The South China Sea (SCS) is affected by two intraseasonal components in summer: the Madden–Julian oscillation (MJO) and the quasi-biweekly oscillation (QBWO). In the present study, the impacts of the MJO and QBWO on tropical cyclones (TCs) locally formed in the SCS (local TCs) in summer are investigated. The results show that both the MJO and QBWO can affect the genesis frequency, location, and motion of the local TCs. More TCs form in the convectively active phases of the MJO and QBWO in the northern SCS. With the northward propagation of the MJO and QBWO convective signals, the major TC genesis location also shifts northward. Since the western Pacific subtropical high shifts eastward (westward) when convection associated with the MJO and QBWO in the northern SCS is enhanced (suppressed), the steering flow in the major TC genesis region is favorable for the eastward (westward) movement of TCs. Results from the composite analysis of the steering flow indicate that both the MJO and QBWO play an important role in controlling the motion of the eastward-moving TCs.
The South China Sea (SCS) is one of the active regions of tropical cyclogenesis, where 13% of the northwestern Pacific (NWP) tropical cyclones (TCs) form (Wang and Fei 1987). The SCS is a unique semienclosed marginal sea; most TCs generated in the SCS (local TCs) have a relatively short time to intensify before they make landfall along the coasts in the surrounding countries, such as China, Vietnam, and the Philippines, and often induce severe casualties and economic losses (Zhang et al. 2009). Previous studies have also demonstrated that TCs can significantly affect the upper-ocean thermal structure and both upper- and deep-ocean circulations, as well as the ocean primary production in the SCS (Chu et al. 2000; Lin et al. 2003; Zhao et al. 2008; Wang et al. 2009; Ling et al. 2011; Wang et al. 2014).
Tropical cyclogenesis and development are controlled by several dynamical and thermodynamic factors, such as relative vorticity in the lower troposphere, vertical wind shear, relative humidity in the middle troposphere, and sea surface temperature (SST) (Gray 1979; Emanuel 2003; Wang et al. 2007; Camargo et al. 2007). Previous studies have revealed that the tropical cyclogenesis in the SCS can be affected by the SCS monsoon (Wang et al. 2007, 2012), mei-yu frontal systems (Lee et al. 2006), the NWP subtropical high (Ling et al. 2015), east Indian Ocean SST anomalies (Zhan et al. 2011), the SST gradient between the eastern Indian Ocean and the western Pacific (Li and Zhou 2014), El Niño-Southern Oscillation (ENSO; Chan 2000; Wang and Chan 2002; Zuki and Lupo 2008), and the Pacific decadal oscillation (PDO; Goh and Chan 2010). The TC genesis locations in the SCS show an obvious seasonality in response to the seasonal evolution of the SCS monsoon system (Wang et al. 2007). TCs form in the northern SCS in summer (southwesterly summer monsoon period), while TCs form in the southeastern SCS in winter (northeasterly winter monsoon period).
In the summer monsoon season, the SCS is influenced by two major components of intraseasonal oscillation—namely, the northeastward-propagating Madden–Julian oscillation (MJO, 30–60 days; Madden and Julian 1971; Li and Zhou 2013a) and the northwestward-propagating quasi-biweekly oscillation (QBWO, 10–20 days; Jia and Yang 2013). Huang et al. (2011) found that intraseasonal oscillation can modulate TC geneses in the NWP, including the SCS. Li and Zhou (2013a) discussed the impacts of both MJO and QBWO on the frequency of local TC genesis in the SCS and found that in general tropical cyclogenesis is enhanced (suppressed) in the convective (nonconvective) phases of MJO and QBWO in the SCS. They showed that in the convective phases of QBWO, tropical cyclogenesis is significantly enhanced, while in the nonconvective phases of MJO, tropical cyclogenesis is significantly suppressed. Li and Zhou (1995, 2013a) also showed that the kinetic energy of QBWO is larger than that of MJO, and thus the effect of QBWO on tropical cyclogenesis in the NWP (including the SCS) should be considered.
Compared to the genesis of TCs, understanding and forecasting the motion of TCs in the SCS is even more important for disaster prevention. This is mainly because TCs formed in the SCS, in general, have a relatively short time to intensify before they make landfall because of the small basin size. The motion of a TC is mainly controlled by the large-scale steering flow (Neumann 1992) and also affected by other factors, such as beta drift (e.g., Chan and Williams 1987; Wang and Holland 1996a,b), spatial distribution of SST (Chan 1995), vertical wind shear (Wu and Emanuel 1995a,b; Wang and Holland 1996c), and diabatic heating (Wu and Wang 2000, 2001). Besides the above factors, some studies also show that intraseasonal oscillation (ISO) plays a major role in the sudden change of TC tracks (Wu et al. 2011a,b; Liang et al. 2011; Wu et al. 2013). Yang et al. (2015) studied the impacts of intraseasonal oscillation (20–100-day period) on the motion of TCs in the SCS. They separated the TCs into westward-moving TCs and eastward-moving TCs. Their results show that the motion of the westward-moving TCs is controlled by the background steering flow, while the motion of the eastward-moving TCs is controlled by the steering flow induced by intraseasonal oscillation. This suggests that intraseasonal oscillation can affect not only TC genesis but also TC motion in the SCS.
Previous studies have demonstrated that the activity of TCs, including genesis and motion, in the SCS may be affected by both MJO and QBWO. However, few studies have focused on the individual effects of MJO and QBWO on the local TCs in the SCS, and some issues still remain to be addressed. For example, Li and Zhou (2013a) discussed only the effect of these two intraseasonal oscillations on the genesis frequency of local TCs in the SCS, while the details of the effect were not investigated, such as the genesis location and motion of local TCs, because they focused mainly on the whole NWP. Yang et al. (2015) analyzed only the impact of intraseasonal oscillations with periods larger than 20 days on the motion of local TCs in the SCS and did not investigate the individual effects of MJO and QBWO.
The major objectives of this study are to provide a detailed analysis on the individual effects and their relative importance of MJO and QBWO on the local TC genesis frequency and location and prevailing tracks (westward and northeastward movements) in the SCS. The analysis is performed for TCs formed during May–September in the southwesterly summer monsoon season over the SCS from 1979 to 2013. We will show that the MJO and QBWO have considerably different impacts on both TC genesis location and motion in the SCS. The rest of the paper is organized as follows. Section 2 briefly introduces the data and methodology used in this study. The patterns of the MJO and QBWO modes and their impacts on TC genesis and motion are discussed in section 3. The major results are summarized in the last section.
2. Data and methodology
The TC best-track data between 1979 and 2013 from the Japan Meteorological Agency (JMA) are used in this study. The dataset includes 6-hourly latitude and longitude of the TC center, minimum central sea level pressure, and 10-min averaged maximum surface wind speed.
The atmospheric data used in our analysis are the daily European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim) data, which have a spatial resolution of 0.75° longitude × 0.75° latitude on 37 pressure levels in the vertical. The relative vorticity at 850 hPa is used to represent the relative vorticity in the lower troposphere. The 500-hPa geopotential height is used to represent the western Pacific subtropical high (WPSH). The wind fields at 200 and 850 hPa are used to calculate the vertical wind shear. The 500-hPa winds are used to represent the steering flow of TC motion (Chan and Gray 1982).
Similar to previous studies (Kim et al. 2008; Huang et al. 2011; Jia and Yang 2013; Li and Zhou 2013a,b), we used the empirical orthogonal function (EOF) analysis to get the dominant patterns of the two intraseasonal modes—namely, the MJO and QBWO—based on the 30–60-day and 10–20-day bandpass-filtered data using a Gaussian filter. The EOF analysis is performed in the domain of 0°–30°N, 100°–170°E, the same as that used in Jia and Yang (2013). The MJO and QBWO modes are then divided into eight phases based on their respective first two principal components (PC1 and PC2), also the same as in Jia and Yang (2013, see their Fig. 5).
a. Patterns in different MJO and QBWO phases
Figures 1 and 2 show the composite anomalies of OLR and 850-hPa winds in summer in the different MJO and QBWO phases, respectively, overlapped with the initial genesis positions of local TCs in the corresponding phases. For the MJO, the OLR anomalies exhibit a northeastward propagation in the SCS. Initially, enhanced convection with its associated low-level cyclonic circulation occurs in the southern SCS in phase 5 and then propagates northeastward gradually until it attains its maximum in the northern SCS in phases 8 and 1. In phase 2, the anomalous convective activity weakens when its center shifts to the northeast SCS, as inferred from the reduced area coverage of negative anomalies in OLR, while positive OLR anomalies with suppressed convection occupy the southern SCS. The suppressed convection with the associated anticyclonic circulation then propagates northeastward and reaches the maximum positive OLR anomaly in the northern SCS in phases 4 and 5.
For the QBWO, a spatial pattern in the anomalous OLR, very similar to that for the MJO, occurs in the SCS. The only difference is that an alternative positive and negative OLR anomaly train propagates northwestward (Fig. 2) instead of northeastward in the MJO (Fig. 1). The enhanced convective signal (negative OLR anomalies) with a cyclonic circulation enters the southern SCS in phase 2 from east of the Philippine Sea. The negative OLR anomalies propagate northwestward slowly and dominate the whole SCS in phases 3 and 4, while they cover only the northern SCS in phase 5. In phase 6, the enhanced convective signal almost propagates out of the SCS, and the suppressed convection signal enters the southern SCS. The suppressed convection signal with an anticyclonic circulation propagates northward and is strengthened in both phases 7 and 8 and then weakens in phase 1 (Fig. 2).
Note that both MJO and QBWO are large-scale modes with a much larger scale in the zonal direction. Although they exhibit two centers with one over the SCS and one over the east of the Philippine Sea (Figs. 1 and 2), we consider that the two centers belong to the same system, which is split by the effect of the Philippine Islands. Although both MJO and QBWO also affect TC activities over the NWP, we focus on our analysis over the SCS since our interests are for local TCs formed over the SCS.
b. Impacts of MJO and QBWO on tropical cyclogenesis
As listed in Table 1, there are 87 TCs formed in summer over the SCS from 1979 to 2013. Most TCs formed in phases 1, 6, 7, and 8 of the MJO (73) and in phases 2, 3, 4, and 5 of the QBWO (70), when convection in the northern SCS associated with the two intraseasonal oscillation modes is enhanced (Figs. 1 and 2), while few TCs formed in the other phases. The daily genesis rate (DGR), defined as the daily TC number in each phase, shows that tropical cyclogenesis is significantly enhanced in phases 1, 7, and 8 of the MJO and in phases 3 and 4 of the QBWO, while it is significantly suppressed in phases 3, 4, and 5 of the MJO and in phases 1, 6, and 7 of the QBWO. The significance test is estimated using the same algorithm as that used in Li and Zhou (2013a).
In addition to affecting the TC genesis frequency, the MJO and QBWO can also influence the genesis location of local TCs. As shown in Fig. 1, the genesis locations of TCs shift northward during phases 6, 7, and 8 and northeastward during phase 1 of the MJO. The mean latitudes in phases 6, 7, 8, and 1 are 14.35°, 15.87 °, 16.86°, and 17.36°N, respectively. This is consistent with the northeastward propagation of the enhanced convective activity associated with the MJO. Similar to the MJO, the QBWO can also influence the genesis locations of local TCs. As we can see from Fig. 2, as the enhanced convective signal propagates northwestward, the genesis location also shifts northward during phases 2, 3, 4, and 5 and then shifts northwestward in phase 6 (Fig. 2). The mean latitudes in phases 2, 3, 4, and 5 show that the change of the genesis location is not significant between phases 2 and 3 or phases 4 and 5, while it is significant between phase 3 and phase 4. The mean latitudes in phases 3 and 4 are 15.73° and 16.96°N, respectively.
Wang et al. (2007) pointed out that the genesis locations of TCs in summer over the SCS are mainly controlled by the dynamical factors. We thus analyzed the low-level relative vorticity, vertical wind shear, and vertical pressure (p) velocity at 500 hPa in different phases of the MJO and QBWO, respectively, to investigate how the MJO and QBWO affect the environmental parameters and accordingly influence the summer tropical cyclogenesis over the SCS. As done in previous studies (Kim et al. 2008; Li and Zhou 2012, 2013a,b), we divide the MJO and QBWO into four combined phases labeled 8 + 1, 2 + 3, 4 + 5, and 6 + 7 based on the numbers of TC genesis.
Figure 3 shows the composite anomalies of 850-hPa relative vorticity, 200–850-hPa vertical wind shear, and 500-hPa vertical p velocity in different phases of the MJO. Note that vertical wind shear is defined as the magnitude of the difference between the total vector winds at 200 and 850 hPa, and thus the negative (positive) value of vertical wind shear anomaly means that the vertical wind shear weakens (strengthens). In phase 6 + 7 (2 + 3) of the MJO, changes in vertical wind shear are not significant, but the low-level relative vorticity is significantly increased (decreased) in the central SCS with anomalous upward (downward) motion dominating. With the northward propagation of the MJO, the relative vorticity increases (decreases) in the northern SCS, the vertical wind shear weakens (strengthens) in the northwestern SCS, and strong anomalous upward (downward) motion dominates the northern SCS in phase 8 + 1 (4 + 5) of the MJO. All these changes in phases 8 + 1 and 6 + 7 (phases 2 + 3 and 4 + 5) of the MJO are favorable (unfavorable) for tropical cyclogenesis, resulting in more (fewer) TCs forming in the SCS (Table 1).
The same composite analysis of the large-scale dynamical parameters was done for the QBWO. In phase 2 + 3 (phase 6 + 7), the 850-hPa relative vorticity significantly increases (decreases) from the Luzon Strait to central Vietnam, the 200–850-hPa vertical wind shear weakens (strengthens) in the whole northern SCS, and anomalous upward (downward) motion dominates in the central SCS (Fig. 4). In phase 4 + 5 (8 + 1), these changes are enhanced and shift northward with the northward propagation of the QBWO. The strong positive (negative) relative vorticity anomalies and anomalous upward (downward) motion dominate the whole northern SCS. The vertical wind shear significantly weakens (strengthens) in the northwestern SCS. All these changes in phases 2 + 3 and 4 + 5 (phases 8 + 1 and 6 + 7) are favorable (unfavorable) for tropical cyclogenesis, leading to more (fewer) TCs forming in the SCS (Table 1).
To examine the combined influence of the MJO and QBWO on tropical cyclogenesis, we define those phases as active phases when more TCs formed in the SCS (phases 1, 6, 7, and 8 of the MJO and phases 2, 3, 4, and 5 of the QBWO) and the other phases as inactive phases. Among 87 local TCs, 60 formed in both the MJO and QBWO active phases, 10 (13) TCs formed in the MJO inactive (active) phases with the QBWO active (inactive) phases, while only 4 TCs formed in both the MJO and QBWO inactive phases, confirming that the MJO and QBWO have significant impacts on local TC genesis over the SCS (Table 2).
Figure 5 shows the composite anomalies of 850-hPa relative vorticity, 200–850-hPa vertical wind shear, and 500-hPa vertical p velocity in the combined active and/or inactive phases of the MJO and QBWO. In both the MJO and QBWO active phases, the 850-hPa relative vorticity [Fig. 5a(1)] and upward motion [Fig. 5c(1)] are significantly enhanced in the whole northern SCS, with peaks located west of Luzon. Concurrently, the vertical wind shear significantly weakens in the northwestern SCS, with the largest reduction east of Hainan [Fig. 5b(1)]. All these changes are favorable for local TC genesis, leading to most TCs (60 out of 87) being formed during this period and suggesting that TC genesis is highly associated with the active phases of the MJO and QBWO. There were 13 local TCs formed in the MJO active phases with the QBWO inactive phases and 10 local TCs formed in the MJO inactive phases with the QBWO active phases, indicating that the effects of the MJO and QBWO on local TC genesis in the SCS are roughly comparable. In the inactive phases of both the MJO and QBWO, changes in relative vorticity, vertical wind shear, and vertical motion are significant in the northern SCS and unfavorable for TC genesis over the SCS, corresponding to the formation of only four TCs in the studied period (1979–2013 summers).
Note that the TC circulation has not been removed in the above filtered and composite analyses. Therefore, the results above are partly contaminated by the TCs themselves. Hsu et al. (2008) showed that synoptic-scale motions (including TCs) may contribute significantly to intraseasonal oscillation (explaining as much as 50% of the total intraseasonal variance over the NWP). They revealed that TCs contribute about 10%–40% in the northeastern SCS to the intraseasonal variance of the 850-hPa vorticity. That suggests that TCs may contribute considerably to intraseasonal oscillation (both the MJO and QBWO). However, because of the large difference in the time and spatial scales between intraseasonal oscillation and synoptic-scale motions, we consider that intraseasonal oscillation provides the large-scale background for TC genesis and the TC system would feed back into (enhance) the large-scale intraseasonal oscillation after its formation or development. Therefore, the major conclusions based on filtered and composite analyses above should not be altered by the possible contamination of TC circulation in the intraseasonal components since we mainly focused on the TC genesis phase and the subsequent motion.
c. Impacts of MJO and QBWO on TC motion
After genesis, the motion of a TC is mainly controlled by its large-scale environmental flow. Because of the semienclosed, relatively small ocean basin of the SCS, TCs formed in the SCS would mostly make landfall over surrounding countries. Here, on the intraseasonal time scale, we divide all local TCs into two groups: westward- and eastward-moving TCs, as in Yang et al. (2015). Among the 87 TCs in the studied period, 53 TCs moved westward and 34 TCs moved eastward (Fig. 6; Table 2). As listed in Table 2, westward-moving TCs could occur in any phase of the MJO and QBWO, while no eastward-moving TCs occurred in phases 4 and 5 of the MJO and phases 1, 7, and 8 of the QBWO. Most eastward-moving TCs (22 out of 34) formed in phases 1 and 8 of the MJO, while most eastward-moving TCs (28 out of 34) formed in phases 3 and 4 of the QBWO. Only in phases 2 and 8 of the MJO and in phase 4 of the QBWO, more TCs moved eastward than westward, indicating that the background steering flow on the intraseasonal time scale is favorable for TCs to move eastward.
In the NWP, the motion of TCs on the intraseasonal time scale is highly associated with the WPSH (Li and Zhou 2013b). To understand the prevailing motion of TCs in different phases of the MJO and QBWO, we first examine the position of the WPSH and the associated steering flow in each phase of the MJO and QBWO (Figs. 7 and 8). As shown in Fig. 7, the WPSH shifts eastward (westward) when convection is enhanced (suppressed) by the MJO in the northern SCS. The position of the WPSH retreats eastward from phase 7 to phase 2 as the convective MJO signal propagates northeastward. In phase 6, a westerly flow appears in the central Vietnam when the enhanced convection associated with the MJO enters the SCS (Fig. 1) and then intensifies and shifts northward in phases 7, 8, 1, and 2, causing the WPSH to retreat eastward. The position of the WPSH is much more eastward in phases 1, 2, and 8 of the MJO with the steering flow (500-hPa wind) favorable for TCs to move eastward, corresponding to more eastward-moving TCs. In phases 4 and 5 of the MJO, the WPSH intensifies and extends westward and the strong easterly flow in the northern SCS prevents TCs from moving eastward, leading to no eastward-moving TCs in the two phases in the studied period (Table 2).
Similar to the case in the MJO, the position of the WPSH is much more westward (eastward) in the inactive (active) phases of the QBWO (Fig. 8). With the northward propagation of the enhanced convective signal in the QBWO, the associated cyclonic circulation also moves northward and the strong westerly flow to the south pushes the WPSH eastward from phase 2 to phase 5, while the WPSH extends westward into the northern SCS and the strong easterly flow dominates the northern SCS from phase 6 to phase 1 (Fig. 8). With the eastward shift of the WPSH, the steering flow is favorable for TCs to move eastward in phases 3 and 4, leading to most TCs (28 out of 51) formed in these two phases moving eastward. In particular, in phase 4 the westerly flow extends to about 130°E and steers most TCs (16 out of 27) eastward. In phases 7, 8, and 1 of the QBWO, with the westward shift of the WPSH, the northern SCS is dominated by easterly flow and thus no TC moves eastward.
The variation of the WPSH in association with the MJO and QBWO corresponds to changes in the steering flow that controls TC motion. Here we calculate the steering flow anomalies for the eastward- and westward-moving TCs, which are defined as the 3-day average wind field after a TC genesis, as used in Yang et al. (2015). Figure 9 shows the TC genesis locations and three different steering flows calculated using the original wind fields, 30–60-day bandpass-filtered wind fields, and 10–20-day bandpass-filtered wind fields, respectively. As seen in Figs. 9b(1) and 9c(1), the eastward-moving TCs mostly formed in the western part of the low-frequency cyclonic circulation and then moved eastward under the influence of westerlies south of the cyclonic circulation, while the westward-moving TCs mostly formed in the eastern part of the low-frequency cyclonic circulation and are then often steered by the easterlies north of the cyclonic circulation to move westward [Figs. 9b(2) and 9c(2)]. These results are consistent with the numerical results of Liang and Wu (2015) and Liang et al. (2014), in particular for those TCs that moved westward.
For the original steering flow anomaly of the eastward-moving TCs, an anomalous strong cyclonic circulation lies in the northeast SCS. The strong anomalous westerly flow comparable to the background steering flow (Fig. 7) to the south of the TC extends eastward to 150°E and steers the TCs eastward [Fig. 9a(1)]. For the original steering flow anomaly of the westward-moving TCs, an anomalous cyclonic circulation lies in the northern SCS. The associated anomalous southwesterly flow is much weaker than that for the eastward-moving TCs. As a result, the summer mean strong background easterly flow dominates, steering these TCs westward. In addition, an anomalous easterly flow in the latitudinal band between 20° and 30°N also favors TCs moving westward.
Patterns of the steering flow anomalies associated with the MJO and QBWO for the eastward-moving TCs are similar to the original steering flow anomalies [Figs. 9b(1) and 9c(1)]. The sum of the two steering flow anomalies is comparable to the original steering flow [Fig. 9a(1)], indicating that the two intraseasonal components contribute dominantly to the eastward movements of the eastward-moving TCs. Note that the anomalous steering flow associated with the QBWO is comparable to that associated with the MJO [Figs. 9b(1) and 9c(1)]. This demonstrates that the QBWO contributes equally to the motion of local SCS TCs. We also checked the steering flow of each eastward-moving TC. The results show that the motion of most of the eastward-moving TCs (25 out of 34) was affected by both the QBWO and MJO, and among the other nine TCs, three are dominated by the QBWO, three are mainly controlled by the MJO, and the other three are not significantly affected by either the MJO or QBWO. This further confirms that the MJO and QBWO play a predominant role in controlling the motion of the eastward-moving TCs. For the westward-moving TCs, the intraseasonal steering flow is much weaker than the background steering flow [Figs. 9b(2) and 9c(2)] and thus plays a secondary role, and TCs are predominantly steered westward by the background flow, consistent with the result of Yang et al. (2015).
In this study, the effects of the MJO and QBWO on the genesis and motion of TCs in summer locally formed over the SCS have been investigated. The results show that both the MJO and QBWO can affect the genesis frequency and location and the motion of local TCs over the SCS. In the active (inactive) phases of the MJO and QBWO, convection in the northern SCS is enhanced (suppressed) with positive (negative) 850-hPa relative vorticity anomalies, reduced (enhanced) vertical wind shear, and anomalous upward (downward) motion, leading to more (fewer) TCs forming in the SCS. With the northward propagation of the MJO and QBWO, the TC genesis locations also shift northward in the convectively active phases.
Both the MJO and QBWO influence the TC motion over the SCS through their modulation of the zonal movement of the WPSH. When convection in the northern SCS is enhanced (suppressed) in the active (inactive) phases of the MJO and QBWO, the WPSH shifts eastward (westward), inducing an anomalous westerly (easterly) flow in the northern SCS, which is favorable for the eastward (westward) movement of TCs. The composite anomalies in the steering flow show that both the MJO and QBWO can comparably affect the movement of TCs. In particular, for the eastward-moving TCs, the strong westerly steering flow anomalies associated with the MJO and QBWO are comparable to the original steering flow, indicating that both the MJO and QBWO play an important role in controlling the motion of the eastward-moving TCs. For the westward-moving TCs, the steering flow anomalies associated with the MJO and QBWO are much weaker than the original steering flow, indicating that the westward-moving TCs are predominantly controlled by the background easterly steering flow.
This study has focused mainly on TCs locally formed in the summer monsoon season over the SCS. Besides local TCs, the SCS is also often affected by TCs that form in the NWP and then enter the SCS (nonlocal TCs). Actually, the number of nonlocal TCs is considerably larger than that of local TCs in summer (Ling et al. 2015). According to our knowledge, the impacts of the MJO and QBWO on nonlocal TCs over the SCS have not been investigated in the literature, which will be discussed in a companion paper.
This study was supported by the National Basic Research Program of China (2013CB430301), National Natural Science Foundation of China (NSFC; 41125019, 41306024, and 41321004), and the basic scientific research of Second Institute of Oceanography, State Oceanic Administration of China (JT1301). Y. Wang was supported in part by NSFC Grants 41375093 and 41375098 and in part by USGS Grant G12AC20501 to the University of Hawai‘i at Mānoa.