1 INTRODUCTION

East Taiwan Channel (ETC, Fig. 1) is an important channel through which the Kuroshio and waters from the Pacific Ocean enter the East China Sea (ECS) (Johns et al., 2001; Yan and Sun, 2015; Yan et al., 2016; Andres et al., 2017; Yin et al., 2019). More importantly, cross-shelf water exchange off the northeastern Taiwan, China, is modulated substantially by the intensity of the ETC inflow under the influence of mesoscale eddies (Yin et al., 2017, 2019, 2020; He et al., 2019b). Furthermore, the cross-shelf waters off the northeastern Taiwan, China, influence the circulation (Xu et al., 2018; Yang et al., 2018a, b), nutrient supplement (Li et al., 2014; Chen et al., 2017; Wang et al., 2018; Zhou et al., 2018), and biological systems in the ECS (Lu and Lee, 2014; Xu et al., 2019; Zhao et al., 2019; Dai et al., 2020). Thus, it is essential to explore the variation of ETC volume transport (ETCVT) in detail. In previous related studies, based on 20-month mooring observations from the PCM-1 array, the mean ETCVT was determined as 21.5 (±2.5) Sv (Johns et al., 2001), and the ETCVT time series showed a significant 100-d period, as well as shorter periods of near 40, 18, and 10 d (Johns et al., 2001; Zhang et al., 2001, Yang et al., 2015). Zheng et al. (2014) indicated that significant extrema could also be found on certain days in the ETCVT time series. Generally, oceanic mesoscale eddies and typhoons both have substantial influence on the northward volume transport off eastern Taiwan, China (Lee et al., 2013; Vélez-Belchí et al., 2013; Zheng et al., 2014; Yin et al., 2017, 2019, 2020; Liu et al., 2019a, b). However, while the influence of an oceanic mesoscale eddy is more likely to occur over a long period (i.e., > 10 d) (Zhang et al., 2001; Yin et al., 2017; He et al., 2019b), typhoons occur much more quickly and thus influence the surface waters of local seas over a shorter period. Therefore, the focus of this study was on the influence of typhoons on significant short-term extrema of the ETCVT.

Previous researches indicated that typhoons can have substantial influence on the physical (Chen et al., 2003; Kuo et al., 2011; Jan et al., 2013; Wei et al., 2014; Hsu and Ho, 2019; Yin and Huang, 2019), biological (Shiah et al., 2000; Chen et al., 2009; Siswanto et al., 2009; López-López et al., 2012; Zhang et al., 2014; Xu et al., 2017), and biogeochemical processes (Siswanto et al., 2008; Hung et al., 2013; Xu et al., 2017) in the upper ocean. In particular, three physical properties, i.e., flow velocity (Sun et al., 2009; Chang et al., 2010; Zheng et al., 2014; Hsu et al., 2018), the temperature and salinity structure (Tsai et al., 2008, 2013; Liu and Wei., 2015; Zheng et al., 2017; Hsu and Ho, 2019; Liu et al., 2019a, b), and the mixed layer depth (Wu and Chen, 2012; Zhang et al., 2014; Hsu and Ho, 2019; Zhang et al., 2019; He et al., 2020), can each exhibit marked variation in response to the passage of typhoons. Recent studies showed that the strong cyclonic wind fields of typhoons can have considerable influence on the surface velocity and flow structure of the upper ocean (Morimoto et al., 2009; Sun et al., 2009; Hsin et al., 2010; Zheng et al., 2014; He et al., 2020). For instance, Zheng et al. (2014) noted that the maximum northward surface velocity decreased from > 1.3 m/s to < 1.1 m/s when Typhoon Morakot moved westward toward the seas off eastern Taiwan, China, and that the surface volume transport east of Taiwan, China, decreased substantially. More importantly, the analysis in this study indicates that the short-term variations of ETCVT in response to the effects of typhoons differed depending on the track of the typhoons over the seas off eastern Taiwan, China, some of the short-term variations of ETCVT even exhibited an inverse pattern. Therefore, specific investigation of the impact of different tracks of typhoons on the ETCVT is needed.

Based on the International Best Track Archive for Climate Stewardship and a combination of data from the National Hurricane Center and Joint Typhoon Warning Center, it has been established that on average there are 25 recorded typhoons (≥35 kt) annually in the western Pacific Ocean (Schreck et al., 2014), and on average 3.6 typhoons will cross the seas off eastern Taiwan, China. Focusing on the impact of typhoons on the ETCVT, this study investigated historical cases of typhoons that occurred during 1993-2017 (Fig. 2). The details of typhoons characterized by three types of track were summarized. Each type causes specific changes to the ETCVT, and representative cases of each type of typhoon were examined in this study.

2 DATA AND METHOD 2.1 Study area

The sea area off eastern Taiwan, China (20°N-25°N, 120°E-125°E) was designated the study sea area (area shaded red, Fig. 1). The Kuroshio Current and waters from the Pacific Ocean flow northward into the ECS through the ETC, and the ETCVT is sensitive to the flow field east of Taiwan, China (Zheng et al., 2014; Yan et al., 2016; Yin et al., 2020). Previous studies (Zheng et al., 2014; Hsu and Ho, 2019) indicated that typhoons that pass over the seas off eastern Taiwan, China have considerable influence on the surface flow in this region and can result in significant extrema of the northward transport into the ECS.

2.2 Best track data

The best track data of typhoons that passed over the seas off eastern Taiwan, China, during 1993-2017 were acquired from the UNISYS database (http://weather.unisys.com/hurricanes/search). This website provides detailed historical hurricane and tropical cyclone data from 1851 with 6-h temporal resolution. The UNISYS weather server provides comprehensive weather information provided as graphics and data in a meteorological format. In addition, further information regarding cases of typhoons off eastern Taiwan, China was supplied by the China Meteorological Administration tropical cyclone database (Ying et al., 2014), and the data are available from the following website: http://www.typhoon.org.cn/.

2.3 HYCOM data

Reanalysis and analysis data generated by the dataassimilative global HYbrid Coordinate Ocean Model (HYCOM) were used to reveal the flow field off eastern Taiwan, China. HYCOM applies the Navy Coupled Ocean Data Assimilation system (Cummings and Smedstad, 2013) to assimilate available satellite altimeter, sea surface wind and sea surface temperature observations, in situ sea surface temperatures, and vertical temperature and salinity profiles derived from expendable bathythermographs, Argo floats, and moored buoys. The three vertical diffusion mixing submodels of HYCOM are capable of resolving both geostrophic shear and ageostrophic wind-driven shear in the upper ocean (Chassignet et al., 2007) and boundary currents in the deep ocean (Jiang et al., 2020). HYCOM data have been used to study various phenomena in the seas around Taiwan, China (Yin et al., 2014, 2017; Wu et al., 2017; Hsu et al., 2018; He et al., 2019a, 2020) and on the global scale (Yu et al., 2015, 2017, 2018). More importantly, HYCOM output has been compared with both the volume transport off eastern Taiwan, China derived from in situ observations from the PCM-1 array, and the Kuroshio intensity off eastern Taiwan, China derived from the tidal sea level difference across the Kuroshio axis (Yan and Sun, 2015), and it has proven reliable in revealing the northward volume transport off eastern Taiwan, China. As GOFS 3.0 reanalysis data only cover the period from January 1993 to December 2012 (Yu et al., 2015), GOFS 3.0 analysis data were also used to provide the flow field off eastern Taiwan, China, in the recent five years (January 2013 to December 2017). The daily data are available from the HYCOM server (https://www.hycom.org/) with 0.08° horizontal resolution and 41 vertical levels.

2.4 Method

For quantitative estimation of the northward volume transport, ETCVT was calculated based on the vertical integral of volume transport through the ETC section:

where v indicates the normal velocity through the ETC section, and s indicates the area of each grid cell from the bottom to the surface along the ETC section.

To reveal the short-term extrema of ETCVT (i.e., within 10 d), the ETCVT short-term variation (DVT) was calculated as follows:

DVT=ETCVT-ETCVTs,

where ETCVT is the daily time series, and ETCVTs is the 10-d low-pass filtered time series of ETCVT.

The DVT extrema during the passage of each typhoon through the study area were defined as follows:

(1) the DVT maximum values satisfied the conditions:

DVT(n) > DVT(n−1) & DVT(n) > DVT(n+1);

(2) the DVT minimum values satisfied the conditions:

DVT(n) < DVT(n−1) & DVT(n) < DVT(n+1);

(3) the DVT extrema of each typhoon case were the maximum of the DVT maximum values and the minimum of the DVT minimum values when typhoons were passing nearby. The transport extrema must have occurred within the period from 3 d before the typhoon center entered the study area to 3 d after the typhoon center left the study area.

3 ETCVT RESPONSES TO DIFFERENT TRACKS OF TYPHOONS

Based on historical best track data spanning 25 years (1993-2017), on average 5 cyclones, 3.6 typhoons, and 1 super typhoon cross the study sea area annually. However, the actual numbers of these typhoons fluctuate markedly (Fig. 2). For instance, six typhoons passed over the study area in 2003, whereas only two typhoons crossed the region in 2014. The fluctuation of typhoon occurrence in the study area might be related to the migration of typhoon occurrence in the North Pacific Ocean (Sun et al., 2019; Hung et al., 2020). To reveal the response of ETCVT to these typhoons, typical typhoon cases were identified, where cases of tropical storms and tropical depressions were also included within the category "Cyclone" (Fig. 2).

It can be seen from Fig. 3 that ETCVT showed significant short-term extrema when typhoons passed over the study sea area, and that 28% of the DVT absolute values exceeded 2.7 Sv (twice the standard deviation of DVT) occurred under the influence of historical typhoons. Moreover, 71% of the DVT absolute values that exceeded 5 Sv occurred under the influence of typhoons crossing the study sea area. The most substantial positive and negative extrema that occurred under the influence of typhoons were 12.5 and -10.9 Sv, respectively, while the mean ETCVT derived from the HYCOM data was 26.6 Sv with a standard deviation of 6.3 Sv.

More importantly, ETCVT showed different extrema responses to typhoons following specific types of track. To study the ETCVT response off eastern Taiwan, China, we identified three types of typhoon based on their track. Specifically, Type I typhoons moved westward across the study sea area, Type II typhoons moved northward over the eastern side of the study sea area, and Type III typhoons moves northward over the western side of the study sea area. It should be noted that atypical cases were included in Type Others (Figs. 4 & 5). Clusters of the above three types of cyclone are shown in the upper row of Fig. 4, and significant cases of each type are presented in the lower row of Fig. 4.

As revealed in Fig. 4, The typhoons that passed over the study sea area were predominantly Type I. There were 54, 21, and 12 cases of Type I, II, and III tracks, respectively, and the above three types of cyclone accounted for 70.2% of the historical cyclones (details of typhoon cases are listed in Table 1). The statistics of the significant typhoon cases also showed a similar pattern, i.e., the significant cases of the three types of typhoon (bold in Table 1) accounted for 74.2% of the significant historical cases.

Further statistical analysis of the DVT in response to the different tracks of the typhoons is illustrated in Fig. 5. The extrema of DVT intensity under the influence of typhoons showed reasonable positive correlation with typhoon intensity, i.e., stronger typhoon intensity was more likely to result in greater DVT extrema (Fig. 5). However, the DVT variation in response to the three types of typhoon differed. To reveal the specifics of typical typhoons of each type and the associated ETCVT and DVT responses, representative typhoon cases were selected. Details of each representative typhoon case selected are listed in Table 1.

3.1 Type I: crossing the study area from east to west

Herb was the 10^th typhoon (category: Super Typhoon-5) to strike the western Pacific Ocean during 1996. Detailed information on Herb is shown in Table 1. Herb moved westward across the study sea area from east to west (Fig. 6). It entered the study sea area at 00:00 on July 31, 1996, and exited at 00:00 on August 1, 1996.

As revealed in Fig. 6a, when the center of Herb entered the study area on July 31, 1996, ETCVT exhibited a negative extremum (NE). Then, when the center of Herb exited the study area on August 1, 1996, a positive extremum (PE) in ETCVT appeared. The vertical profile of ETCVT reveals that the transport in the surface layer above 100 m decreased substantially when the center of Herb entered the study sea area, thereby resulting in the NE response (Fig. 6b). When Herb left the study sea area, ETCVT in the surface layer above 100 m increased considerably, resulting in the PE response. On August 3, i.e., the third day after Herb exited the study area, the vertical profile of ETCVT returned to normal levels (Fig. 6b).

It can be seen from Fig. 6c-f that the sea surface velocity field was influenced markedly by the cyclonic wind field of Herb. The wind field to the west (east) of the typhoon center was directed southward (northward) according to the typhoon wind profile. On July 31, 1996, when the center of Herb entered the study sea area, the surface flow field was disturbed. The northward flow velocity in the sea surface layer was weakened substantially (the sea surface flow velocity even changed direction and became directed southward) (Fig. 6d) because it was forced by the southward wind field to the west of the center of Herb. This led to marked decrease of the ETCVT in the surface layer above the water depth of 100 m. On August 1, 1996, when the center of Herb left the study area, the flow velocity in the sea surface layer became directed northward and enhanced considerably (Fig. 6e), because it was forced by the northward wind field to the east of the center of Herb. This process led to the marked increase of the ETCVT in the surface layer above the water depth of 100 m. On August 3, 1996, the third day after Herb had left the study area, the distribution of the horizontal velocity returned to the normal geostrophic pattern (Fig. 6f).

3.2 Type II: northward across the eastern side of the study area

Jelawat was the 18^th typhoon (category: Super Typhoon-5) to strike the western Pacific Ocean during 2012. Detailed information on Jelawat is shown in Table 1. Jelawat moved northward across the eastern side of the study sea area (Fig. 7). The center of Jelawat entered the study sea area at 06:00 on September 27, 2012, and it exited at 18:00 on September 28, 2012.

As revealed in Fig. 7a, when the center of Jelawat entered the study sea area on September 27, 2012, a weak increase of ETCVT occurred. The weak increase of ETCVT continued to strengthen when the center of Jelawat left the study sea area on September 28, 2012. However, on September 30, 2012, two days after the center of Jelawat exited the study sea area, a notable NE (i.e., more than -5 Sv) in ETCVT appeared. The vertical profile of ETCVT reveals that the transport in the surface layer above 100 m increased slightly when Jelawat entered the study area from the south to the north (Fig. 7b). Then, when the center of Jelawat exited the study sea area northward, the ETCVT in the surface layer above 50 m was weakened, while that in the surface layer between 50 and 100 m was enhanced steadily, resulting in a weak PE of ETCVT (Fig. 7b). On September 30, 2012, two days after the center of Jelawat had exited the study area, the ETCVT in the surface layer gradually recovered to its normal magnitude, whereas the ETCVT in the subsurface layer below 100 m was diminished substantially, resulting in a delayed NE of the ETCVT (Fig. 7c). On October 3, 2012, the fifth day after Jelawat had left the study sea area, the vertical profile of ETCVT returned to normal levels (Fig. 7c).

It can be seen from Fig. 7d-k that the sea surface velocity field was influenced markedly by the cyclonic wind field of Jelawat, and that the surface geostrophic flow was substantially disturbed. Furthermore, the velocity in the subsurface layer also showed notable and unique responses. On the first two days after the center of Jelawat had entered the shaded area, the northward velocity through the ETC in the surface layer above 50 m weakened gradually (Fig. 7d & e). In contrast, in the deeper surface layer between 50 and 100 m, the northward velocity increased gradually, leading to a weak increase in the ETCVT (Fig. 7a & b). On September 29, 2012, when the surface velocity near the ETC was influenced strongly by the southward wind field to the west of the center of Jelawat, the northward Kuroshio velocity weakened substantially. This led to a notable decrease of the ETCVT in the surface layer above the water depth of 100 m (Fig. 7c). On September 30, 2012, two days after the Jelawat had left the study sea area, the surface velocity gradually returned to normal geostrophic patterns (Fig. 7c & f), while the velocity through the ETC in the subsurface layer below 100 m was weakened considerably (Fig. 7h-j). Consequently, the ETCVT passing through the ETC exhibited a delayed NE response (Fig. 7a). On October 3, 2012, the fifth day after the center of Jelawat had left the study area, the horizontal velocity field returned to its normal geostrophic patterns (Fig. 7g & k).

3.3 Type III: northward across the western side of the study area

Zeb was the 18^th typhoon (category: Super Typhoon-5) to strike the western Pacific Ocean during 1998. Detailed information on Zeb is shown in Table 1. Zeb moved northward across the western side of the study sea area (Fig. 8). The center of Zeb entered the study sea area at 06:00 on October 15, 1998, and it exited at 12:00 on October 16, 1998.

As revealed in Fig. 8, when the center of Zeb passed over the study sea area on October 16, 1998, a significant PE (> 5 Sv) of ETCVT appeared (Fig. 8a). The vertical profile reveals that ETCVT in the surface layer above the water depth of 100 m increased gradually on October 15, 1998 (Fig. 8b) when the center of Zeb entered the study sea area. On October 16, 1998, when the center of Zeb left the study sea area, the ETCVT in the surface layer above the water depth of 100 m was enhanced markedly (Fig. 8b), which led to a PE response in ETCVT. On October 20, 1998, the fourth day after Zeb had left the study sea area, the vertical profile of ETCVT returned to its normal levels (Fig. 8b).

It can be seen from Fig. 8c-f that the sea surface velocity field was influenced substantially by the cyclonic wind field of Zeb, and that the surface geostrophic flow was evidently disturbed. On October 15, 1998, when Zeb entered the study sea area, the northward velocity through the ETC above the water depth of 100 m increased gradually (Fig. 8b, On October 16, 1998, when the velocity near the ETC was strongly influenced by the northward wind field to the east of the center of Zeb, the northward velocity through the ETC was enhanced markedly (Fig. 8b & e), which led to a notable increase of the ETCVT in the surface layer. On October 20, the fourth day after Zeb had left the shaded area, the horizontal velocity field returned to the normal geostrophic pattern (Fig. 8b & f).

4 DISCUSSION

The results above provide strong evidences to support the previously stated viewpoint that a typhoon is an important influencing process of the seas off eastern Taiwan, China (Wei et al., 2014; Zheng et al., 2014, 2017; Hsu and Ho, 2019). The short-term northward transport through the ETC in response to the passage of typhoons was proven significant and shown to exhibit unique patterns. In Section 3, three types of typhoon were identified, and the response of transport through the ETC was shown to be closely related to the typhoon wind field above the sea surface.

Generally, the variation of the volume transport through the ETC depends on changes of the flow velocities to the east of Taiwan, China (upstream region of the ETC) (Zhang et al., 2001; Hsin et al., 2010, 2013; Wu et al., 2014, 2017; Yin et al., 2014, 2017, 2019; Yan and Sun, 2015; Yan et al., 2016), and the flow velocities to the east of Taiwan, China, are strongly influenced by the cyclonic wind field of passing tropical cyclones. The surface wind stress is against (follows) the northward velocities to the east of Taiwan, China, when a typhoon is located to the east (west) of the study sea area, which reduces (enhances) northward transport via the ETC.

The case studies of Type I (Fig. 6) and Type III typhoons (Fig. 8) are representative of the above mechanism. However, it should be noted that ETCVT variations are more complex under the influence of Type II typhoons. First, Type II typhoons move northward across the upstream region of the ETC. Ignoring the southward wind field of typhoons, which is against the surface northward flow velocities, the northward movement of a Type II typhoon could push the water to the east of Taiwan, China, northward into the ETC. This process could induce an increase in the transport in the surface mixing layer. As revealed in Fig. 7b, the northward transport in the surface layer of 50-100 m increased gradually as typhoon Jelawat moved northward. Taking into account the southward wind stress of the typhoon, the northward velocity in the near-surface layer above the water depth of 50 m was diminished substantially. The combination of these two processes in the surface layer induced a weak PE of the northward transport in the ETC. However, the delayed NE of ETCVT occurred when the seas to the east of Taiwan, China, were free from the influence of Typhoon Jelawat, and the surface transport gradually recovered to normal levels (Fig. 7c). The unique NE responses might be induced by a combination of topography, internal gravity waves (Chang et al, 2019), and bifurcation of northward sea waters (Yan et al., 2016) in the local seas, but further specific study on this topic is needed. Additionally, further study on Type II typhoons could enhance understanding of the interaction between typhoons and the local seas.

This study contributed to improved understanding of typical significant extrema of ETCVT (within days) off the eastern coast of Taiwan, China, and provided new perspective on the interactions between typhoons and the local seas. The three basic types of typhoon identified in this study could be used as a reference to understand extrema exhibited by the volume transport of sea water channels in other regions. It would also be worth studying the differences of these three typical typhoon cases on other physical processes (e.g., the temperature structure and mixed layer depth) and biogeochemical processes in local and downstream regions.

5 CONCLUSION

First, based on historical best track data spanning the past 25 years (1993-2017), on average 5 cyclones, 3.6 typhoons, and 1 super typhoon cross the study sea area off eastern Taiwan, China, annually. The northward ETCVT is strongly influenced by these typhoons, resulting in significant short-term extrema of ETCVT, as revealed by HYCOM data. Overall, 71% of short-term (within 10 d) ETCVT absolute values with > 5 Sv were affected by typhoons that passed over the study sea area, and the maximum typhoon induced ETCVT extrema (within 10 d) were 12.5 and -10.9 Sv. The ETCVT extrema intensity induced by typhoons showed reasonable positive correlation with the mean maximum wind speed of typhoons during the time they passed by.

Second, the response of ETCVT to typhoons with different tracks was revealed characteristically different. Accordingly, three typical types of typhoon were identified and their corresponding influence on ETCVT changes was summarized. Of the typhoons that passed over the seas off the east coast of Taiwan, China, Type I typhoons are dominant and they produce an NE response followed by a PE response in ETCVT. Type II typhoons result in a delayed NE response, whereas Type III typhoons result in a PE response. The interactions between the typhoon winds and the induced surface flows are very important for understanding these typical differences. The variations in the vertical profile of ETCVT in response to the different typhoon cases indicate that typical changes in ETCVT occur mainly within the surface layer above a water depth of 100 m. However, in addition to surface transport fluctuations, Type II typhoons result in substantial and delayed changes in ETCVT in the subsurface layer below 100 m, which is a subject worthy of further attention.

6 DATA AVAILABILITY STATEMENT

The datasets generated and analyzed during the current study are available to the public. The best track data of typhoons are available from http://weather.unisys.com/hurricanes/search and http://www.typhoon.org.cn/, the bathymetry data obtained from the General Bathymetric Chart of the Oceans (GEBCO) are available from http://www.gebco.net, and the HYCOM datasets are available from https://hycom.org/.

7 ACKNOWLEDGMENT

We wish to thank the CAS Key Laboratory of Ocean Circulation and Waves, Institute of Oceanology, Chinese Academy of Sciences for their support. Furthermore, the critical and constructive comments and suggestions of the reviewers are very helpful and valuable for improving this paper.