.g., Ninomiya and Muraki 1986). Consequently, the WPSH has a close relationship with baiu rainband activity.
However, the mechanisms responsible for the WPSH intensification remain to be elucidated, because many previous studies have focused on the influence of the WPSH on the baiu rainband and few studies have discussed the WPSH response to the baiu convection. In the present study, we attempt to understand that baiu rainband activity plays an important role in recent intensification in the WPSH, based on data analysis and model experiments. The remainder of this paper is organized as follows: Section 2 describes the data and model experiments. Section 3 presents observational diagnosis, and section 4 examines atmospheric response to baiu heating using a regional climate model. Section 5 is a summary and discusses broad implications.
2. Data and model
We mainly used data from the Interim European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA-Interim; Dee et al. 2011), but we also used data from the Global Precipitation Climatology Project (Huffman et al. 1997) and the International Satellite Cloud Climatology Project (ISCCP; Rossow and Schiffer 1991). Sea level pressure data at observational stations were obtained from the Japan Meteorological Agency. Our analysis focuses on long-term changes over the past three decades (1980–2012). Following a convectional approach, we analyzed the data based on the linear trend, and significance is based on a Student’s t test.
The model used in this study is the Weather Research and Forecasting Model (WRF), version 3.2.1, with the Advanced Research WRF dynamical core (Skamarock et al. 2008). Our model experiments followed Sato and Sugimoto’s (2013) methods under the same model setup. For example, the cloud scheme uses a cumulus parameterization (Grell and Devenyi 2002) and a cloud microphysics scheme (Hong et al. 2004) that adopts six categories of hydrometeors which allow representing mixed-phase processes as well as ice-phase processes. The model domain focuses on the western North Pacific (see Fig. 7) with a horizontal resolution of 25 km (170 × 150 grid points) and 40 vertical layers. We used the 6-hourly data from ERA-Interim as the atmospheric lateral boundary condition. The daily mean Optimum Interpolation SST (OISST; Reynolds et al. 2007) on a 0.25° grid was used as the ocean surface boundary condition because it has similar horizontal resolution with the atmospheric model. Model integrations were conducted from 20 May through 31 August for the period 1982–2012 (CTRL run), since the daily mean OISST is available after 1982.
To elucidate the atmospheric response to baiu heating, we turned off latent heating from the cloud microphysics scheme within the entire model domain. This sensitivity experiment was conducted with the same model configuration as the CTRL run (NOMP run). Thus, the NOMP run can suppress baiu heating compared with the CTRL run. Indeed, the temperature difference between the CTRL and NOMP runs averaged for the period 1982–2012 shows a maximum in the upper troposphere (not shown), only reflecting a response to condensational heating. Consequently, the CTRL − NOMP difference can provide the atmospheric response derived from the baiu heating.
3. Observational analysis
We will first consider changes in the North Pacific subtropical high during the past three decades. Figure 1 shows the spatial pattern of linear trends in summer SLP, and in the 850- and 300-hPa geopotential heights, for the period 1980–2012. Summertime mean precipitation shows a contrasting west–east distribution over the North Pacific (Fig. 1a), forming a part of the East Asian summer monsoon over the western Pacific. Over the eastern Pacific, SLP and the 850- and 300-hPa geopotential heights increase over the center of action of the North Pacific subtropical high. This intensification of the eastern Pacific subtropical high (EPSH) can be explained by an increase in thermal contrast between the land and ocean (e.g., Miyasaka and Nakamura 2005), possibly resulting from recent La Niña–like pattern.
On the other hand, over the western Pacific, both SLP and 850-hPa geopotential height increase over the western ridge of the North Pacific subtropical high and also increase farther northeastward, along the zonally elongated baiu rainband (black contours in Fig. 1a). Both the increasing trend and variance in the 300-hPa geopotential height are significant over the Kamchatka Peninsula. Comparing the two highs in Fig. 1, the EPSH intensifies over its center of action throughout the entire troposphere, reflecting a nearly equivalent barotropic change, while the intensification in the WPSH exhibits a northwestward tilt with height. This indicates that the intensification mechanism of the North Pacific subtropical high differs between the western and eastern Pacific. The intensified EPSH is found in both observations and in a simulated warming climate (Li et al. 2012), but the intensification of the WPSH along the baiu rainband has not been identified in earlier studies.
In contrast, SLP decreases over the Sea of Okhotsk, which is located to the north of the baiu rainband, resulting in contrasting north–south changes across 40°N. This means that the quasi-stationary surface Okhotsk high (OH) (e.g., Tachibana et al. 2004) has weakened during the past three decades, although the upper-tropospheric geopotential height has an increasing trend (Fig. 1c). Observational station data also show the contrasting north–south SLP trends between Nemuro and Chichi–Jima since the early 1980s (Fig. 2).
To examine the intensification of the WPSH, we focus on July, since baiu rainband activity over the western North Pacific is most active in midsummer. Over the western Pacific, lower-tropospheric southwesterlies and specific humidity significantly increase along the baiu rainband (Fig. 3a), suggesting an enhancement of baiu rainband activity. Indeed, the occurrence rate of deep convection (i.e., cumulus convection) for the period 1983–2009 (when ISCCP satellite cloud estimates are available) increases northeastward along the region of increased specific humidity (Fig. 3b). In contrast, to the south of the enhanced baiu rainband, lower-level specific humidity decreases and northeasterlies increase, forming an intensified anticyclonic circulation.
We also show the corresponding changes in diabatic heating to clarify the influence of baiu convective heating. Figure 3c shows the trends in diabatic heating at 500 hPa based on a diagnostic heat budget analysis (e.g., Matsumura and Yamazaki 2012). Stronger diabatic heating occurs northeastward from Japan along the north side of the peak of the mean diabatic heating, which appears to correspond with intensified lower-level cyclonic circulation to the north of the baiu rainband (Fig. 3a). Furthermore, the trend in diabatic cooling is significant to the south of the enhanced diabatic heating, corresponding to the decrease in specific humidity. These results indicate that baiu convective activity has moved northward and enhanced over the past three decades.
Significant trends in diabatic heating and cumulus convection occur also over the subtropical Pacific around 160°–170°E (Figs. 3b,c), indicating an enhancement of subtropical convection (Ueda et al. 1995). Figure 3d shows trends in divergence and ageostrophic wind at 950 hPa. Stronger convergences occur along the baiu rainband and subtropical convection, while along 30°–35°N strong divergence occurs as a result of the enhanced diabatic cooling. The enhanced diabatic cooling contributes to the lower-level subtropical convergence by inducing lower-level divergent ageostrophic northerlies, likely forming enhanced subtropical convection over the warm ocean. These results suggest that the intensified surface WPSH contributes to the enhanced subtropical convection.
We now analyze how the enhanced and northward-moved baiu convection leads to the intensification of the WPSH. Figure 4 shows meridional sections of linear trends in July averaged over the enhanced baiu rainband, at 140°–170°E. Lower-level specific humidity increases over 40°–45°N (Fig. 4a), indicating that lower-level southwesterlies supplies the moisture over the baiu rainband (Fig. 3a), where stronger ascent occurs in conjunction with lower-level convergence and upper-tropospheric divergence (Fig. 4b). Consequently, the lower-level convergence associated with the upward motion also supplies much of the moisture to the upper troposphere. To the south of the enhanced baiu rainband, lower-level specific humidity decreases with the intensified downward motion. Over the baiu rainband, stronger diabatic heating occurs, especially in the middle troposphere, which also increases lower-level southwesterlies to supply the moisture over the baiu rainband (Sampe and Xie 2010). Furthermore, the diabatic cooling trend and enhanced subtropical convection are also significant in the lower and middle troposphere, respectively. Another region of significant diabatic cooling at 200 hPa extends over the peak of mean diabatic heating, suggesting the rise in tropopause height as a result of the enhanced baiu convective activity.
The northward-moved and enhanced condensational heating accounts for the mid and upper-tropospheric warming trends over the baiu rainband (Fig. 4c), while the lower-tropospheric warming can be explained by southwesterlies-induced warm advection (not shown). As a result of the baiu heating, the upper-tropospheric ridge intensifies over the baiu rainband and the lower-level geopotential height decreases north of the baiu rainband (Fig. 4d), over the Sea of Okhotsk (Fig. 1a), although there is no significance in the lower troposphere. Consequently, to the south of the baiu heating, enhanced diabatic cooling associated with the downward motion accounts for the intensification of the lower-level ridge. The intensification in the WPSH exhibits a northward tilt with height, corresponding to the surface WPSH over 30°–35°N (Fig. 1a). Over northern latitudes 60°–65°N, on the other hand, adiabatic heating associated with the downward motion contributes to the tropospheric warming trends (Fig. 4c), forming a strong upper-tropospheric ridge (Figs. 1c and 4d).
The baroclinic atmospheric response to baiu convection is most active in July, while in August, when the baiu rainband disappears, the baroclinic forcing weakens, but atmospheric circulation changes are still significant. Figure 5 shows linear trends in August corresponded to July trends. Although the baiu forcing weakens, enhanced diabatic heating over 40°–45°N contributes to intensify the upper-tropospheric ridge, resulting in the baroclinic intensification of the WPSH, similar to July atmospheric circulation changes. The upper-tropospheric ridge in August becomes weaker than that in July, whereas in the lower-tropospheric ridge it is stronger, suggesting a transition from a baroclinic to a nearly barotropic-like change. Over the northern latitudes of 60°–65°N, on the other hand, enhanced diabatic heating contributes to the upper-tropospheric warming trends, likely resulting from land surface warming over eastern Siberia (Matsumura and Yamazaki 2012). Since baiu rainband activity over the western North Pacific is most active in midsummer, July trends well capture the summer [June–August (JJA)] western North Pacific climate changes (Fig. 6), although June trends are not robust. Therefore, baiu convective activity in midsummer can act as a major driver for the WPSH changes. A possible mechanism for the baiu convective activity change will be discussed in section 5.
4. Model experiments
As noted earlier, since the WPSH is interactively linked to baiu rainband activity through the lower-tropospheric southwesterlies, it might be difficult to clearly divide the cause and result based on only the observational analysis. To further verify the causality between the WPSH and baiu convection in long-term changes, we perform long-term numerical experiments using a regional climate model, which can better simulate baiu convection and rainband, compared with global climate models (Ninomiya 2011). To focus on the atmospheric response to only baiu convection, our model domain excludes the subtropical convective forcing (e.g., Figs. 3c and 4b).
Figure 7a shows summer SLP trends in the CTRL run for the period 1982–2012. The CTRL run well replicates the ERA-Interim trends (Fig. 1a), which show the north–south contrasting changes between the OH and WPSH, even if we analyze it using the same period as the CTRL run. The CTRL − NOMP difference also captures the north–south contrasting SLP changes, although the intensified surface WPSH is relatively weak and the weakened OH moves slightly eastward (Fig. 7b). Figures 7c,d show trends in July cloud water mixing ratio, which is the sum of cloud water, rainwater, ice, snow, and graupel, in the CTRL run and CTRL − NOMP difference. Both the CTRL run and CTRL − NOMP difference well reproduce the northward-moved and enhanced baiu cloud in July over the western North Pacific, suggesting that baiu heating increases the baiu convective cloud. The lower-level southwesterlies also increase over the western North Pacific in both the CTRL run and CTRL − NOMP difference. In addition, the lower-tropospheric circulation changes in July well capture the JJA SLP changes (Figs. 7a,b), corresponding to the ERA-Interim trends. In the CTRL − NOMP difference, however, the increased baiu cloud also occurs over the climatological baiu rainband to the south of 40°N, because we suppressed baiu heating within the entire model domain. As a result, the increased cloud to the south of 40°N might be overestimated, while the decreased cloud underestimated.
Figure 8 shows meridional sections of linear trends in July averaged over the enhanced baiu rainband over 155°–170°E. In the CTRL run (Fig. 8a), low clouds significantly increase over 40°–45°N and the lower-level convergence promotes the upward motion, enhancing the baiu convection. Since the western North Pacific SST around 40°–45°N has a strong northward-decreasing gradient, the lower-level southerlies suppress surface evaporation, causing low cloud or sea fog (Tokinaga et al. 2009). Indeed, OISST used in our experiments significantly increases in the western North Pacific (see Fig. 9a) and the occurrence rate of ISCCP low cloud also increases along the enhanced baiu convection (not shown). To the south of the enhanced baiu rainband, lower- to upper-tropospheric clouds decrease with the intensified downward motion. These baroclinic atmospheric circulation changes are consistent with the ERA-Interim trends (Fig. 4b). Similarly, the CTRL − NOMP difference also enhances the baroclinic atmospheric circulation change (Fig. 8b), indicating a response to baiu heating. In particular, the middle and upper baiu clouds significantly increase because of no condensational heating from the cloud microphysics.
The intensification in the WPSH in the CTRL run exhibits a northward tilt with height (Fig. 8c), corresponding to the surface WPSH over 30°–35°N (Fig. 7a). The upper-tropospheric ridge intensifies over the enhanced baiu convection and the lower-level geopotential height decreases, resulting in contrasting north–south changes between the OH and WPSH across the baiu convection. In the CTRL − NOMP difference (Fig. 8d), lower-level geopotential height significantly decreases under the enhanced baiu convection, which accounts for a weakening in the surface OH. The upper-level geopotential height appears to strongly reflect a baroclinic response to condensational heating–like tropical region. However, the upper-tropospheric ridge intensifies to the south of the enhanced baiu convection, indicative of the baroclinic intensification in the WPSH. The weaker response in the lower-level WPSH is most likely caused by competition with the significantly intensified lower-level cyclonic circulation under the enhanced baiu convection. Although we performed the NOMP run by turning off latent heating from the cloud microphysics within the entire model domain, the baroclinic atmospheric circulation changes are consistent with results of the response to observed diabatic heating along the baiu rainband from a linear baroclinic model (Lu and Lin 2009; Sampe and Xie 2010).
5. Discussion and conclusions
We have examined the intensification mechanism of the WPSH over the past three decades. Synthesis from comprehensive data analysis and model experiments indicates that the north–south contrasting changes between the OH and WPSH reflect a response to baiu heating. The northward-moved and enhanced baiu heating intensifies the upper-tropospheric ridge, resulting in the baroclinic intensification of the WPSH along the baiu rainband. Therefore, baiu convective activity in midsummer can act as a major driver for the summer western North Pacific climate changes.
The region of the WPSH found in this study appears to be different from the conventional definition (e.g., Wang and Fan 1999; Sui et al. 2007; Kosaka et al. 2012), which is located to the west edge of the North Pacific subtropical high rather than its north edge. Our study focuses on long-term changes in the WPSH, not its interannual variability, which has been mainly discussed under the conventional definition. Thus, the region of WPSH in long-term changes could be different from that in the interannual variability, since there are many factors that affect the WPSH in the interannual variability. At least, because the WPSH intensification is located within the range of the North Pacific subtropical high (Fig. 1b), we discussed the WPSH found in this study as the “western Pacific subtropical high.” Indeed, it appears that southwestward-extended ridge over the western North Pacific in the 2000s intensifies northwestward, corresponding to the region of increased SLP (not shown). This suggests that the WPSH may extend northwestward during the recent decades. Further research is needed to better understand the WPSH difference between the long-term changes and interannual variability.
Our results suggest that baiu convection has moved northward and enhanced over the western North Pacific, resulting in the intensification of the WPSH. What induces the baiu rainfall trend? Over the tropical Indian Ocean, SST warming excites a warm Kelvin wave in tropospheric temperature, resulting in the western North Pacific climate anomalies such as the WPSH (e.g., Xie et al. 2010; Huang et al. 2010; Chowdary et al. 2011). The tropical Indian Ocean SST anomalies also act as the anchor of the PJ pattern via the capacitor effect (Xie et al. 2009). Since the 1980s, the tropical Indian Ocean SST forcing of the PJ pattern has strengthened, resulting in an increase in El Niño–Southern Oscillation influences on the baiu rainfall over East Asia (Xie et al. 2010). It is possible that the PJ pattern induced by subtropical convection (e.g., Figs. 3c and 4b) has enhanced the baiu rainband, resulting in the intensification of the WPSH, which also enhances the baiu rainband through the lower-tropospheric southwesterlies. However, baiu convective variability also plays a crucial role in maintaining the PJ pattern (Lu and Lin 2009). Although the present study focuses on only baiu forcing, the baiu–PJ interaction may further amplify the summer western North Pacific climate changes.
However, the northward-shifted baiu convection over the western North Pacific could not be explained by only the PJ pattern. In the summer western North Pacific with a strong northward-decreasing SST gradient (i.e., the Kuroshio and Oyashio Extension region), increased SST can shift the subtropical jet northward by modifying the near-surface baroclinic atmosphere (e.g., Nakamura and Miyama 2014). Indeed, OISST used in our experiments has significantly increased in the Kuroshio and Oyashio Extension region (Fig. 9a), and the CTRL run in July clearly shows the northward-shifted subtropical jet (Fig. 9b). As a result, baiu convection is also expected to shift northward along the subtropical jet. In addition, baiu heating also contributes to accelerate the subtropical jet (Fig. 9c), which is consistent with the results from a linear baroclinic model (Lu and Lin 2009; Sampe and Xie 2010). Although in climatology the baiu rainband is generally active over East Asia, baiu convection has enhanced only over the western North Pacific to the east of 140°E (Fig. 3c), implying that the increased western North Pacific SST may contribute to the northward shift of the baiu rainband. Both the PJ pattern and western North Pacific SST will help to better understand the WPSH intensification associated with the baiu rainband.
We thank three anonymous reviewers for providing useful comments for the improvement of this paper. We acknowledge support from the Environment Research and Technology Development Fund [S-8-1(2)] of the Ministry of the Environment and the Research Program on Climate Change Adaptation (RECCA) of the MEXT, Japan.