1 INTRODUCTION

The South China Sea (SCS) is the largest semi-enclosed marginal sea in the northwestern Pacific Ocean. Located in the northeast of SCS, Luzon Strait is the primary channel connecting the SCS and Pacific Ocean (Qu, 2002; Tian et al., 2006). Originating from the North Equatorial Current, the Kuroshio flows northward along the eastern Philippine coast and easily penetrates into the northern SCS when passing through the east side of the Luzon Strait (Caruso et al., 2006; Nan et al., 2015; Zhang et al., 2017). Influenced by Kuroshio intrusion, monsoon forcing, abrupt topography, and other factors, Luzon Strait has complex multiscale dynamic processes and biological responses (Xian et al., 2012; Liu et al., 2020; Tsutsumi et al., 2020).

Phytoplanktons are a key component of marine ecosystems. Chlorophyll-a (Chl-a) concentration is an important indicator of phytoplankton biomass, which is helpful in understanding the distribution and variation of marine primary productivity (Siswanto et al., 2005; Lin, 2012; Gaube et al., 2014). The magnitude of Chl a is generally affected by light and nutrients, which are further regulated by atmospheric and oceanic dynamic processes (Babin et al., 2004; Lin et al., 2010; Zhang et al., 2016). Based on years of satellite remote sensing data and voyage observations, previous studies have found that the Chl a in the SCS changes greatly from the coast and continental shelf to the open sea, which is high in coastal areas and low in the open sea (Ning et al., 2004; Zhao et al., 2005; Gao et al., 2010). Using two large-scale biological oceanographic surveys, Ning et al. (2004) preliminarily depicted the seasonal distributions of phytoplankton in the SCS and identified their link to the coupled physical-chemical-biological oceanographic processes. Gao et al. (2010) statistically estimated the vertical distribution of Chl a in the SCS, finding that the high Chl a was mostly concentrated in the surface layer in the nearshore and estuarine areas, while in the deep-water basin, the maximum of Chl a appeared in the subsurface layer. The subsurface chlorophyll maximum (SCM) is ubiquitous in stratified upper oceans (Gong et al., 2012). Zhang et al. (2016) found that there was a permanent SCM ranging from 48 m to 96 m in the central basin of the SCS. While in the northern basin, the SCM disappeared in winter, replaced by enhanced surface layer phytoplankton with high Chl a. Recently, Chen and Zhao (2021) identified 4 different patterns in the vertical Chl-a profiles in the northern SCS, and pointed out that they are mainly regulated by alternative limitations of nutrients and light from the surface to the bottom of the euphotic layer. In addition, Xu et al. (2022) investigated the characteristics of SCM in the SCS in summer through years of cruise observation data, and found that light attenuation and physical processes play a fundamental role in determining the SCM depth and SCM magnitude.

With regard to the Luzon Strait region, Chl a shows a significant annual cycle that is high in winter but low in summer, which is closely related to the coupled processes driven by the East Asian Monsoon (Ning et al., 2004; Xian et al., 2012). Phytoplankton blooms are often reported in the southwest of the Luzon Strait during winter, which is likely related to the nutrient supplies from upwelling (Tang et al., 1999; Peñaflor et al., 2007). However, the water over the Luzon Strait is generally stratified during summer, which leads to an oligotrophic environment and low phytoplankton biomass in the upper ocean (Ke et al., 2013; Xu et al., 2018). In addition to monsoon forcing, other oceanic dynamic processes, such as Kuroshio intrusion, mesoscale eddies, and turbulent mixing can also affect the ecological processes. Compared with the SCS water, the Kuroshio water is featured by high temperature, high salinity, and low nutrients in the upper layer, thus reducing local primary productivity once intruding into the SCS (Du et al., 2013; Liu et al., 2020). Nevertheless, Guo et al. (2017) proposed that the Kuroshio intrusion fronts enhanced Chl a in the northern SCS in winter because of high nutrients provided by intense upwelling in the frontal region. Mesoscale eddies can influence biogeochemical cycles through both vertical flux and horizontal advection of nutrients and ecosystems (Lin et al., 2010; Gaube et al., 2014). Chen et al. (2007) found increased Chl a and different phytoplankton compositions in a cyclonic eddy relative to the surrounding Kuroshio and SCS waters. He et al. (2016) investigated the effects of eddies on surface Chl a in the northern SCS and found that the positive Chl-a anomalies within cyclonic eddies are generally larger in magnitude than the negative Chl-a anomalies within anticyclonic eddies. Xiu et al. (2016) also pointed out that Chl a in the Luzon Strait region is mainly controlled by an eddy-pumping mechanism. Recently, Shang et al. (2021) found that the strong turbulent mixing over the western Luzon Strait can induce nutrient fluxes that transport nutrients not only to the SCM layer but also to the upper water, which scatters the nutrients in the water column and diffuses the SCM layer.

Despite there are many studies on the spatial distribution of surface Chl a in the SCS, the knowledge of temporal and vertical variations of Chl a is still insufficient, especially in the Luzon Strait. Satellite observation provides an opportunity to comprehensively understand the distribution of Chl a in the global oceans. However, it is limited to the surface ocean and sometimes is interfered by clouds. To compensate for the deficiency of vertical observations, it is not enough to rely on traditional voyage surveys. In recent years, some new observation methods, such as biogeochemical-Argo and underwater gliders, which can conduct frequent autonomous observations and provide continuous vertical profiles, have been applied to ocean investigations (Shu et al., 2019; Chai et al., 2021). In this study, the data of two gliders with a period of 44 days of measurements in the Luzon Strait were analyzed. We investigated the temporal and vertical variations of Chl a in the Luzon Strait in summer 2019 and attempted to examine the related influencing factors. Section 2 describes the dataset used in the analysis. Section 3 presents the variation of Chl a and analyzes the related physical processes. The conclusion and discussion are formulated in Section 4.

2 DATA AND METHOD 2.1 Glider observation

Autonomous, buoyancy-driven underwater gliders have been successfully used to investigate ocean dynamics in the SCS in recent years (Rudnick et al., 2013; Li et al., 2019; Qiu et al., 2019; Shu et al., 2019). The glider collects various measurements from its equipped sensors as it dives and climbs the water column. Two Sea-wing underwater gliders, developed by the South China Sea Institute of Oceanology, Chinese Academy of Sciences, were used in this study. Glider 1000J031 was equipped with Sea-Bird conductivity-temperature-depth (CTD) measuring temperature and salinity profiles. In addition to CTD, glider 1000K012 was equipped with a dissolved oxygen sensor and an optical sensor to measure dissolved oxygen, chlorophyll concentration, and turbidity. The parameters used in this study were temperature, salinity, and chlorophyll concentration. The gliders were deployed west of the Luzon Strait, diving to a depth of 1 000 m and tracking back and forth along 120.30°E from July 25 to September 6, 2019 (Fig. 1). Note that the glider 1000J031 was configured to run continuously along the section (120.30°E/20.25°N–21.15°N), while glider 1000K012 was configured to operate between two points (120.30°E/20.60°N and 120.30°E/21.00°N). The sampling cycle of the glider included a diving profile and a climbing profile, which took about 4 hours. The sample interval was set to be 6 s and the vertical resolution was about 1.0 m. During the observation period, a total of 510 temperature/ salinity profiles were collected by glider 1000J031, and 574 temperature/salinity/chlorophyll profiles were collected by glider 1000K012. In order to avoid the error caused by the difference of diving profiles and climbing profiles, only the climbing profiles were used in this study. Note that the preliminary calibration of CTD and chlorophyll fluorometer has been completed by the manufacturer before the deployment of gliders. The chlorophyll fluorometer has been characterized by the manufacturer using a fluorescent material. Based on the scale factors provided by the manufacturer, the chlorophyll concentration can be derived. Data quality control was conducted before using the data. Firstly, the temperature/salinity/chlorophyll profiles were processed by a vertical 7-bin moving average to reduce the burrs existing in the measurements. Then, the salinity data were calibrated by removing the thermal lag effect (Morison et al., 1994). Lastly, to eliminate the influence of diurnal variability and other high-frequency processes, the profiles were daily averaged and interpolated to a vertical resolution of 1 m.

2.2 Other data

The seasonal climatology data of surface chlorophyll is satellite-derived (MODIS-Aqua) Level-3 product, with a horizontal resolution of 4 km, which was obtained from the NASA's OceanColor Web (https://oceancolor.gsfc.nasa.gov/l3/). Daily sea level anomaly (SLA) and absolute geostrophic velocity from July 25 to September 6, 2019 were obtained from the Copernicus Marine Environment Monitoring Service (CMEMS, http://marine.copernicus.eu). The specific product used was SEALEVEL_ GLO_PHY_L4_REP_OBSERVATIONS_008_047, which was merged from different altimetry satellites, including Jason, TOPEX/Poseidon, GFO, ERS, and so on. A full description of the dataset is available at http://cmems-resources.cls.fr/documents/PUM/CMEMS-SL-PUM-008-032-051.pdf. Daily precipitation data with a spatial resolution of 0.25°×0.25° were extracted from the 3B42 product provided by the Tropical Rain Measuring Mission (TRMM) project. Satellite sea surface salinity (SSS) data from the Soil Moisture Active Passive (SMAP) platform were derived from Remote Sensing Systems (RSS), with a resolution of 40 km and 8-day running averages. The sea surface winds at 10 m above the sea surface were derived from the Cross-Calibrated Multi-Platform (CCMP) gridded product, with a temporal resolution of 6 h and spatial resolution of 0.25°× 0.25°. The tropical cyclone (TC) information, including center positions, maximum sustained winds (MSW) and minimum sea level pressures (MSLP) at interval of 6 h, is provided by the China Meteorological Administration (CMA) best track dataset (tcdata.typhoon.org.cn). This dataset contains information on all TCs that have passed through the western North Pacific and the SCS since 1949 (Ying et al., 2014; Lu et al., 2021). For typhoons landing in China, the best track dataset provides TC information with an interval of 3 hours from 24 hours before landing. In addition, seasonal climatological temperature and salinity from WOA13 were used to offer the basic characteristics of the northern SCS water and Kuroshio water.

2.3 Variables calculation

The mixed layer depth (MLD) is defined as the depth at which the potential density increase relative to the surface (10 m) equals the increase in surface potential density when the surface temperature decreases by 0.5 ℃ (de Boyer Montégut et al., 2004). According to previous studies, the depth of 22 ℃ isotherm is used to represent the thermocline depth in the northern SCS (Liu et al., 2001; Zhang et al., 2016; Xu et al., 2022).

3 RESULT 3.1 The variation of chlorophyll in the Luzon Strait

Figure 1 shows the spatial distribution of Chl a in the northeastern SCS in summer. High Chl a is generally distributed in the coastal area of Chinese mainland, and gradually decreases from the coast to the open sea. The Luzon Strait, as the main channel connecting the SCS and the western Pacific, has complex atmospheric and oceanic dynamic processes. The Chl a in the Luzon Strait is lower than that in the nearshore but higher than that in the western Pacific from satellite observations. To investigate the temporal and vertical variations of Chl a in the Luzon Strait, two gliders were deployed along 120.30°E and continuously measured for more than a month (from July 25 to September 6, 2019). The observation results are as follows.

Figure 2 shows the vertical distribution of Chl a observed by glider 1000K012. There was a prominent SCM in the Luzon Strait, ranging from 50 m to 110 m, with a mean depth of 75 m. The mean magnitude of SCM was 0.65 mg/m³. The SCM was located below the mixed layer and above the thermocline, where the optimal combination of light and nutrients was conducive to the growth of phytoplankton. The high Chl a (> 0.1 mg/m³) was limited to the layer above 200 m. Thus, the vertical integral of the Chl a in the upper 200 m was taken as the total phytoplankton biomass of the upper water column. There was a high correlation between the depth-integrated Chl a and the magnitude of SCM, with the correlation coefficient (more than 95% confidence) being as high as 0.84. Take 50% of the SCM magnitude as the threshold to determine the upper and lower boundaries of the SCM layer. The Chl a in the SCM layer accounted for 38%–75% of total Chl a in the upper water column during the study period, implying that the phytoplankton biomass was mainly concentrated in the SCM layer. Based on in-situ observations, Williams et al. (2013) found that the SCM depth corresponds to the base of the thermocline with a sharp nitrate gradient. The thermocline depth in the open basin of the SCS can be indicated by the depth of the 22 ℃ isotherm (Liu et al., 2001). Zhang et al. (2016) used the vertical displacement of the 22 ℃ isotherm to reflect the change of thermocline and available nutrients, pointing out that the magnitude of the SCM in the northern SCS was influenced by the vertical displacement of isotherms. Similar results can be seen in Fig. 2a that 22 ℃ isotherm was consistent with 1 024 kg/m³ isopycnal and located below the depth of the SCM in the Luzon Strait. The correlation coefficient between the thermocline depth and the SCM magnitude was -0.68, with 95% confidence level. This indicates that the shallow thermocline is favorable for the upward of nutrients and promotes the growth of phytoplankton, while the deep thermocline is unfavorable for the upward of nutrients and inhibits the growth of phytoplankton. The prominent variation of Chl a was mainly concentrated in the SCM layer, including the displacement of SCM depth and the change of the SCM magnitude. The temporal variation of Chl a can be divided into three main stages: July 30 to August 5 (Stage 1), August 6 to August 16 (Stage 2), and August 17 to August 28 (Stage 3). Before Stage 1, the SCM depth was relatively shallow and the SCM magnitude was up to 0.92 mg/m³. Subsequently, the SCM depth sharply deepened from 68 m to 100 m on July 30, when the SCM magnitude also reduced from 0.82 mg/m³ to 0.57 mg/m³. The deep SCM depth and low SCM magnitude remained until August 5. Afterward, a significant Chl-a bloom was observed in the SCM layer during Stage 2, lasting for nearly two weeks. In particular, the SCM depth was uplifted to 58 m and the SCM magnitude reached 1.12 mg/m³ on August 6. During Stage 3, the SCM magnitude fell to a low level once again. However, the SCM depth hardly changed and remained as shallow as that in Stage 2. In addition, the surface Chl a was slightly increased in Stage 3, especially on August 24. After August 28, the Chl a first decreased and then increased to a relatively high level, similar to that during and before Stage 1, respectively. The changes of Chl a were different in these stages, which may be related to different dynamic processes.

3.2 The related physical processes 3.2.1 Hydrodynamic environment

To better understand the variation of Chl a in Luzon Strat, we futher analyzed the hydrological variables during the study period. Figure 3 shows the temporal variations of vertical temperature and salinity observed by the gliders. The results of the two gliders were consistent, which further proves the authenticity and reliability of the glider observations. It was clear that the isotherms fluctuated greatly during the observation period. The thermocline was uplifted in Stage 2, and deepened in Stage 1 and Stage 3. In particular, Stage 1 and Stage 3 corresponded to the processes of subsurface salinization and surface desalination, respectively. From July 30 to August 5, a high salinity core was observed in the subsurface layer with the maximum value exceeding 34.8. This saline water may come from the western Pacific because the salinity of typical SCS water is generally no higher than 34.7 (Nan et al., 2011; Wang et al., 2021). Correspondingly, the Chl a was reduced during Stage 1. Afterward, the subsurface high-salinity core gradually disappeared in Stage 2, with the uplift of isotherms and the bloom of subsurface Chl a. In particular, there was a significant salinity decrease above 60 m in Stage 3. The surface salinity (2 m) observed by glider 1000J031 decreased from 34.21 (August 17) to 33.13 (August 25), with a reduction of 1.08. Similar results were also seen from glider 1000K012. Additionally, the MLD and thermocline depth both deepened in Stage 3, with the magnitude of SCM decreased and surface Chl a increased. After August 28, the low-salinity signal gradually disappeared, and was replaced by a body of high-salinity water similar to Stage 1. The underlying processes of the hydrodynamic environment change and their biological responses will be discussed in the following paragraphs.

3.2.2 Kuroshio intrusion

To better explore the corresponding physical processes of these stages, we first plotted the temperature-salinity (T-S) diagrams to determine the water characteristics during the study period. Figure 4 depicts the average T-S diagrams and Chl-a profiles for the three stages. During Stage 1, the average T-S diagram was close to that of Kuroshio water in the upper layer above 26 σ_θ, implying that this stage was affected by Kuroshio intrusion. The average depth of the SCM deepened to 87 m, and the SCM magnitude was as low as 0.47 mg/m³. The Kuroshio water is characterized by high temperature and salinity but low nutrients in the upper layer, thus leading to low primary production in the upper ocean (Du et al., 2013; Liu et al., 2020). The results observed by gliders were consistent with previous studies. In general, Kuroshio intrusion into the SCS is strong in winter and weak in summer in response to the seasonal reversing monsoon (Xu and Su, 1997; Nan et al., 2015). During the study period of 44 days, the glider observed two obvious high-salinity processes. Apart from Stage 1, a similar high-salinity disturbance occurred from August 30 to September 1 (Fig. 3d), and its biological response was also similar to Stage 1 (Fig. 2). The depth of SCM deepened and its magnitude decreased with the appearance of a subsurface high-salinity core. After that, the increase of Chl a may be affected by other dynamic processes. As seen from the satellite altimeter observations, there was no strong Kuroshio intrusion during the study period (Fig. 5). These results indicate that the high-salinity processes observed by gliders were only high-frequency Kuroshio disturbances near the Luzon Strait.

3.2.3 Effect of a cyclonic eddy

During Stage 2, the T-S curve showed the mixing characteristics of Kuroshio water and SCS water (Fig. 4a). Unlike the other stages, there was a significant phytoplankton bloom in the SCM layer from August 6 to August 16. The average depth of the SCM was uplifted to 67 m, and the magnitude of SCM was as high as 0.74 mg/m³ (Fig. 4b). Although there are many studies on the surface Chl-a blooms in the Luzon Strait, subsurface Chl-a blooms are still rarely reported. To explore the possible factors related to the subsurface Chl-a bloom, we examined the temporal evolution of the sea level anomaly (SLA) and geostrophic current during the study period (Fig. 5). From July 30 to August 14, there was a cyclonic eddy in the west of Luzon Strait. As the eddy developed, it slowly moved northward and eventually affected the area of glider observation. Cyclonic eddies are generally reported to have high primary production and phytoplankton biomass due to the vertical supply of nutrients caused by eddy pumping (McGillicuddy et al., 1998; Chen et al., 2007; He et al., 2016). In our study, elevated Chl a was also observed within the cyclonic eddy. Although the space-time scale of this cyclonic eddy observed by gliders was relatively small, the uplift of the thermocline and the increase of Chl a were significant (Figs. 2–3). The Chl-a bloom only occurred in the subsurface layer, which was probably related to the eddy intensity and the background stratification of the upper ocean, that is, weak cyclonic eddy and strong stratification in summer was unfavorable for the transport of deep nutrients to the surface layer. The SCS is rich in mesoscale eddy activity, which has a great effect on the phytoplankton productivity and biogeochemical cycle of the region. Satellite remote sensing can only detect the changes of Chl a on the sea surface but cannot identify the subsurface Chl-a bloom, thus may underestimate the new primary productivity caused by mesoscale eddies.

3.2.4 Summer rainfall and typhoon process

Compared with the previous two stages, the T-S curve in Stage 3 was closer to the northern SCS water. One of the distinct features was that there was a significant salinity decrease in the upper ocean. The salinity above 21.8 σ_θ was even lower than that of typical northern SCS water. The process of salinity desalination was further confirmed by satellite observations (Fig. 6). Satellite images showed that influenced by the summer southwest monsoon, a large area of heavy rainfall occurred in the east and north of the SCS during Stage 3. The variation and distribution of sea surface salinity (SSS) were highly consistent with precipitation, implying that the change of SSS at Stage 3 was mainly affected by freshwater forcing. During Stage 3, the average depth of the SCM was slightly increased to 75 m, and its magnitude was reduced to 0.50 mg/m³ (Fig. 4b). The depth of SCM layer was almost consistent with the isopycnal depth at 23 σ_θ throughout the study period, but it deviated largely during Stage 3. Previous studies have revealed that light and nutrients are the two major factors controlling the growth of phytoplankton (Letelier et al., 2004; Chen and Zhao, 2021). Xu et al. (2022) found that there was a positive linear relationship between the SCM depth and the euphotic layer depth in the SCS, and suggested that light may play an important role in the adjustment of SCM depth in summer. During the period of continuous rainfall, the light radiation was reduced on the sea surface. The growth of phytoplankton was limited by light, thus the SCM layer was maintained at a shallow depth in Stage 3. In addition, there was a positive SLA appeared in the south of Luzon Strait on August 9 (Fig. 5). As the high SLA signal expanded and moved northward, it gradually affected the area of glider observation. The positive SLA corresponded to the convergence of the upper low-salinity water, which deepened the MLD and the thermocline depth in Stage 3. Therefore, the depth deviation of the SCM and the 23 σ_θ isopycnal was mainly affected by light limitation and the isotherm displacement. Because of the deep MLD, the Chl a in the shallow SCM layer was easier to be entrained to the surface layer. Thus, the surface Chl a increased in Stage 3. However, the magnitude of the SCM at this stage decreased due to the limitation of nutrients.

In addition, it is worth noting that there was a peak rainfall in the study area on August 24, with the precipitation up to 64 mm/d. Further analysis showed that this was related to a typhoon process. Figure 7 shows the track of typhoon Bailu and its intensity. Bailu developed as a tropical depression in the western Pacific and strengthened to a severe tropical storm on August 23, 2019. The typhoon moved northwest to the Luzon Strait and entered the SCS on August 24, accompanied by strong winds and heavy rainfall. The maximum sustained wind (MSW) was up to 30 m/s and the minimum sea level pressure (MSLP) was 980 hPa on August 24. Many studies have reported that surface Chl a substantially increases during the passage of a typhoon (Babin et al., 2004; Pan et al., 2017; Lee et al., 2020). Fortunately, the Chl-a response to the typhoon Bailu was also observed by the glider 1000K012. The surface Chl a significantly increased on August 24, consistent with previous studies. However, both the depth-integrated Chl a and the SCM showed an insignificant increase during this period (Fig. 2b). Liu et al. (2013) pointed out that a deep thermocline is unfavorable for the increase of surface Chl a because the deep thermocline makes wind-driven nutrient pumping less efficient. Similarly, the thermocline in Luzon Strait was deep during Stage 3 (Fig. 2a). Thus the increase of surface Chl a may not be the effect of deep nutrient transport, but was more likely associated with entrainment of deep chlorophyll maxima caused by strong mixing.

4 DISCUSSION AND CONCLUSION

Luzon Strait is the main channel for water exchange between the SCS and the western Pacific, with complex dynamic processes and biological responses. Based on glider observations, this study investigated the temporal and vertical variations of Chl a in the Luzon Strait from July 25 to September 6, 2019. The vertical Chl a in the Luzon Strait exhibited a prominent SCM in summer, ranging from 50 m to 110 m, with a mean depth of 75 m. The Chl a in the SCM layer accounted for 38%–75% of total Chl a in the upper water column (0–200 m). The magnitude of the SCM was highly negative with the thermocline depth, implying that shallow (deep) thermocline is favorable (unfavorable) for the upward of nutrients and promotes (inhibits) the growth of phytoplankton. During the study period, the Chl a in the Luzon Strait experienced several significant changes. Combined with satellite data, the physical processes related to Chl-a changes were further analyzed.

Under the influence of Kuroshio intrusion, there was a high-salinity (> 34.7) core observed in the subsurface layer in Stage 1. Due to the low nutrients in the Kuroshio water, the average depth of the SCM deepened to 87 m, and the SCM magnitude was as low as 0.47 mg/m³. Apart from Stage 1, a similar high-salinity disturbance occurred from August 30 to September 1, and its biological response was also similar to that in Stage 1. Although there are few strong Kuroshio intrusions in summer, Luzon Strait is frequently affected by the high-frequency Kuroshio disturbances.

During Stage 2, the gliders passed through a cyclonic eddy where the Chl a was elevated in the SCM layer. The average depth of the SCM was uplifted to 67 m, and its magnitude was increased to 0.74 mg/m³. The cyclonic eddy uplifted the thermocline and pumped up the nutrient-rich deep water, promoting the subsurface Chl-a bloom. Different from the surface Chl-a bloom often reported in the Luzon Strait, this Chl-a bloom was only found in the subsurface layer, which was probably related to the weak intensity of the cyclonic eddy and the strong stratification of the upper ocean in summer. Satellite remote sensing can only detect the surface Chl a but cannot identify the subsurface Chl-a bloom, thus may underestimate the new primary productivity caused by mesoscale eddies.

Compared with the previous two stages, Stage 3 was mainly affected by heavy summer precipitation, which resulted in a significant salinity decrease in the upper ocean. The average depth of the SCM was slightly increased to 75 m, and its magnitude was reduced to 0.50 mg/m³. Except for Stage 3, the depth of SCM layer was almost consistent with the isopycnal depth at 23 σ_θ during the study period. The SCM layer was maintained at a shallow depth due to the light limitation, while the convergence of the upper low-salinity water deepened the MLD and the thermocline. Therefore, the Chl-a decrease in the SCM layer due to the limitation of nutrients, but increase in the surface layer due to the deepening of MLD. In particular, a typhoon passed through the Luzon Strait on August 24, and the Chl a above the upper 50 m was substantially increased. However, both the depth-integrated Chl a and the SCM showed an insignificant increase during this period. The surface Chl-a increase was more likely associated with entrainment of deep chlorophyll maxima caused by strong mixing, indicating that the typhoon process may simply alter the vertical redistribution of Chl a and have a limited effect on net ocean primary production. Similar results were also reported by Chai et al. (2021) and Qiu et al. (2021). Nevertheless, the responses of the upper ocean to typhoons are extremely complex, depending on many factors, including the typhoon intensity, its translation speed, and the pre-typhoon ocean state (Lin, 2012). Our analysis is only a case study, which cannot represent the whole picture of the biological response to the typhoon. To comprehensively quantify oceanic biological responses to a typhoon, more high-resolution observations and case studies are needed in the future.

5 DATA AVAILABILITY STATEMENT

All data generated and/or analyzed during this study are available from the authors upon reasonable request, except for the glider data, which require permission from the South China Sea Special Management Office.

6 ACKNOWLEDGMENT

The authors are grateful for the reviewers' careful review and constructive suggestions to improve the manuscript. The glider data were provided by the South China Sea Institute of Oceanology, Chinese Academy of Sciences. The WOA13 data were obtained from the Asia-Pacific Data Research Center (http://apdrc.soest.hawaii.edu/data). The surface chlorophyll data were downloaded from the NASA's OceanColor Web (https://oceancolor.gsfc.nasa.gov/l3/). The SLA and absolute geostrophic velocity data were provided by CMEMS (http://marine.copernicus.eu). Rainfall data were retrieved from the Tropical Rainfall Measuring Mission. SSS and 10-m wind data were provided by Remote Sensing Systems (http://www.remss.com/). The TC information was obtained from the China Meteorological Administration (CMA) best track dataset (tcdata.typhoon.org.cn).