This study emphasizes the importance of sea surface temperature (SST) over the tropical western Pacific and the ocean–atmosphere coupling in the extratropical Pacific for the climate in East and Southeast Asia. Specifically, it demonstrates that the anomalies of tropical SST explain many features of the climate variability in those regions during the summers of 1993 and 1994.
Very different atmospheric circulation patterns appeared in East and Southeast Asia between 1993 and 1994. Many regions including northern China, Korea, and Japan suffered from extremely high temperatures and severe droughts in the summer of 1994 but experienced reverse climate anomalies in the summer of 1993. To the south of these regions, the opposite climate patterns occurred. These climate features do not really resemble those associated with the El Niño–Southern Oscillation, which usually exerts a moderate impact on the East Asian climate. However, different SST anomalies have been observed in the tropical western and extratropical Pacific in the spring and summer between these two years. The authors carried out a series of simulations using an atmospheric circulation model and a slab oceanic model to understand the influences of these SST anomalies on the climate features.
Both the uncoupled atmospheric and coupled oceanic–atmospheric experiments indicate that the tropical western Pacific SST affects the East and Southeast Asia climate significantly. Warming in the tropical western Pacific produces hot, dry conditions in northern China, Korea, and Japan, and opposite climate signals to their south. These climate anomalies produced by the local SST resemble the observed climate difference between the summers of 1994 and 1993 when positive and negative SST anomalies, respectively, existed in the tropical western Pacific.
The coupled experiment also shows that the changes in extratropical atmospheric circulation caused by tropical SST anomalies generate changes in the extratropical Pacific SST, which, in turn, reinforces the climate signals produced by the tropical SST.
On the other hand, the uncoupled experiments forced by the extratropical Pacific SST anomalies show that the extratropical SST exerts an insignificant impact on the East and Southeast Asian climate. The change in this SST between 1994 and 1993 generates unrealistic climate patterns in some East Asian regions, accompanying an unnatural shift of the atmospheric circulation.
The variability of the summer climate in East and Southeast Asia is usually characterized by very complex features. The multiple-cell structure of the regional atmospheric circulation, in contrast to the relatively simpler single-cell monsoon overturning over South Asia, has been attributed to these complex features (e.g., Lau et al. 2000). Indeed, the locations and the shapes of these cells vary significantly from one year to another, and any shift and deformation of them will cause remarkable changes in the local temperature and precipitation patterns. Many difficulties exist in modeling and predicting these changes in this region, compared to many other regions of the world (Sperber et al. 1994; Slingo et al. 1996; Liang et al. 1997; Liang and Wang 1998; Kusunoki et al. 2001; Kang et al. 2002). As discussed by Ding et al. (2002), the reasons for these difficulties may include the strong internal dynamics of the East Asian monsoon, the strong interaction between the Tropics and extratropics, and the influence of the subtropical western Pacific high. The complex land–sea distribution, especially in Southeast Asia, may also increase the difficulties partially.
Not surprisingly, the El Niño–Southern Oscillation (ENSO) phenomenon is an important, if not the most important, factor affecting the interannual variability of the global climate and has been considered a predominant factor in controlling the variability of the East Asian climate. Numerous studies have linked ENSO to the temperature and precipitation anomalies in East and Southeast Asia for both summer and winter seasons (e.g., Kurihara and Kawahara 1986; Ropelewski and Halpert 1987; Huang and Wu 1989; Lau 1992; Zhang et al. 1996, 1997; Compo et al. 1999). However, it is now understood that the impact of ENSO on the climate in East Asia is only moderate, compared to many other regions. Even in North America, which is affected by ENSO more significantly, ENSO can only explain a small portion of the total climate anomalies (e.g., Namias and Cayan 1984). Lau et al. (2000) have shown that the response of the East Asian monsoon to ENSO is generally weak. Yang et al. (2002) have even suggested that the role of ENSO in the East Asian climate has been overemphasized, especially for subtropical– extratropical regions. Furthermore, the relationship between ENSO and the Asian monsoon appears to have declined during the past decade (Kumar et al. 1999; Torrence and Webster 1999; Chang et al. 2001; Wang 2001, 2002).
Other factors besides ENSO that have been considered to affect the summer climate in East and Southeast Asia include the Tibetan Plateau (Murakami 1987; Park and Schubert 1997; Wu and Zhang 1998) and sea surface temperatures (SSTs) in the western Pacific and the Indian Ocean (e.g., Nitta 1987; Huang and Sun 1992; Chang et al. 2000a; Hu et al. 2003). Previous studies have also investigated the effects of large-scale climate phenomena such as the Arctic Oscillation (Gong and Ho 2003), the East Asian jet stream (Liang and Wang 1998), and the subtropical western Pacific high (Chang et al. 2000a; Gong and Ho 2002).
In spite of its importance for East and Southeast Asian climate, it is only since the 1980s that the effect of tropical western Pacific SST has been studied extensively. It has been widely accepted that the warm SST in the tropical western Pacific and its associated atmospheric heating generate a Rossby wave–like response in the atmosphere and affect the climate over the western Pacific and East Asia (Huang 1984; Kurihara and Tsuyuki 1987; Nitta 1987; Lau and Peng 1992). Evidence has been provided by both observational and modeling studies that anomalously warm SST causes significant changes in the atmospheric heating over the Philippine Sea and in the subtropical western Pacific high on intraseasonal and interannual time scales (Huang and Lu 1989; Chang et al. 2000a). These changes are in turn associated with variations in the meridional structure of the local atmosphere (Chang et al. 2000b; Lau et al. 2000) and anomalies in precipitation and surface temperature (Tsuyuki and Kurihara 1989; Huang and Sun 1992; Simmonds et al. 1996).
In this study, we investigate the impact of regional SSTs over the tropical western and extratropical Pacific on the East and Southeast Asian climate during the summers of 1993 and 1994. These SSTs may or may not vary in phase with the equatorial central-eastern Pacific SST, which is commonly used to monitor ENSO events. Specifically, we study the climate response to these “local” SSTs during these two particular summers when opposite climate features occurred in East Asia and different SST patterns appeared in the adjacent oceans. We analyze various datasets and conduct a number of ensemble experiments using an atmospheric general circulation model (GCM) and a slab-oceanic mixed layer model.
In section 2, we describe the datasets and models used for this study and outline the major features of various experiments using the models. In section 3, we discuss the major climate anomalies in East and Southeast Asia observed during the summers of 1993 and 1994, and review previous studies that have been conducted in order to understand these climate anomalies. We present the main results from the current study in section 4, and summarize the results in section 5.
2. Data, models, and experiments
The data used in this study include the National Oceanic and Atmospheric Administration (NOAA) reconstructed monthly SST (Smith et al. 1996), surface air temperature from the National Aeronautics and Space Administration (NASA) Goddard Institute for Space Studies analysis (GISS_Ts; Hansen et al. 1999), the NOAA Climate Prediction Center Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997), and winds and geopotential height from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCER–NCAR) reanalysis (Kalnay et al. 1996). Most of these datasets have been used widely and detailed descriptions are available in many previous studies. The data source of the GISS_Ts analysis is a global historical climatology network. The analysis also includes results for a global temperature index, constructed by combining the meteorological station measurements over land with SST obtained primarily from satellite measurements.
1) Atmospheric GCM
This study applies the Seoul National University (SNU) atmospheric GCM, which has been modified based on the original Center Climate Systems Research/ National Institute for Environmental Studies (CCSR/ NIES) model of the Tokyo University (Numaguti et al. 1995). The GCM is a spectral model, with a triangular truncation at a two-dimensional wavenumber of 31. It is based on three-dimensional primitive equations, with 20 sigma levels in the vertical and an approximately 3.75° × 3.75° spatial resolution in the horizontal. Model physics include the two-stream k-distribution radiation scheme (Nakajima and Tanaka 1986), the simplified Arakawa–Schubert cumulus convection scheme based on the relaxed Arakawa–Schubert scheme (Moorthi and Suarez 1992) with diffusion-type shallow convection, and the large-scale condensation scheme of Letreut and Li (1991). The GCM also uses the Bonan (1996) land surface model, the Holtslag and Boville (1993) nonlocal convection scheme for vertical diffusion, and the McFarland (1987) scheme for gravity wave drag. A more detailed description of the model can be found in Kim (1999).
The SNU GCM has been widely used in studies of the Asian monsoon (Shen et al. 1998), El Niño events (Lee et al. 2002), atmospheric low-frequency variability (Yamazaki and Shinya 1999), and atmospheric convective activities (Wang et al. 2000). It is one of the major atmospheric GCMs involved in the recent International Research Programme on Climate Variability and Predictability (CLIVAR) Monsoon GCM Intercomparison Project (Kang et al. 2002).
2) Slab oceanic model
The oceanic model used in this study is from the NCAR Community Climate Model version 3 (Kiehl et al. 1996). It is a thermodynamic slab-ocean and sea ice model with specified ocean heat transport, which varies with month, longitude, and latitude, and has been coupled with the SNU atmospheric GCM (Kim 1999) and NASA Goddard Space Flight Center atmospheric GCM (Ho et al. 1998) for a variety of studies. The model allows for the simplest interactive surface for the ocean and sea ice components of the climate system. The prognostic equation for ocean mixed layer temperature, T0, is
where ρ0 is the density of ocean water (1.026 × 103 kg m−3), C0 the heat capacity of ocean water (3.93 × 103 J kg−1K−1), and Q the ocean mixed layer heat flux (W m−2) simulating the deep-water heat exchange and ocean transport. In addition, h0 is the ocean mixed layer depth (m), which is set to a uniform depth of 40 m, and F is the net atmosphere-to-ocean heat flux (W m−2), which is defined in the absence of sea ice as
where FS is solar flux absorbed by the ocean, FL the longwave cooling flux at the ocean surface, SH the sensible heat flux, and LH the latent heat flux from the ocean to the atmosphere.
At model iteration n, one has Tn0, Fn, and Qn. The forecast value of T0 is then
where Δt is the model time step.
1) Global SST Simulations
We have carried out two experiments, GLOBAL_1 and GLOBAL_2, forced by globally observed SST. Each of these experiments had six ensemble members, which used the initial conditions of 1 January of 1991, 1992, 1993, 1994, 1995, and 1996, respectively. The only difference between the two experiments is in the time period of SST forcing. In GLOBAL_1, the model was forced by the SST of January–August 1993. In GLOBAL_2, the SST of January–August 1994 was used to force the model. Integrations of eight months (from January to August) were conducted for all simulations. We carried out these two experiments to provide a benchmark for the assessment of model performance.
Correspondingly, we have carried out another ensemble experiment, CLI_SST, in which the model was forced by the monthly climatology of observed global SST. Results from this experiment will be applied as a reference for the sensitivity experiments using local SST features.
A list of the above experiments using global SST is given in Table 1, which includes the ocean–atmosphere coupled experiment CLI_SST_ML that will be discussed shortly in this section.
2) Regional SST Simulations
As shown in Table 2, we have carried out several experiments using regional SSTs to understand the influence of tropical western and extratropical Pacific SST on the East and Southeast Asian climate. Like the global SST experiments described above, these regional SST experiments also consisted of six ensemble members, used the same initial conditions mentioned above, and were integrated from 1 January to 31 August. However, unlike experiment CLI_SST, the variations of SST in the extratropical and tropical western Pacific during 1993 and 1994 are included in these experiments.
Figure 1 illustrates the target domains for the various regional SST experiments. In particular, we focused on the influences of SST in the extratropical Pacific Ocean (EPO), the tropical western Pacific Ocean (TWPO), and both (TWEPO). In experiment TWPO, for example, we superimposed the SST difference between 1994 and 1993 onto the monthly SST climatology in the box shown in Fig. 1b. That is, the SST within the targeted box is the sum of the monthly SST climatology and the monthly SST difference between 1994 and 1993. Outside the box, the SST was set into the monthly climatological SST only, at each grid point.
3) Coupled ocean–atmosphere experiments
We also carried out two additional, 15-yr coupled experiments using the SNU atmospheric GCM and a slab-oceanic model: CLI_SST_ML and TWPO_ML (see Tables 1 and 2). For these experiments, we first carried out an atmospheric GCM experiment using the observed monthly climatological SST of 1980–94 to generate the climatological ocean heat flux. In experiment CLI_SST_ML, the climatological ocean heat flux is prescribed at all ocean grid points and SST is predicted by the coupled models. Experiment TWPO_ML is similar to CLI_SST_ML except that the prescribed SST difference between 1994 and 1993 is used in the tropical western Pacific within the box in Fig. 1d. To ensure that the coupled simulations are not affected by the model spinup process, we only analyze the results from the integration of the last 11 yr.
3. Observed features
In this section, we depict the observed climate features in East–Southeast Asia and over the Pacific Ocean during 1993 and 1994.
a. Precipitation, surface temperature, and atmospheric circulation
Very different climate features appeared in East and Southeast Asia in 1993 and 1994. As shown in Figs. 2a and 2b, in June–August (JJA) 1993, cool and wet conditions dominated in the midlatitudes including northern China, Korea, and Japan, but the opposite climate features appeared in southern China, the South China Sea, and much of Southeast Asia. During the summer of 1994, on the other hand, many Asian countries experienced generally opposite climate conditions: hot and dry in the midlatitudes, and cool and wet to the south (Figs. 2d and 2e). The most robust signals in both temperature and precipitation fields occurred in Japan and Korea. The atmospheric circulation patterns also differed significantly during the two summers (Figs. 2c and 2f). For example, cyclonic patterns appeared over East Asia and the extratropical Pacific in 1993, but they were replaced by anticyclonic patterns in 1994. The atmospheric circulation over eastern Russia and the South China Sea also differed between the two summers.
Since warm SST anomalies appeared in the Niño-3.4 region (5°N–5°S, 170°–120°W) during the summers of 1993 and 1994 (0.47° and 0.48°C, respectively), we compare the features shown in Fig. 2 with the typical climate features associated with ENSO. Figure 3 shows the composite anomalies of temperature, precipitation, and atmospheric circulation for the summers (see figure captions) when ENSO is in its warm phase (information online at http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensoyears.html). While Figs. 2a and 2d depicts major features in East Asia, Fig. 3a displays strong signals over the tropical central and extratropical Pacific Ocean. During El Niño, the temperature is generally lower than normal in the extratropical North Pacific, extending slightly into East Asia. However, the signal in East Asia is so weak that it cannot be compared with the much more robust features shown in Figs. 2a and 2d. Note also the apparent difference over Southeast Asia and the tropical Pacific between Fig. 3a and Figs. 2a and 2d. A comparison of the precipitation field indicates a similarity over the Indian Ocean and the western-central Pacific between Figs. 2 and 3. However, the El Niño–related signal in JJA precipitation is mixed and insignificant in East Asia, especially in Korea and the eastern China (Fig. 3b). This is in contrast to the more apparent, opposite-signed signal shown in Figs. 2b and 2e. The atmospheric circulation of JJA 1993 (Fig. 2c) is similar to the El Niño–related pattern (Fig. 3c) over the tropical central-eastern and extratropical Pacific. However, the features over most of the Asian continent, the South China Sea, and tropical western Pacific are very different between the two figures. More remarkably, the El Niño–related features (Fig. 3c) are quite different from those in 1994 (Fig. 2f) except for the winds over the tropical western Pacific.
In short, the climate signals observed during the summers of 1993 and 1994 are different from those associated with El Niño events. More importantly, many of the signals of 1994 tend to have an opposite sign to those of 1993. These features suggest that the differences in the temperature, precipitation, and atmospheric circulation over East Asia between the two summers are not largely linked to the impact, at least the direct impact, of ENSO. Thus, we conclude that ENSO was not a significant impact factor during these seasons.
b. SSTs in the western Pacific and adjacent oceans
The change in surface air temperature, depicted in Figs. 2a and 2d, implies different SST patterns between 1993 and 1994. Figure 4 shows the differences in monthly SST between the two years (1994 minus 1993), from March to August. Several features can be identified from the figure. First, in the springtime, the northern tropical oceans (except the Bay of Bengal) were warmer in 1994. The warmer SST in spring was accompanied by a stronger South Asian monsoon in summer, as measured by the Webster–Yang monsoon index [−0.1 in 1993 and 1.6 in 1994; Webster and Yang (1992)]. Second, the stronger monsoon drove the ocean and resulted in cooler SST over the South China Sea and adjacent oceans in the summer of 1994. Such premonsoon warming and postmonsoon cooling in the oceans are in agreement with the general discussion of Lau and Yang (1997) and Han and Webster (2002). Third, the warmer tropical western and extratropical Pacific SST in 1994 was separated by a relatively cooler band around 20°N associated with the heavier precipitation shown above.
c. Results from previous studies
A number of studies have attempted to describe and explain the climate anomalies in East Asia during the summers of 1993–94. For example, Lee et al. (1997) showed that the cool, wet conditions of 1993 were preceded by a southward outbreak of cold air associated with a wavelike polar jet into the Korean Peninsula. For the summer of 1994, Heo et al. (1997) attributed the hot, dry conditions to the anticyclonic circulation associated with a blocking high around Korea, which can be related to the intensification and westward expansion of the subtropical western Pacific high (Kim et al. 1998). Park and Schubert (1997) explained the 1994 climate anomalies by the changes in orographic forcing associated with zonal wind variations over the Tibetan Plateau. Other factors that have been used to explain the 1993–94 climate anomalies include the gradient of SST between the South China Sea and western Pacific east of the Philippines (Kawamura et al. 1998) and the vorticity flux forcing of transient activities (Geng et al. 2000). In addition, the climate anomalies of 1994 in southern Asia have been explained by the change in the tropical Indian Ocean (Behera et al. 1999).
4. Modeling results
In this section, we discuss the results from various model experiments with a focus on the influence of tropical western Pacific SST and the importance of ocean–atmosphere coupling.
a. Model performance: Experiments GLOBAL_1 and GLOBAL_2
Figure 5 shows the differences in the six-member ensemble means of JJA surface temperature, precipitation, 500-mb geopotential height, and 200-mb wind between experiments GLOBAL_2 and GLOBAL_1 (see Table 1). As explained in section 2c, in GLOBAL_1 (GLOBAL_2), the model was forced by the global SST observed in 1993 (1994). A comparison of Figs. 5a and 6a, which shows the differences in the corresponding observed fields between the summers of 1994 and 1993, indicates that the SNU GCM successfully captures many of the observed features in surface temperature including the higher temperature during 1994 in East Asia and over the latitude band of 25°–50°N, and the lower temperature north of the band and over South–Southeast Asia. In particular, a great similarity can be found in eastern China, Korea, and Japan, where the climate anomalies are the focus of this study.
The main features of precipitation captured by the model are the reduced rainfall in 1994 over the tropical Indian and Pacific Oceans and in the extratropical East Asia, and the increased rainfall between the two regions mainly in the subtropical latitudes of 10°–25°N (see Figs. 5b and 6b). Consistent with the common problem of climate modeling by GCMs, the simulations of precipitation are less realistic than those of temperature. The SNU GCM does not simulate successfully the change in rainfall over the East China Sea and southern Japan. According to Kang et al. (2002), none of the 10 GCMs that participated in the CLIVAR–Monsoon GCM Intercomparison Project is capable of reproducing the observed rainband related to the mei-yu, baiu, or changma.
Figures 5c and 6c show that the model reasonably simulates the changes in circulation patterns over East Asia and the adjacent regions. It realistically captures many circulation features over the subtropical and extratropical Asia and the western Pacific, including the areas that are of interest in this study. However, the model overemphasizes the changes in geopotential height over the western Philippine Sea and the South China Sea.
b. Role of extratropical Pacific SST: Experiment EPO
Before focusing on the climate impact of tropical western Pacific SST in the next subsection, we assess here the influence of extratropical Pacific SST. Figure 7 shows the differences of ensemble means in surface temperature, precipitation, 500-mb geopotential height, and 200-mb winds between experiments EPO (see Fig. 1b and Table 2) and CLI_SST (see Table 1). This figure should be compared with Fig. 5 in which results from experiments with global SST are shown. Between Figs. 5 and 7, large similarities exist in temperature simulations in East and South–Southeast Asia (see Figs. 5a and 7a). Similarities also appear in precipitation simulations especially in Japan, Korea, and the oceans east of China. These features imply that the extratropical SST anomalies play a certain role in accounting for some of the climate signals in East Asia and the western Pacific. However, extratropical SST anomalies by themselves fail to ensure successful simulations of tropical precipitation variations (see Figs. 5b and 7b). A southward shift of about 10° latitude in the atmospheric circulation pattern over subtropical and extratropical East Asia and the western Pacific is observed in the extratropical experiments (Fig. 7c), compared to the global experiments (Fig. 5c). In addition, the extratropical SST anomalies also produce a smaller change in temperature in eastern-central Asia.
Previous studies have demonstrated that, in general, extratropical SST is less important than tropical SST in influencing the overlying atmosphere. Some studies have even claimed that the effect of extratropical SST is insignificant, although this is still an issue of much debate (see Frankignoul 1985; Robinson 2000). That this aspect appeared in the uncoupled experiments is probably attributed to the lack of air–sea interaction, at least partially. Experiments that allow interaction between the atmosphere and oceans in the extratropics are believed to yield more realistic simulations. More discussions about this issue will be provided in section 3d.
c. Role of tropical western Pacific SST: Experiments TWPO and TWEPO
In this section, we discuss the results from experiment TWPO, in which the model is forced by SST anomalies in the tropical northwestern Pacific and the South China Sea. Figure 8 shows the differences of ensemble means in surface temperature, precipitation, 500-mb geopotential height, and 200-mb winds between experiments TWPO and CLI_SST. The patterns are similar to the global results (Fig. 5) in many aspects especially in the Tropics and subtropics. For example, the tropical western Pacific SST enables a successful simulation of the atmospheric circulation. Between Figs. 5 and 8, similar circulation patterns appear over the western Pacific and a large portion of Asia, in spite of the discrepancy over the Sea of Okhotsk and eastern Russia. Accordingly, the differences (between 1994 and 1993) in temperature and in precipitation are simulated relatively realistically. The differences include the higher temperature (in 1994) in extratropical Asia and the lower temperature in South–Southeast Asia and southeastern Russia. They also include the reduced precipitation (in 1994) north of 20°N and in the equatorial region, and the enhanced precipitation in between. These similar features in Figs. 5 and 8 suggest the importance of tropical western Pacific SST.
Two specific features that mark a significant improvement from experiment EPO (Fig. 7) to experiment TWPO should be emphasized. First, the atmospheric wave train over East Asia and the western Pacific (Fig. 8c) is much more similar to that shown in Fig. 5c. Second, the simulations of precipitation over the East China Sea, the Yellow Sea, and the adjacent regions are more realistic (Fig. 8b).
We now discuss the results from experiment TWEPO as shown in Fig. 9. We conduct this analysis to further demonstrate the importance of tropical western Pacific SST since in the experiment the model was forced by both tropical and extratropical SST anomalies as in experiments TWPO and EPO, respectively. Not surprisingly, the simulations of TWEPO (Fig. 9) are much more similar to the results of TWPO (Fig. 8) than to those of EPO (Fig. 7), especially in the Tropics and subtropics. This is consistent with the discussions in the last two subsections.
It is worth noting two additional features by comparing Figs. 7–9. First, the tropical western SST (Fig. 8) tends to shift the wave train pattern more eastward than the extratropical SST (Fig. 7) and the combined tropical and extratropical SST (Fig. 9). Second, the difference in precipitation over the northern East China Sea between TWPO and TWEPO suggests again the importance of the tropical western Pacific SST in simulating the East–Southeast Asian precipitation.
The uncoupled experiments with the SNU GCM as discussed above have indicated that the tropical western Pacific SST plays an important role in regulating the summer climate in East and Southeast Asian climate. It is reasonable to speculate that the climate response varies from year to year because of the changes in the location and magnitude of SST anomalies even though other influencing factors remain unchanged. The nature of climate response may also behave differently from one model to another. However, the results presented here are consistently within the context of previous studies about the general impact of tropical western Pacific SST on East and Southeast Asian climate (Huang 1984; Nitta 1987; Huang and Sun 1992; Simmonds et al. 1996; Kawamura et al. 1998). More importantly, the current analysis is aimed intentionally at the specific climate events in 1993–94 using a GCM.
d. Importance of ocean–atmosphere coupling: Experiment TWPO_ML
Here, we discuss the results from two coupled ocean– atmosphere experiments: TWPO_ML and CLI_SST_ML (see description and Tables 1 and 2 in section 2c). The design of the model experiments enables an assessment of the importance of extratropical air–sea interaction and tropical western Pacific SST.
Figure 10 shows the differences in land and sea surface temperature, precipitation, 200-mb winds, and 500-mb geopotential height between experiments TWPO_ML and CLI_SST_ML. In general, the coupled runs produce smaller-amplitude signals than the SST-forcing runs (cf. Fig. 10 with Figs. 8 and 9). The reason for this feature is that the atmosphere and the ocean interface with each other through heat exchange in the coupled experiments. At least two important features shown in the coupled experiment should be noticed. First, the anomalous anticyclone in the wave train pattern over East Asia is simulated reasonably well. Second, the change in precipitation appears to be realistic over the East China Sea, the Yellow Sea, and nearby ocean domains. These features suggest an apparent improvement from the uncoupled to coupled experiments. In the uncoupled EPO experiment, the higher SST in the extratropical Pacific leads to more regional evaporation, cloud, and precipitation. However, this process does not decrease the local SST because of the experimental design of the fixed SST and thus excessive precipitation is produced unrealistically. This argument of an atmospheric effect is helpful for explaining the decrease in SST in the East China Sea in the coupled experiment (cf. Figs. 9 and 10). However, the air–sea coupling also reduces the surface temperature in northern China unrealistically.
Figure 11 shows the differences in fluxes of surface shortwave radiation, surface longwave radiation, latent heat, and sensible heat between TWPO_ML and CLI_SST_ML. There is an obvious consistency between Figs. 10 and 11. The higher tropical western Pacific SST generates a wave train pattern with an anticyclonic circulation over eastern Asia (Fig. 10c), accompanied by less precipitation (Fig. 10b) and more downward shortwave radiation (Fig. 11a). As a result, the surface temperature increases (Fig. 10a), and evaporation and sensible heating also increase correspondingly (Figs. 11c,d). Relatively smaller changes occur in the surface longwave radiation fluxes (Fig. 11b). It should also be pointed out that since the interaction between the ocean and atmosphere is allowed in TWPO_ML, the coupled experiments generate more reasonable surface fluxes compared to the uncoupled experiments (figures not shown).
As shown in Fig. 12, the response of outgoing longwave radiation to the high tropical western Pacific SST in the coupled ocean–atmosphere system is characterized by a wave train over eastern and southeastern Asia and the western Pacific Ocean. The warm SST produces strong convection over the Philippine Sea. However, atmospheric convection is suppressed over eastern China, Korea, Japan, and the adjacent domains. This feature is consistent with the results of Huang (1984), Nitta (1987), Huang and Lu (1989), and Huang and Sun (1992).
Interestingly, Fig. 10a indicates that the warming in the extratropical Pacific may result from high SST in the tropical western Pacific, which plays an active role in extratropical ocean–atmosphere interaction (Peng and Whitaker 1999), as a response to the changes in atmospheric circulation as suggested by Park and Schubert (1997). According to Nitta (1987), a warming in the tropical western Pacific generates an anomalous high over Japan and the adjacent regions. The subsidence associated with this anomalous high increases the local (extratropical) SST naturally or in coupled ocean–atmosphere experiments such as the EPO in this study. The physical processes in which tropical SST leads to a response in the extratropical oceans have been discussed comprehensively by Lau and Nath (1996, 2001) and Alexander et al. (2002). As seen by comparing Figs. 7 and 10, the response in the extratropical Pacific to the tropical western Pacific SST in turn reinforces climate signals such as in the change of precipitation. These processes clearly demonstrate the importance of tropical western Pacific SST and extratropical Pacific air–sea interaction for explaining the climate signals in East and Southeast Asia.
5. Summary and discussion
Between the summers of 1993 and 1994, very different atmospheric circulation patterns were observed over East and Southeast Asia. Many regions including northern China, Korea, and Japan suffered from extremely high temperatures and severe droughts in 1994 but experienced the reverse climate anomalies in 1993. To the south of these regions, the opposite climate patterns appeared. Previous studies have explained these signals, or part of the signals, by the southward outbreak of cold air associated with the polar jet, the blocking circulation related to the intensification and westward expansion of the subtropical western Pacific high, and the changes in orographic forcing associated with zonal wind variations over the Tibetan Plateau. Others have attributed them to the effect of the SST gradient between the South China Sea and the western Pacific, and the vorticity flux forcing of transient activities.
Since very different SST anomalies were observed in the tropical western, and especially the extratropical, Pacific Ocean in the spring and summer seasons between 1993 and 1994, we were motivated to investigate the influence of these SST anomalies on the climate signals, although it is difficult to separate the relative contributions of all the factors whose influence has been claimed. We carried out a series of ensemble simulations using the SNU atmospheric GCM and a slab-oceanic model to accomplish our objectives.
It has been demonstrated that the variability of the tropical western Pacific SST explains many features of the East and Southeast Asian climate during the summers of 1993 and 1994. The experiment forced by tropical western Pacific SST anomalies produces changes in precipitation, temperature, and atmospheric circulation that are similar in many aspects to those observed. Warming in the ocean domain leads to hot and dry conditions in northern China, Korea, and Japan, and the opposite climate signals to the south of these regions. On the other hand, the influence of warm SST anomalies over the extratropical Pacific on the East and Southeast Asian climate during the summers of 1993–94 is much weaker. The extratropical SST experiment also generates hot–dry and cool–wet climate patterns although these patterns shift unrealistically southward compared to those generated by the warming in the tropical western Pacific.
Compared with the atmospheric model experiments forced by SST, coupled ocean–atmosphere experiments improve the results noticeably. The coupled experiments also explain the extratropical Pacific SST as a response to the changes in atmospheric circulation caused by the tropical western Pacific SST. However, the feedback of extratropical SST in the coupled experiments reinforces some of the climate signals caused by the tropical SST anomaly.
We have also examined the relationship between the tropical western Pacific SST and the East Asian climate for other summers. A relationship similar to that of 1994 (warm SST and dry–hot climate conditions) also appeared in 1990, 1995, 1996, and 2001. On the other hand, in 1983, a negative SST anomaly in the tropical western Pacific was accompanied by wet and cool conditions in East Asia, as occurred in 1993.
Results from this study demonstrate the influence of tropical western Pacific SST on East–Southeast Asian climate in the summers of 1993–94 and are in agreement with those of many previous studies that focus on the impact of the SST on the summer climate in China, Japan, Korea, and other countries (e.g., Huang and Lu 1989; Huang and Sun 1992; Simmonds et al. 1996; Kawamura et al. 1998; Chang et al. 2000a,b). Although there are only a few GCM studies targeting this issue for these two summers in particular, the simulations in the work of Kawamura et al. indicate that the warming in the tropical western Pacific east of the Philippines, with a cooling in the South China Sea, causes increased cumulous convection over the Philippines and increased geopotential height (thus surface temperature) over Japan. This situation is similar to the 1994 case that is discussed in this study. This influence of the tropical western Pacific SST can be explained by the physical mechanisms involved in the Rossby wave response as proposed by Huang (1984, 1985) and Nitta (1987). In particular, according to Nitta (1987; see his Fig. 18), warming in the tropical western Pacific causes low geopotential height anomalies over the South China Sea, southern China, and Southeast Asia, and high anomalies over northern China, the Koreas, Japan, and the adjacent oceans. These circulation anomalies are accompanied by a wet and cool climate in the former regions and dry and hot conditions in the latter. Correspondingly, cooling in the tropical western Pacific causes the opposite anomalies in the fields of atmospheric circulation, precipitation, and temperature, as simulated for the case of 1993 (results not shown).
The above-described relationship between the tropical western Pacific SST and the East–Southeast Asian climate is also consistent with the variability of the East Asian monsoon circulation. For example, the negative SST anomaly in the summer of 1993, when wet and cool conditions appeared in East Asia, is linked to a strong East Asian monsoon, as measured by the index of Lau et al. (2000). Similar cases can also be found in 1980 and 1982. On the other hand, the positive SST anomaly in the summer of 1994, when a dry and hot climate emerged in East Asia, is associated with a weak East Asian monsoon and similar conditions occurred in 1981, 1984, and 1990. Thus, the variability of the monsoon circulation as measured by the Lau et al. index is closely related to the wave trains generated by the tropical western Pacific SST anomalies.
This study emphasizes the importance of tropical western Pacific SST in affecting the summer climate in East and Southeast Asia. In the meantime, it does not exclude the influence of ENSO on the climate. Major ENSO events in Northern Hemisphere winters may not always be accompanied by consistent SST anomalies in the western Pacific and other ocean domains like the South China Sea in the following summer because of various reasons. However, ENSO may cause changes in the Eurasian land surface processes in winter, which provide additional memory for the following summer and thus affect the summer climate indirectly (Yang and Lau 1998). This indirect influence of ENSO on the East and Southeast Asian temperature and precipitation has not been discussed in this study.
This project was mainly supported by the BK21 Project of the Korean government. The authors thank John Janowiak of NOAA/Climate Prediction Center (CPC), Editor Siegfried Schubert, and the three anonymous reviewers whose constructive comments have improved the quality of the paper. They also thank Dr. Jong-Khun Kim of the Korea Meteorological Administration for assistance at the early stage of model experiments. Soo-Hyun Yoo is thankful to CPC, which hosted her visit when this work was being completed.
Visiting scientist: NOAA/Climate Prediction Center, Camp Springs, Maryland
Corresponding author address: Prof. Chang-Hoi Ho, School of Earth and Environmental Sciences, Seoul National University, Seoul 151-742, South Korea. Email: email@example.com