Effect of SST in the Northwest Indian Ocean on Synoptic Eddies over the South China Sea-Philippine Sea in June

Abstract

Synoptic eddies (with a period of two to eight days) are active in the South China Sea-Philippine Sea (SCS-PS) and control weather variations. In addition, the intensity and frequency of synoptic eddies may change along with variations in sea surface temperatures (SST). This paper presented the influence of SST in the northwest Indian Ocean on synoptic eddies in the lower troposphere over the SCS-PS in June. Our statistical analysis showed a significant negative correlation between the SST in the northwest Indian Ocean and the synoptic scale eddy kinetic energy (EKE) in the SCS-PS. By analyzing the EKE budget of synoptic eddies, we found that the variation in the synoptic scale EKE over the SCS-PS is mainly due to the change in the monthly zonal wind gradient, which affects the barotropic energy conversion between the monthly mean flow and the synoptic eddies. Additionally, the northwest Indian Ocean SST modulates the monthly flow over the SCS-PS by alternating the strength of the Walker circulation in the west Pacific and Indian Ocean. Finally, the influence of SST in the northwest Indian Ocean on EKE in the SCS-PS was reproduced using the simplified atmospheric general circulation model, SPEEDY.

1. Introduction

In the South China Sea-Philippine Sea, synoptic eddies regularly propagate in a northwestern direction from the western equatorial Pacific Ocean [1]. Synoptic eddies, a type of tropical disturbance, can trigger tropical cyclone genesis in the SCS-PS and its surrounding regions [2]. Li (2006) [3] used an atmospheric general circulation model to prove this evolution from tropical disturbances to tropical cyclones. Other studies have shown that around 30% of typhoons are associated with synoptic eddies [4,5]. Furthermore, Xu et al. (2014) [6], who investigated typhoon Manyi in 2001 using a sensitivity experiment, showed that a synoptic wave train was crucial for the generation of the typhoon. Additionally, Zhou et al. (2018) [7] showed that the phases of quasi-biweekly oscillation can affect the development of synoptic disturbances, which leads to the intraseasonal variation in the intensity of tropical cyclones over the western North Pacific Ocean.

In addition to tropical cyclone genesis and intensity, Liu et al. (2022) [8] pointed out that the multiscale interaction between synoptic eddies and low-frequency disturbances also causes extreme pre-summer precipitation over South China. The work of Hu et al. (2022) [9] also showed that the early and late onset of the South China Sea summer monsoon may alter synoptic eddy activity, which affects the frequency of extreme precipitation events in the Indochina Peninsula, Hainan Island, and the Philippines. In addition, synoptic weather patterns also modulate extreme temperature and the range of temperature on diurnal and synoptic timescales in southeast China [10,11,12]. Hence, synoptic eddies also play a role in controlling precipitation and temperature fluctuations in China.

The synoptic eddies in the lower troposphere over the South China Sea are subject to sea surface temperature variations in the equatorial eastern Pacific. Hsu et al. (2009) [13] showed that the El Niño/Southern Oscillation is associated with the southeastward extension of the monsoon trough and the intensification of the westerly jet over the tropical western Pacific, which is favorable for the baroclinic energy conversion of synoptic eddies. Additionally, Zeng et al. (2012) [14] proved the relation between the El Niño/Southern Oscillation and synoptic eddies in the South China Sea, based on mooring observations.

Previous studies have demonstrated that the SST in the Indian Ocean can influence atmospheric circulation in the South China Sea and its surrounding regions. Yuan et al. (2008) [15] found that SST anomalies in the tropical Indian Ocean influenced the onset of the South China Sea summer monsoon by affecting the Walker circulation and regulating convection in the western Pacific. A recent study pointed out that the increase in June precipitation in the South China Sea and decrease in June precipitation in the Indochina Peninsula are caused by warming in the tropical western Indian Ocean [16]. Hence, previous studies have reported the impact of the western Indian Ocean on atmospheric circulation in the South China Sea, on monthly and seasonal timescales. However, its impact on a synoptic timescale has still not been well investigated.

Lorenz proposed an energy cycle theory including the potential and kinetic energies of the mean flow and eddy components on a global scale [17]. In the energy cycle, energy is converted from the available potential energy of mean flow to eddy potential energy and then to eddy kinetic energy. Subsequently, Ulbrich and Speth (1991) [18] divided the eddy component into stationary and transient parts. The energy cycle has been utilized in past studies on atmospheric dynamics, including multiscale eddy interaction, eddy–mean flow interaction, and the influence of SST on synoptic eddies [7,19,20,21,22]. For instance, Hu et al. (2020) [20] demonstrated that synoptic eddies will be stronger (weaker) during the years with late (early) withdrawal date.

Furthermore, it has also been noted that the SST in the central equatorial Pacific is positively correlated with synoptic scale eddy kinetic energy over the northwest Pacific during boreal summer [21]. In short, the energy cycle has been widely utilized in the study of synoptic eddies. In this study, the energy cycle was applied to investigate the potential influence of the SST in the Indian Ocean on synoptic eddies in the SCS-PS, which remains unclear.

The main purpose of this paper was to explore and discuss the influence of the SST in the northwest Indian Ocean on synoptic eddies in the South China Sea-Philippine Sea using statistical analysis and numerical simulation. This paper is divided into four sections. Section 2 introduces the data, methods and numerical models employed in this study; Section 3 describes the underlying process and SST forcing, leading to anomalous synoptic eddies in the SCS-PS. Finally, the results are summarized in Section 4.

2. Data and Methods

Daily atmosphere parameters were obtained from the National Centers for Environmental Prediction (NCEP2; Kanamitsu et al., 2002 [23]), including geopotential height, air temperature, horizontal wind, and vertical velocity. The NCEP2 reanalysis contained 17 vertical levels extending from 1000 to 10 hPa, where each level had a horizontal resolution of 2.5° × 2.5°. The NOAA Extended Reconstruction Sea Surface Temperature version 5 (ERSSTv5; Huang et al., 2017 [24]) dataset, with a horizontal resolution of 2° × 2°, was employed to determine SST. The study period was from 1979 to 2020 (42 years).

Previous studies have isolated the synoptic signal by using 2–10-day [20,21], 3–8-day [1,25] or 2–8 day [26,27] bandpass filters. In this study, the synoptic signal (2–8 days) was obtained using the Lanczos bandpass filter [28]. There are several variables, such as wind speed, geopotential height, relative vorticity, air temperature, etc., that can demonstrate synoptic variation at different levels of troposphere [1,29]. In this study, the signal of horizontal wind and geopotential height were primarily considered to represent the intensity of synoptic eddies.

To calculate the synoptic scale EKE and diagnose its variation, two equations described by Hsu et al. (2009) [13] were employed, as shown below:

$$EKE=\frac{1}{2}{u^{\prime }}^2+\frac{1}{2}{v^{\prime }}^2,$$

(1)

$$\overline{\frac{\partial EKE}{\partial t}}=\underset{CK}{\underbrace{-\overline{\left(V^{\prime }·\left(V_3^{\prime }·{\nabla }_3\right)\overline{V}\right)}}}\underset{CE}{\underbrace{-\frac{R}{P}·\overline{T^{\prime }{\omega }^{\prime }}}}\underset{BG}{\underbrace{-\overline{{\nabla }_3·\left(V_3^{\prime }{\phi }^{\prime }\right)}}}\underset{BK}{\underbrace{-{\overline{V}}_3·{\nabla }_3\overline{EKE}-\overline{V_3^{\prime }·{\nabla }_{3}EKE}}}+D,$$

(2)

where overbar and prime represent the June mean and 2–8-day bandpass filtered values, respectively; $V$ is the horizontal velocity with zonal and meridional components $u$ and $v$, respectively; ${\nabla }_3$ denotes a three-dimensional gradient; $t$ is the time; $T$ is the air temperature; $P$ is the pressure; $\phi$ is the geopotential; $R$ is the gas constant of dry air; $\omega$ is the vertical velocity; and $D$ denotes the dissipation of EKE by frictional effects.

EKE is calculated using Equation (1). Equation (2) is the budget equation of EKE. On the right-hand side of Equation (2), CK represents barotropic energy conversion, BG is the boundary flux of geopotential, BK indicates the boundary contribution of EKE, and CE denotes baroclinic energy conversion. In addition, the term CK in Equation (2) is divided into six terms, as presented in Equation (3), which can be used to quantify the contributions of different eddy-mean flow components to the variation in barotropic energy conversion.

$$CK=\stackrel{C{K}_1}{\overbrace{-\overline{{u^{\prime }}^2\frac{\partial \overline{u}}{\partial x}}}}\stackrel{C{K}_2}{\overbrace{-\overline{{v^{\prime }}^2\frac{\partial \overline{v}}{\partial y}}}}\stackrel{C{K}_3}{ \overbrace{-\overline{u^{\prime }{v}^{\prime }\frac{\partial \overline{u}}{\partial y}}}}\stackrel{C{K}_4}{\overbrace{-\overline{u^{\prime }{v}^{\prime }\frac{\partial \overline{v}}{\partial x}}}}\stackrel{C{K}_5}{\overbrace{-\overline{u^{\prime }{\omega }^{\prime }\frac{\partial \overline{u}}{\partial P}}}} \stackrel{C{K}_6}{\overbrace{-\overline{v^{\prime }{\omega }^{\prime }\frac{\partial \overline{v}}{\partial P}}}},$$

(3)

In this study, we further divide CK₁ and CK₃ to Equations (4) and (5) for figuring out the contribution of perturbation anomaly and mean flow anomaly, respectively. “[]” represents climatological mean and “{}” represents anomaly.

$$C{K}_1=\underset{climatological mean}{\underbrace{-\overline{[{u^{\prime }}^2][\frac{\partial \overline{u}}{\partial x}]}}}\underset{ perturbation}{\underbrace{-\overline{\{{u^{\prime }}^2\}[\frac{\partial \overline{u}}{\partial x}]}}}\underset{ mean flow}{\underbrace{-\overline{[{u^{\prime }}^2]\{\frac{\partial \overline{u}}{\partial x}\}}}}\underset{nonliear}{\underbrace{-\overline{\{{u^{\prime }}^2\}\{\frac{\partial \overline{u}}{\partial x}\}}}}$$

(4)

$$C{K}_3=\underset{climatological mean}{\underbrace{-\overline{[u^{\prime }{v}^{\prime }][\frac{\partial \overline{u}}{\partial y}]}}}\underset{ perturbation}{\underbrace{-\overline{\{u^{\prime }{v}^{\prime }\}[\frac{\partial \overline{u}}{\partial y}]}}}\underset{ mean flow}{\underbrace{-\overline{[u^{\prime }{v}^{\prime }]\{\frac{\partial \overline{u}}{\partial y}\}}}}\underset{nonliear}{\underbrace{-\overline{\{u^{\prime }{v}^{\prime }\}\{\frac{\partial \overline{u}}{\partial y}\}}}}$$

(5)

In addition, several statistical analysis methods were used in this study. The dominant mode of EKE in the SCS-PS (0–35° N, 100–150° E) was identified using the empirical orthogonal function (EOF) analysis. In addition, spectral analysis was employed to investigate the frequency geopotential height at 850 hPa in the SCS-PS [30,31]. A Student’s t-test is employed to examine the difference between two-sample means. The results of the numerical simulations are examined using a paired z-test.

For numerical simulations, this study used a simplified atmospheric general circulation model—Simplified Parameterizations, privitivE-Equation DYnamics (SPEEDY; Molteni 2003; Kucharski et al., 2006, 2013 [32,33,34])—which was developed by the International Centre for Theoretical Physics to simulate anomalous SST forcing on EKE over the SCS-PS. SPEEDY uses semi-implicit processing for gravity waves and parameterizes large-scale physical processes. Thus, it can easily simulate global climate anomalies caused by anomalies in SST or other meteorological elements in a local area. As a global model, SPEEDY’s boundary conditions are mainly the boundary of the underlying surface, including SST, sea ice fraction, soil temperature in the deep soil layer (about 1 m), moisture in the top soil layer and the root-zone layer, snow depth, bare-surface albedo (in the absence of snow or sea ice), and the fraction of land-surface covered by vegetation. In this study, only the boundary condition of SST was modified, and the other conditions were set to default. In addition, the horizontal and vertical resolution were set to T30, and 8 levels, respectively.

Previous studies have shown that this model can reasonably simulate the atmospheric response to anomalous SST forcings [35,36,37,38,39]. In this study, the model was firstly driven with the climatological mean of global monthly SST for 165 years. Subsequently, each of the last 160 years was used as an initial condition for sensitive simulations.

3. Results

3.1. Link between Synoptic Eddies in the SCS-PS and Northwest Indian Ocean SST

The climatological mean of the monthly synoptic scale EKE at 850 hPa is presented in Figure 1a–c, which demonstrates an active region of synoptic eddies in the SCS-PS in June and July. It was noted that synoptic eddies were substantially weaker in May. In addition, as illustrated in Figure 1d, the EOF1 of May EKE showed weak signals in the SCS-PS. For the EOF1s of June and July EKE, a negative center was located in the SCS-PS, which implied a suppression of synoptic eddies (Figure 1e,f). The three EOF1s accounted for 37.6% to 46.9% of EKE variance and passed the 0.05 significance level of the North test [40]. The corresponding time series of the EOF1s are presented in Figure 1g–i.

The relation between the SST in the Indian Ocean and EKE in the SCS-PS was examined using the correlation coefficient between monthly SST and the time series of EOF1. For SCS-PS, a significant correlation was found, as shown in Figure 2a,b, which led to a result similar to that reported by Huangfu et al. (2022) [21]: local SST is connected to the intensity of synoptic eddies. In this study, we were more interested in the modulation of the Indian Ocean. In June (Figure 2a), a significant positive correlation was found in the northwest Indian Ocean, with a maximum center in the eastern Arabian Sea. In addition, the correlation of detrended values is shown in Figure 2b, which is similar to Figure 2a. This implies that the relation between the SST in the northwest Indian Ocean and EKE is significant in June, even when the trends are removed. On the other hand, for the correlation in July, no significant signal was noted in the northwest Indian Ocean (figure not shown). This suggests that the linkage between the SST in the northwest Indian Ocean and synoptic EKE in the SCS-PS occurs on a monthly timescale. Hereafter, we will focus solely on the linkage in June.

The northwest Indian Ocean SST index (NWIO) was used to measure the temporal variation in SST, which was obtained by standardizing and averaging SST over 5°–20° N and 55°–75° E (black box in Figure 2a,b). As shown in Figure 2c, a significant increasing trend was found in the NWIO. Thus, detrended variables were used for subsequent analyses. The detrended NWIO is shown in Figure 2d. Subsequently, a standard deviation of 0.82 standard deviation was chosen as the threshold to identify the 11 warmest and coldest Junes in the northwest Indian Ocean (Figure 2d), which were termed as Pos_year and Neg_year groups. The two groups were used to study anomalous atmospheric circulation in association with SST anomalies in the northwest Indian Ocean.

The difference in geopotential height between Pos_year and Neg_year (Pos_year–Neg_year) is presented in Figure 3a; a significant positive difference was found in the West Pacific. This implies a southwestward extension of the western north Pacific subtropical high. On the other hand, as presented in Figure 3b, the difference in the standard deviation of synoptic geopotential height at 850 hPa between Pos_year and Neg_year manifested a negative center in the SCS-PS. This suggests that synoptic eddies in the SCS-PS are suppressed along with warm northwest Indian Ocean SST. In addition, spectral analysis was applied to the daily geopotential height at 850 hPa over 2°–15° N,115°–140° E (black box in Figure 3a,b), for Pos_year and Neg_year. As shown in Figure 3c,d, the magnitude of the 5-day variation was notably weaker in Pos_year than in Neg_year.

3.2. Change in the Energy Conversion of EKE over the SCS-PS in Response to Anomalous Northwest Indian Ocean SST

According to Equation (2), we calculated the difference in the terms $CK$, $BK$, $BG$, and $CE$ between Pos_year and Neg_year, as illustrated in Figure 4. $CK$ demonstrated a spatial pattern that was consistent with EOF1, with a negative difference in synoptic scale eddy kinetic energy in the SCS-PS. In contrast, BK showed a positive difference in the SCS-PS. For $BG$ and $CE$, the difference in synoptic kinetic energy was not significant in the SCS-PS. Therefore, EOF1 is likely to be controlled by the barotropic energy conversion (term $CK$). In addition, this also suggests a that a warm northwest Indian Ocean SST (Pos_year) is concurrent with the suppression of barotropic energy conversion.

In order to clarify the cause of the difference in barotropic energy conversion between Pos_year and Neg_year, the term was divided into subterms, according to Equation (3), as shown in Figure 5. It was found that the subterms $C{K}_1$ and $C{K}_3$ dominated the difference in barotropic energy conversion between Pos_year and Neg_year. The contribution of the remaining subterms was negligible (not shown). In addition, as presented in Figure 5a, a significant negative difference in $C{K}_1$ was noted at the upstream region (around 5° N and 145° E), which could play a role in suppressing the intensification of synoptic eddies during its development stage. This could weaken their intensification afterward, because of the smaller magnitude of $u^{\prime }$ and $v^{\prime }$.

In Equation (3), the subterms $C{K}_1$ and $C{K}_3$ are subject to the variation in the perturbation component (${u^{\prime }}^2$ and $u^{\prime }{v}^{\prime }$) and June mean flow component ($\frac{\partial \overline{u}}{\partial x}$ and $\frac{\partial \overline{u}}{\partial y}$). To quantify the importance of these two components for $C{K}_1$ and $C{K}_3$, we replace each component with its climatological mean in each calculation, which can eliminate its contributions to the difference in subterms $C{K}_1$ and $C{K}_3$. For instance, $C{K}_1 w/o perturbation$ was calculated with the climatological mean ${u^{\prime }}^2$. Therefore, four combinations were obtained (Figure 6). For $C{K}_1$ (Figure 6a,b), the negative center in the upstream was mainly due to the meridional zonal wind gradient ($\frac{\partial \overline{u}}{\partial x}$). In contrast, the perturbation component (${u^{\prime }}^2$) was more important in the SCS-PS. For $C{K}_3$ (Figure 6c,d), both the perturbation and mean flow components are also essential to the difference in barotropic energy conversion between Pos_year and Neg_year.

3.3. Atmospheric Circulation Anomaly Driven by Anomalous Northwest Indian Ocean SST

As mentioned above, the change in mean flow can influence the barotropic energy conversion, which alters the intensity of synoptic eddies in the SCS-PS. The difference in horizontal wind at 850 hPa and the zonal and meridional gradients of zonal wind between Pos_year and Neg_year is presented in Figure 7. We observed an anomalous anticyclone flow over the SCS-PS along with an anomalous zonal wind gradient. At the southern part of the anomalous flow (Figure 7a), an easterly anomaly led to positive difference in the zonal gradient of zonal wind at the upstream region of the SCS-PS (around 5° N and 145° E). In addition, the anomalous anticyclone corresponded to the positive difference in the meridional gradient of the zonal wind in the SCS-PS. Thus, the variation in zonal wind affects the barotropic energy conversion between synoptic eddies and mean flow.

Figure 8 depicts the difference in atmospheric circulation over the tropical western Indian Ocean and the western Pacific Ocean between Pos_year and Neg_year. The difference showed upper-level divergence and a rising motion over the northwest Indian Ocean, and convergence and a sinking motion in the SCS-PS (Figure 8a,b). In addition, the easterly wind in the lower troposphere formed a closed loop of the anomalous overturning circulation (Figure 8c). Therefore, the Walker circulation was suppressed, concurrent with the warm northwest Indian Ocean SST.

SPEEDY was utilized to examine the forcing of a warm northwest Indian Ocean SST on Walker circulation and zonal wind in the SCS-PS. June SST anomalies over northwest Indian Ocean (black box in Figure 2c) were added to drive the warm and cold northwest Indian Ocean simulations. The difference between the two simulations (warm—cold northwest Indian Ocean simulations) demonstrated the suppression of EKE around the Philippine Sea (Figure 9a) and an anomalous anticyclonic flow in the SCS-PS (Figure 9b,c). For the difference in the zonal gradient of zonal wind (Figure 9b), a positive center was noted upstream of the SCS-PS. Additionally, for the difference in the meridional gradient of zonal wind (Figure 9c), a positive center was located in the SCS-PS. Thus, the results of the numerical simulations were generally consistent with the difference between Pos_year and Neg_year, as presented in Figure 7a,b. This implies that a warm northwest Indian Ocean SST can induce anomalous anticyclonic flow in the SCS-PS, which suppresses the barotropic energy conversion and synoptic scale EKE.

Figure 10 shows the differences in atmospheric circulation across the tropical western Indian Ocean and the western Pacific. In the upper troposphere, there was divergence and rising motion at the Arabian Sea area, while convergence and sinking motion were observed around the SCS-PS (Figure 10a,b). There was also an obvious difference in easterly wind from the Philippine Sea to the Arabian Sea, which suggests a weakened overturning circulation (Figure 10c). A positive geopotential height difference also appeared at 850 hPa over the SCS-PS, which corresponded to a center of anomalous anticyclone flow. Therefore, the numerical model was able to simulate the influence of northwest Indian Ocean SST on Walker circulation.

4. Discussion

The results of this study suggest that the SST in the northwest Indian Ocean modulates the intensity of synoptic eddies in the South China Sea-Philippine Sea (SCS-PS) in June. The possible mechanism is summarized in a schematic diagram presented in Figure 11. Northwest Indian Ocean SST warming (cooling) may induce anomalously warm (cold) near-surface temperatures and anomalous rising motion in the western Indian Ocean, which weakens (enhances) the Walker circulation in the tropical western Indian Ocean and the western Pacific. Thus, an anomalous anticyclonic (cyclonic) flow is induced in the SCS-PS. The change in June mean flow decreases (increases) the barotropic energy conversion between synoptic eddies and mean flow and subsequently weakens (intensifies) the synoptic eddies. The results of the numerical simulations also support this conclusion.

In the numerical simulation, it was noted that the anomaly of EKE determined using SPEEDY was much weaker than that determined via observation. This could possibly be due to the bias of June mean flow between the model and real atmosphere. In addition, the contribution of other ocean basins, inter-ocean basin interactions, and atmospheric intraseasonal oscillation also influence June mean flow and synoptic eddies in the SCS-PS [2,10,13]. Additionally, the bias in the simulation could be also due to excluding local SST in the SCS-PS. As presented in Huangfu et al. (2022) [21] and Figure 2a,b in this study, a significant relation is found between synoptic-scale eddies and local SST in the SCS-PS. Hence, future studies are needed to clarify the relative contributions of the northwest Indian Ocean, other ocean basins and local SST to synoptic eddies in the SCS-PS.

Additionally, the trend of global warming has been shown to be reflected in the state of the ocean and atmosphere. Bader and Latif (2003) [41] pointed out that the equatorial Indian Ocean SST has shown a significant upward trend since 1950. The work of Sharma et al. (2022) [42] reported that with the gradual increase in SST in the northwest Indian Ocean, the Walker circulation in the Indian Ocean is gradually weakening and may disappear. However, long-term changes in the intensity of synoptic eddies in the SCS-PS and the contribution of SST in the northwest Indian Ocean to these future changes remain unclear.