Near-Inertial Oscillations Induced by Winter Monsoon Onset in the Southwest Taiwan Strait

Abstract

The near-inertial motion in ocean surface currents directly reflects the energy transported by wind towards the surface layer, playing an important role in climate regulation and energy balance. Previous studies have mainly focused on near inertial oscillations (NIOs) induced by tropical cyclones in the Taiwan Strait, with few reports on near inertial oscillations induced by monsoon onset. Using high-frequency radar observations, we detected an amplification of NIOs induced by the winter monsoon onset. While not as strong as NIOs induced by tropical cyclones, the near-inertial current (NIC) induced by winter monsoon onset in the Taiwan Strait has peak speeds reaching up to 5.2 cm/s and explaining up to 0.7% of non-tidal variance. This study presents observational results of NIOs during three monsoon onset events, and analyzes the impact of winds and temperature changes on NIOs. Temporal and spectral analysis reveals that the monsoon onset is the primary driver behind the formation of NIOs. Results indicate that near-inertial kinetic energy is relatively lower in shallower waters, such as the Taiwan Bank, compared to deeper regions. Furthermore, by integrating the air and sea surface temperature from reanalysis products, we have examined the abrupt changes in sea surface temperature (SST) before and after monsoon onset and their correlation with NIOs. The findings suggest that temperature falling favors the intensification of NICs during monsoon onset, and a lack of significant SST changes precludes the triggering of notable NICs. These insights enhance our understanding of the mechanisms driving NIOs and their roles in seawater mixing.

1. Introduction

Inertial oscillations are a prevalent form of motion in the ocean, facilitating the transfer of energy and momentum between seawater in different depth layers and regulating large-scale ocean circulation systems [1]. These oscillations play a pivotal role not only in physical processes, but also in the global energy balance [2], with far-reaching implications for climate systems, biogeochemical cycles, mixing and nutrient supply enhance [3], and weather pattern forecasts [4]. Therefore, ongoing observations and studies are essential for advancing our knowledge of NIOs.

Due to the influence of the Earth’s rotation, inertial oscillations can occur in various oceanic regions, spanning various depths within the ocean such as the mixed layer, shallow water, and deep-water regions. However, although inertial currents can be observed in oceans at different latitudes and depths, they are often superimposed on other water currents, gradually attenuating over time due to the dissipative effect of friction, typically diminishing to varying degrees after a few days. In the deep ocean, inertial motion is notably pronounced and easy to observe [5,6,7]. In nearshore areas, where currents are primarily dominated by tidal or wind-driven currents, inertial motion is not as prominent as in the open ocean, and observational studies on nearshore inertial oscillations are generally scheduled around occurrences of strong wind events [1,8]. Furthermore, these oscillations exhibit frequencies near the local inertial frequency, showing either a redshift or blueshift phenomenon influenced by background currents [9,10,11].

Previous studies have indicated that the energy of inertial oscillations in the ocean is mainly driven by non-stationary wind fields, such as tropical cyclones or strong wind events with wind direction rapidly changing [12,13,14]. These events serve as key triggers for inertial oscillations occurring from the sea surface down to the ocean interior [9,15]. For instance, ref. [16] observed inertial oscillations in the Venetian lagoon excited by the Bora strong wind event (20 m/s). Ref. [17] detected NICs induced by Typhoon Wayne passing through the Taiwan Strait using high-frequency (HF) radar currents. When the upper ocean layer is shallowly stratified, inertial oscillations are more readily excited by winds [18]. Ref. [19] observed the inertial waves induced by two tropical cyclones in the northwestern South China Sea using moored buoys, propagating gradually from the mixed layer to the seafloor.

Recent years have seen a plethora of observational studies focusing on ocean eddies or tropical cyclone-induced processes exciting NIOs [20,21], influenced by stratification and bottom topography [13,22,23,24], enriching our understanding of this oceanic phenomenon and offering valuable insights. Ocean eddies, typhoons, and monsoon onsets trigger different characteristics in terms of the mechanisms of formation, impacts, and durations of NIOs. Eddies are typically generated by instabilities in ocean currents, affecting localized marine areas; typhoons are intense cyclones driven by tropical ocean heat, with widespread impacts [25]; while monsoon onsets are associated with seasonal wind shifts, affecting large regions. All three phenomena significantly influence ocean dynamics and ecosystems.

The primary aim of this study is to analyze the characteristics of monsoon onset-induced NIOs by combining moored buoy data, HF radar observations, reanalysis products, and modeling. The onset of monsoons involves sudden changes in large-scale atmospheric circulation, influencing regional climate and weather patterns and sometimes leading to the generation of NIOs. Monsoon onset is typically characterized by alterations in atmospheric circulation patterns in the lower and upper troposphere. During monsoon onset, there are rapid and significant increases in wind speeds above the sea surface, which can initiate or enhance NIOs. The second aim of this study is to explore the possible dynamic mechanism of NIOs induced by monsoon onset. However, there is limited literature regarding NIOs induced by monsoon onset. Shu et al. [26] utilized ADCP to observe robust NIOs induced by the summer monsoon onset in the central South China Sea in 1998 and 1999, reaching current velocities of 25 cm/s, despite wind speeds (10 m/s) being considerably lower than typical tropical storm speeds. The study presented in this paper focuses on NICs in the southwestern Taiwan Strait region, induced by the onset of the winter monsoon.

To explore the dynamic mechanisms of near inertial oscillation triggered by monsoon onset, we investigate the spatiotemporal distribution characteristics of near-inertial oscillations in the Taiwan Strait, and we extracted near-inertial surface currents from HF radar currents and conducted a dynamical diagnosis using SST, wind, and other reanalysis data. We hope that our observational study will deepen the understanding of this oceanic process.

The remaining structure of this article is outlined as follows: Section 2 introduces the data and methods, Section 3 presents the results, Section 4 delves into the discussion, and Section 5 concludes the study.

2. Data and Method

2.1. Ocean Surface Current

The current datasets for this HF radar field experiment were gathered from two OSMAR071 radar systems deployed along the coast of Fujian Province, China. Data were collected between 29 January 2013 and 26 March 2013. Two radar stations are situated at Dongshan (23.6575°N, 117.4863°E) and Longhai (24.2674°N, 118.1353°E), as illustrated in Figure 1, showing their geographic coordinates and radar footprints. Each station covers a beam angle of 150°, and has a radial current angle resolution of 1.5°.

The transmitting antenna is a three-element Yagi–Uda antenna (see Figure 2a), while the receiving antenna consists a non-linear array of eight antennas (depicted in Figure 2b). The receiving antenna array is organized into two rows, with the first row containing six antennas and the second row containing two antennas. More information can be found in reference [27]. These HF radars at each station are capable of detecting radial currents moving towards or away from the radar site. Dongshan and Longhai stations are approximately located 90 km apart, with a maximum detection range of 200 km for each single-station radar. In the overlapping coverage area, each grid point can detect two different directions of radial current, enabling the calculation of vector currents at these points. To ensure the reliability of the ocean currents obtained, the authors conducted data quality control measures and removed any corrupted data. For a detailed explanation of the methodologies used, please refer to [28].

A dataset spanning 57 days of vector current data was obtained with a spatial resolution of 5 km and a temporal resolution of 10 min. The radial current and vector data from the experiment were validated by comparing them with ADCP and buoy data. For detailed validation results, readers can refer to Wei’s study [29]. The high level of agreement found confirms the accuracy and quality of the radar data, establishing consistency that ensures the reliability of subsequent analyses on tidal, residual, and near-inertial currents characteristics.

2.2. Local Wind

The Taiwan Strait showcases characteristics typical of an East Asian monsoonal region, with distinct seasonal variations in wind patterns. During winter, northeasterlies are predominant, while southwestern or southerly winds dominate in summer [30,31]. The presence of elevated terrain and high mountains on both sides of the strait has resulted in the formation of a classic trumpet-shaped region, where the narrow pipe effect accelerates and converges winds significantly.

During the winter season (December to February), the monsoonal winds peak in speed, averaging between 12 and 14 m/s in the central Taiwan Strait. In the southwestern region of the Taiwan Strait, the average wind speed is slightly lower than that in the central region. A mooring buoy located at 23.46°N, 118.33°E, as marked in Figure 1a, recorded surface wind data every thirty minutes at a height of ten meters above sea level. The margin of error for wind speed was 5%, and for wind direction was 10°. Examination of the wind rose in Figure 2c indicates prevailing northeasterly winds during the experimental period. These winds fluctuated between 0 and 18.5 m/s, and were predominantly coming from the northeast at angles of 22.5° and 45° (with 0° representing true north).

To investigate the spatial characteristics of NIOs across the entire radar footprint, the authors integrated Cross−Calibrated Multi−Platform (CCMP) surface wind data within the radar coverage as a supplementary validation [32]. To assess the accuracy of CCMP wind data within the radar coverage area, the authors initially performed cubic interpolations on the wind field time series at the buoy location, generating an hourly time series with a resolution of 1 h. By averaging the buoy wind field to align with the 1-h resolution time series, we calculated the Root Mean Squared Error (RMSE) and correlation coefficients with respect to the buoy wind field time series, as depicted in Figure 2d,e, where the RMSE and correlation coefficients were 0.93, 1.35 m/s (east component), and 0.94, 1.72 m/s (north component), respectively, indicating the favorable accuracy of CCMP within the radar’s coverage during this experiment. Based on good agreement with the buoy winds, the CCMP winds over areas lacking measured winds could be used to compared with HF radar currents.

Throughout the winter experiment period in 2013, the dominant northeasterly winds were often accompanied by brief periods of high wind events, particularly on 18–19 February, 1–3 March, and 10–13 March (highlighted as elliptical circles in Figure 2).

2.3. Air and Sea Surface Temperature

To investigate the impact of various hydrological and geographical factors on the spatial distribution of ocean surface currents, the author utilized global oceanographic reanalysis data GLORYS12V1 from the Copernicus Marine Environment Monitoring Service (CMEMS). This dataset provides a spatial resolution of 1/12° and a temporal resolution of 24 h for surface sea water temperature. Additionally, the author obtained atmospheric temperature data above the sea surface from ECMWF’s ERA5 reanalysis data for temporal analyses related to surface sea water temperature and wind speed.

While these datasets are derived from global reanalysis data, their accuracy in coastal regions still needs to be confirmed. However, some comparative studies in the literature indicate that GLORYS12V1 and ERA5 are considered the most optimal and suitable choice among different global reanalysis products for analyzing SST in a marine environment [33,34].

2.4. Complex Demodulation

NIOs are commonly observed in global oceans, including both open sea regions and coastal straits. These oscillations are more pronounced in the deep ocean due to weak tidal currents and wind-induced currents. However, in coastal areas affected by local wind patterns, continental shelf slopes, and tidal effects, studying inertial oscillations proves to be more challenging. Given the strong tidal activity in the Taiwan Strait region, the initial approach involves using the T_TIDE tool to perform harmonic analysis on the vector sea currents at each spatial point to determine the tidal currents [35]. By subtracting these tidal currents from the vector currents, residual currents are obtained, from which the inertial currents are then extracted.

Traditionally, NICs are extracted using band-pass filters to isolate the inertial component. However, the accuracy of this method is closely tied to the filter’s performance. In order to mitigate the impact of filter performance, this study adopts the complex demodulation spectrum shifting, which converts the band-pass filter design into a low-pass filter.

The first step involves expressing the residual currents as $u\left(t\right)+iv\left(t\right)$. Subsequently, multiplying by $e^{i\omega t}$ induces spectral shifting, followed by low-pass filtering to isolate the inertial oscillations component, as demonstrated in Equation (1):

$$U_{NIOs}=\left|LPF\{[u\left(t\right)+iv\left(t\right)]·e^{i\omega t}\}\right|$$

(1)

where $\omega$ is the angular frequency, while u and v denote the eastward and northward components of the sea currents, respectively. The study employs a Lanczos filter with a low-pass filtering range set at 0.1 cycles per day.

The near-inertial kinetic energy (NIKE) of surface sea currents is quantified using the formula below, expressed in units of J/m³:

$$NIK{E}_{surface}={\displaystyle \frac{1}{2}}{\rho }_w(u_f^2+v_f^2)$$

(2)

where ${\rho }_w$ signifies seawater density (1024 kg/m³), while $u_f$ and $v_f$ represent the near-inertial meridional and zonal surface ocean currents, respectively.

2.5. Slab Simulation

In this study, the influence of local wind fields on inertial oscillations was assessed by referencing the methodology of [36] to estimate NICs in a slab model. Considering factors such as the Coriolis force, ocean surface wind stress, and given parameters such as Mixed Layer Depth (MLD), density ($\rho$), inertial frequency (f), and wind stress components ${\tau }_s=({\tau }_{sx},{\tau }_{sy})$, a dynamics model for momentum equations in the alongshore and offshore horizontal directions is expressed as follows:

$${\displaystyle \frac{\partial u_f}{\partial t}}-f{v}_f={\displaystyle \frac{{\tau }_{sx}}{{\rho }_{w}H}}-r{u}_f$$

(3)

$${\displaystyle \frac{\partial v_f}{\partial t}}+f{u}_f={\displaystyle \frac{{\tau }_{sy}}{{\rho }_{w}H}}-r{v}_f$$

(4)

Here, f represents the Coriolis parameter at the spatial point’s latitude, the subscripts x and y represent alongshore and offshore directions, H is the ocean depth, ${\rho }_w$ denotes seawater density, and ${\tau }_{sx}$ arises from wind stress affecting the surface pressure of the ocean, acting horizontally on the sea surface per unit area. The subscript s represents the sea surface, and the subscript x corresponds to the x-axis. We utilize the formula proposed by [37] for calculating wind stress,

$${\tau }_s={\rho }_{air}{C}_d\left|U_{wind}\right|U_{wind}$$

(5)

This equation parameterizes wind stress as a function of wind speed at a certain height above the sea surface. Here, a sea surface height of 10 m is chosen for the CCMP wind field data, where ${\rho }_{air}$ is the atmospheric density at 1.22 kg/m³, $U_{wind}$ is the wind speed at a height of 10 m above sea level, and $C_d$ is a dimensionless drag coefficient related to the state of airflow over the sea surface, representing the transfer of wind momentum to the ocean surface through wave drag [37].

$$C_d\times 1000=\left\{\begin{array}{cc}1.2,\hfill & |U_{wind}|\le 11,\hfill \\ 0.49+0.065|U_{wind}|,\hfill & 11<|U_{wind}|\le 19,\hfill \\ 1.364+0.0234|U_{wind}|-0.00023158|U_{wind}{|}^2,\hfill & 19<|U_{wind}|<100.\hfill \end{array}\right.$$

(6)

The parameter r in Equations (3) and (4) is set as a fixed drag coefficient (also known as an empirical damping coefficient) with units of s⁻¹. It is an inverse relationship describing fluid characteristic decay time (slightly less than one day). Alford suggesting the optimal range to be $0.1f$ to $0.3f$ [2]. In this study, r is taken as $1.43\times {10}^{-5}$ [28]. The NICs in the alongshore and offshore directions in the slab model can be numerically solved using a fourth-order Runge–Kutta method, with a time step of 1 h.

The energy flux $\mathrm{\Pi }$ due to the work of the wind on the NIC in the upper ocean layer is calculated as

$$\mathrm{\Pi }={\tau }_s·V_f$$

(7)

where $V_f$ represents the horizontal inertial current at the sea surface. A positive value indicates energy transfer from the wind to the near-inertial motion in the ocean mixed layer. Integrating $\mathrm{\Pi }$ over time can be used to represent the cumulative energy input from the wind into near-inertial motion, given by

$$E=\int \mathrm{\Pi }dt$$

(8)

3. Results

3.1. Characteristics of NIOs Induced by Monsoon Onset

During the 2013 experimental period, the East Asian Northeast Monsoon was dominant, often accompanied by transient high wind events known as monsoon onset. These events occurred from 17 to 21 February, 28 February to 4 March, and 10 to 14 March. The three ellipses in Figure 3 denote these monsoon onset events. In Figure 3a, wind speed and residual current components at point A along the strait are shown, while Figure 3b displays the 850 hPa and 950 hPa air temperatures at point A, along with the SST. It is observed that, during monsoon onset, wind speeds increase rapidly, with maximum speeds along the strait reaching up to 18.5 m/s. At the same time, there is a significant drop in atmospheric temperature, decreasing from 20 °C to 7 °C. A consistent trend is noticed in wind speed, residual current velocity, and atmospheric temperature above the sea surface during monsoon onset. However, despite these changes, the surface oceans layer, acting as a large heat reservoir, did not cool at the same rate. When onset 1 and onset 2 occur, cold air quickly moved over the sea surface, leading to a rapid 5-degree Celsius decrease in SST. In contrast, no significant change in SST is observed during onset 3.

Power spectral analysis can be employed to examine the presence of significant NIOs within ocean currents. Upon conducting rotated power spectral analysis of the residual currents, no clear spectral peak is observed near the near-inertial frequency. This absence is speculated to arise from the relatively weak near-inertial energy induced by monsoon onset, which are transient in nature and thus not sustained over extended periods. Additionally, the near-inertial energy peak is likely smoothed out in long-term power spectral calculations. As a result, temporal segments pre- and post-monsoon onset are extracted from the residual currents and analyzed using rotated power spectral analysis, leading to the results shown in Figure 4.

For a typical monsoon onset event at point A, the clockwise rotation power spectral analysis of surface residual currents pre—(9–18 February) and post−monsoon onset (19–28 February) in the vicinity of the near-inertial frequency reveals an increase in energy spectral density post-monsoon onset.

The study findings indicate that prior to the onset of the monsoon, there is minimal motion of surface sea water in the near-inertial band, with no distinct energy spectral peaks. After the monsoon onset begins, although surface oceanic motion continues to be significantly influenced by winds, there is a noticeable increase in energy spectral density in the near-inertial band, almost doubling the energy levels from before the monsoon onset (see Figure 4a). However, the energy associated with the NIOs remains relatively weak, suggesting they do not play a dominant role.

The NICs extracted from HF radar observations at point A, along with wind curl, NICs simulated by the Slab model, and the inertial energy input from the wind to the surface layer currents of the ocean are shown in Figure 5. Three monsoon onset events occurred during the observational period. During onset 1, a significant increase was observed in near-inertial energy in the surface layer due to strong winds, with near-inertial energy levels starting at approximately 0.2 J/m³ before the monsoon onset. As the monsoon onset progressed, near-inertial energy gradually increased, peaking around 19 February at approximately 1.3 J/m³ with energy input from the wind to the ocean surface. Additionally, the extracted NICs displayed a coherent resonant near-inertial mode in both along-strait and cross-strait directions, as depicted in Figure 5b,c.

The red trajectories in Figure 5e depict the progressive vector diagrams (PVD) of the NICs during onset 1. A preliminary calculation suggests a southward displacement of around 1 km over approximately three periods within the monsoon onset period (19–23 February).

To assess the influence of wind forcing during the monsoon onset on inducing NICs, hourly wind field data were employed. For simplicity, a constant depth of 20 m was utilized in the model configuration as the upper mixed layer depth. A comparison between radar observations and model outputs of the NICs is presented in Figure 5b,c, where the blue curves represent radar observations and the magenta curves denote the results from the Slab model. The results indicate a prompt response of the model output to the wind events, particularly evident during onset 1, showcasing a strong alignment between the two. This suggests that wind forcing predominantly drives the initiation of NICs despite its brief duration. However, in the subsequent two monsoon onset events, the trends between radar observations and model outputs did not align as closely as during the first event, signifying the influence of other factors during these latter monsoon onset events.

3.2. Spatial Distribution of NIKE

Figure 6a illustrates the spatial distribution of the average NIKE throughout the entire experimental period. It is discernible that under the influence of the northeast monsoon conditions, the NIKE values in the central region of the radar footprint are relatively small (0.1 J/m³), contrasting with larger NIKE values observed in the West Waterway near the DS radar station and the eastern region of the Taiwan Bank. The energy associated with near-inertial currents within the surface currents detected by the radar in the study area is comparatively modest. Periodic near-inertial motions occur around the onset of monsoonal bursts when wind directions undergo sharp changes; however, these circular motions are transient and quickly diminish due to the combined effects of background currents and friction.

Figure 6b illustrates the spatial distribution of the proportion of average near-inertial kinetic energy to the total vector mean kinetic energy. The data show that, over the course of the experiment, the proportion of near-inertial kinetic energy decreases significantly, especially in the Taiwan Bank. This decrease in the share of near-inertial kinetic energy may be influenced by the amplification of semidiurnal tidal currents in the Taiwan Bank, which increases the overall magnitude of the vector mean ocean currents and results in a smaller proportion of average near-inertial kinetic energy compared to the total kinetic energy.

To further analyze the NIKE characteristics, we calculated the spatial distribution of the average NIKE and its proportion relative to the total current kinetic energy during three monsoon onset periods. In onset 1, the highest average NIKE values were mainly concentrated in the northern region and along the Taiwan Bank, peaking at 0.45 J/m³ (Figure 7). Moreover, it is evident that the average NIKE during onset 1 surpasses that of onset 2 and onset 3, particularly in the northern part of the study area (Figure 8 and Figure 9).

During subsequent occurrences of the monsoon onsets, the regions characterized by significant values of average NIKE were sparse, primarily concentrated in the northeastern sector of the study area. Additionally, the large values of average NIKE at the edges of the study area exhibited a lack of continuity and coherence, appearing disorganized and lacking discernible patterns.

3.3. Variation in Wind and SST

To further explore the impact of wind fields on triggering NIOs, the spatial distribution of wind field variations during three monsoon onset periods was computed Figure 10. The formula used to calculate the changes in wind fields over three onset periods follows the methodology outlined by [26].

$$Variation=\sqrt{{\displaystyle \frac{1}{n}}\sum _1^n\left[{(u-\overline{u})}^2+{(v-\overline{v})}^2\right]}$$

(9)

where u and v represent the east and north components of wind velocities, while $\overline{u}$ and $\overline{v}$ denote the average wind velocities during the respective period. Notably, the Xiamen–Penghu Trough emerges as the primary region with the highest Variation values, reaching a maximum of 8.5 m/s. Comparatively, during the third onset period, wind speed and direction exhibited the most significant changes, with peak values over 7.5 m/s distributed extensively across the entire study area.

The spatial distribution map of the average flux $\mathrm{\Pi }$ of windwork on the NICs of the ocean surface layer throughout the entire experimental period is presented in Figure 11. Values exceeding 50 mW m⁻² predominantly concentrate in the central and southern regions of the studied area, notably along the northern edge of the Taiwan Bank. This observation indicates a heightened and more efficient energy transfer in the central sea area towards the surface currents in the near-inertial frequency band.

During the onset of the monsoon for the third time, the spatial distribution of the accumulated energy input E of the winds approaching near-inertial motion is illustrated in Figure 12. Noteworthy is that E represents the cumulative integral of energy input over time. In the investigation of NIOs induced by tropical cyclones, when calculating the value of E, wind directions typically exhibit systematic and frequent variations, hence NIOs generally manifest when E assumes larger values. In contrast to NIOs induced by tropical cyclones, this study primarily focuses on NIOs triggered by monsoonal onset. During monsoonal onset, wind speeds significantly escalate, yet wind directions do not undergo frequent changes. This implies that if NICs induced by monsoonal onset exist, the value of E would be relatively diminished. Consequently, under the influence of quasi-steady winds, the integral of work performed by the near-inertial currents in circular motion tends towards zero. In essence, regions exhibiting large amplitudes of E do not harbor induced NIOs. Conversely, areas where E approaches zero might either witness NIOs or remain devoid of such phenomena.

From Figure 12a,b of first two onsets of winds, it is apparent that the cumulative energy input fluctuations towards near-inertial motion are considerable, indicating a significant amount of work performed by the winds on the near-inertial motion (wind work can be both positive and negative). The intensity of NIOs in the surface ocean induced by monsoonal onset is more pronounced than that of the third onset. Particularly, there are substantial differences in the locations of low E values during the first and second onsets, with the signs of work performed by the NICs in the same areas being opposite for these two onsets.

From the commencement of the three monsoonal onsets to their passing, corresponding alterations occur in the SST. Taking the third onset as an example, compared to the first two instances, the values of E during the third onset generally lean towards lower magnitudes. Considering the significant wind variation during the third onset as shown in Figure 10, it is unclear whether there was near-inertial current present across the entire spatial extent. Consequently, a modification was made to Equation (8) as follows:

$$E_{amp}=\int \left|\mathrm{\Pi }\right|dt$$

(10)

This equation computes the integrated amplitude values of the work performed by the winds on the near-inertial motion at each time point. As illustrated in Figure 13c, during the third onset, most regions exhibit relatively small values of $E_{\mathrm{amp}}$ below 3 KJ/m², indicating the absence of distinct NIOs triggered by onset 3.

The variations in SST during the monsoon onset were examined in this study. The SST data are obtained from GLORYS12V1. Figure 14b–k depict the spatial distribution of SST from 14–23 February. In Figure 14a, sea water temperatures at different depths over the ten-day period are shown; each vertical section displays temperature variations with depth and longitude (as shown by the solid black line cross section in Figure 14b) at a fixed latitude of 23.8° (corresponding to point A), the black curves in Figure 14a represent the 20° isotherm. The analysis reveals that, starting from 17 February, the thermodynamic characteristics of the ocean surface are impacted by the monsoon onset. Increased winds lead to enhanced heat exchange between the ocean and atmosphere, causing a reduction in sea water temperature. At point A, temperatures decrease significantly in the sea surface by up to 5 °C, with this cooling effect extending to depths of 30 m. These findings suggest that the influence of the monsoon onset on sea water temperatures extends beyond the surface layer and may reach deeper regions.

4. Discussion

Utilizing complex demodulation spectral shifting, the study extracts NICs in the ocean surface layer. By scrutinizing the temporal characteristics of these NICs, it was revealed that the onset of the East Asian winter monsoon passing through the Taiwan Strait serves as a trigger for near-inertial motion in the ocean.

The study also investigates the spatial distribution of NIKE and energy flux from wind to inertial motions, along with variations in SST during monsoon onset events. The spatial distribution of NIKE indicates that areas with shallow water depths, such as the Taiwan Bank, exhibit lower near-inertial energy compared to regions with deeper waters.

Near-inertial currents (NICs) in the Taiwan Strait are exhibiting different characteristics compared to NICs in other oceans. NICs studied in this research are typically associated with the winter monsoon onsets, whereas in other regions like the South China Sea, NICs may be influenced by typhoon activities or tropical cyclones. For instance, NICs in the South China Sea can be significantly enhanced by typhoons, reaching speeds of up to 1.2 m/s, with kinetic energy peaking in the mixed layer, surpassing even three times the internal tidal energy [38]. Furthermore, the intensity of NICs in the South China Sea is positively correlated with typhoon intensity, including factors such as maximum wind speed, maximum wind radius, and movement speed, showing significant vertical and horizontal propagation characteristics [39]. In contrast, NICs in the Taiwan Strait may be influenced by different environmental factors, such as rapidly changing air temperature and SST.

Three typical monsoon onset events over a two-month period were selected for analysis to calculate kinetic energy and variations in NICs induced by these events. The results indicate that the velocity of NICs induced by monsoon onset can reach up to 5.2 cm/s, slower than those induced by typhoons. However, a significant enhancement in the power spectral peak in the near-inertial band is observed from the onset to the passage of the monsoon, as depicted in Figure 4.

To investigate the energy enhancement of near-inertial motions before and after monsoon onset, a formula was employed to calculate the degree of enhancement in power spectral energy within the near-inertial band:

$$R={\displaystyle \frac{\sum PS{D}_{after}}{\sum PS{D}_{before}}}$$

(11)

where $\sum PS{D}_{after}$ denotes the sum of power spectral densities near the near-inertial band four days after the monsoon onset, while $\sum PS{D}_{before}$ represents the sum of power spectral densities near the near-inertial band four days before the monsoon onset. A value of R greater than 1 signifies that the monsoon onset results in increased energy of near-inertial motions. Figure 15 illustrates the spatial distribution characteristics of R for the three onset events, with the first onset event showing enhanced power spectral areas primarily in the central region of the study area. From a spectral analysis perspective, the second and third onset events do not exhibit a stable and distinct enhancement in energy of near-inertial motions before and after the monsoon onset.

Due to the lack of buoy data during the experiment period, it was not possible to study the vertical propagation of NIOs by analyzing ocean currents at different depth layers. However, the extensive spatial coverage of HF radar synchronous observations allows us to examine the spatial characteristics of near-inertial surface ocean currents in the horizontal plane. Figure 16e–h depicts the spatial distribution of NICs with a temporal interval of around 31 h, which is the median value of the inertial period in the radar footprint. A clear wave propagation signal towards the south is visible, where the lines of constant phase represent the crests and troughs of the waves. Particularly, during the onset 1 and passage from 19–23 February (from Figure 16e–h), the spatial distribution of their phase shows significant similarity.

Figure 17 illustrates the spatial distribution of meridional and zonal NICs at each inertial period from 19–23 February, with a maximum value of approximately 5.2 cm/s. The main energy is found in the horizontal velocities, with minimal impact on surface level changes. The highest values of meridional NICs are mainly located north of the Taiwan Bank, while the peak values of zonal NICs are primarily found at the northern edge of the Taiwan Bank, showing cyclic variations over a continuous four-day period during near-inertial motion.

It is important to note that not every monsoon onset leads to significant NICs in the surface ocean layer. As can be seen in Figure 5b,c, only the first onset event triggers a stable NIC. During onset 1, the currents from Slab model matched particularly well with the extracted NICs from HF radar. Unlike in onset 1, during onset 2 and onset 3, additional influencing factors led to significant discrepancies in the model estimates. This indicates that the Slab model, which only considers dynamical characteristics (as per Equations (3) and (4)), has limited accuracy in predicting results during onset 2 and onset 3. Incorporating more factors into the model will be one of the future work. Advanced radar technologies [40] could provide SST and wind field data with higher temporal and spatial accuracy, which is helpful to solve this problem. The findings from the PVD support this observation (Figure 5e). Figure 3a illustrates residual currents and wind speeds at point A along the strait. The patterns of residual currents at these points closely resemble the variations in the wind field, suggesting that the wind field primarily influences the characteristics of the background detided current during the monsoonal period.

Furthermore, this study integrates wind field data and SST from reanalysis products to examine the abrupt changes in SST around the onset of monsoons and their relation to pre-inertial oscillations. During a vigorous wind episode documented in Zhang’s research, the near-surface temperature in the Ross Sea dropped from approximately 2 °C to −1 °C, with this temperature decline observed from the ocean surface down to a depth of around 50 m [11]. Zhang’s analysis focused on the sudden changes in SST during the period of near-inertial motion induced by the strong wind event. Our results similarly indicate that as obviously NICs are induced, there is a noticeable decrease in seawater temperature.

This reduction in SST could also be attributed to the eastward displacement of the cold tongue in the West Waterway driven by monsoon wind stress vortices, a deflection mechanism discussed in studies by Shen and others [41,42]. During this process, the continuous eastward movement of the front formed by the interaction of cold and warm water masses raises the mixed layer depth, facilitating the excitation of NIOs and vertical mixing, consequently enhancing the more significant drop in temperature within the mixed layer. However, during the second and third onsets, there was no significant variation in surface seawater temperature, resulting in the absence of clearly stable NICs triggered by the monsoonal onset. Furthermore, Figure 3 shows a close link between the wind and background subtidal currents. The abruptly decrease in SST can enhance heat exchange between the ocean and the atmosphere, thereby affecting the intensity and direction of the wind. The altered wind field can lead to changes in surface currents, potentially fostering the local formation of near-inertial oscillations.

Due to the influence of filter performance, tidal and residual current signals may persist in the extracted NICs. Therefore, over the entire experimental period, due to the interplay of background current and friction, NICs extracted at each spatial point might not exhibit entirely regular circular motions. As a result, only the more prominent NIOs may have been identified. Nonetheless, the study’s results still indicate that the monsoonal onset and the decrease in surface seawater temperature (associated with seabed topography or upper-layer depth of the mixed layer) are the primary factors responsible for generating NIOs.

The spatial distribution of NIKE and its proportion in detided current can be observed from the results of the three onset events (Figure 6, Figure 7, Figure 8 and Figure 9), all of which show NIKE being small in the Taiwan Bank. The subtidal current is affected by both the monsoon and the Taiwan Bank, especially in the northern edge area of the Taiwan Bank [43]. However, the contention here is that this phenomenon does not solely stem from water depth determining the spatial characteristics of NIKE. NIKE could likely be related to wind direction, MLD, among other factors. The persistent northeast monsoonal winds continuously transfer energy to the ocean surface, elevating sea surface height and, in this process, triggering the NIOs. These oscillations tend to occur more frequently in the Xiamen–Penghu Depression, where a more suitable MLD exists.

Previous research findings indicate that, under strong wind conditions, regions with shallower Mixed Layer Depths (MLD) are more prone to the generation of NIOs (Shrira, 2015). By integrating the insights from Figure 12 and Figure 13, it can be inferred that during the initial onset, NICs induced by monsoon onset are more prevalent in the Xiamen–Penghu Depression (located north of the Taiwan Bank) compared to the Taiwan Bank where these currents are less pronounced.

This discrepancy can be attributed to several factors. Firstly, the amplified background tidal currents in the Taiwan Bank do not support the significant generation of NIOs [44,45]. Secondly, the shallower depths of the Bank impede the sustained rise in sea surface height, causing a portion of wind input energy to dissipate or redirect towards balancing with the geostrophic current [43]. Thirdly, the Taiwan Bank exhibits shallower depths, thereby failing to establish stable oceanic stratification during the winter monsoon period [46,47].

5. Conclusions

Utilizing HF radar observations of ocean surface currents in the southwestern Taiwan Strait, a dynamic analysis of near-inertial motions was conducted. The study integrated the Slab model with data on wind fields, SST, and other variables. Through examining the temporal and spatial characteristics of NICs, it was observed that the winter monsoon in East Asia intensifies near-inertial motions significantly as it traverses the Taiwan Strait, particularly during the onset of the monsoon. The maximal NIC is with speeds reaching up to 5.2 cm/s.

The study examined the spatial distribution of NIKE and the energy flux from wind to inertial motions, as well as variations in SST during a monsoon onset event. Findings showed that shallow regions like the Taiwan Bank had lower levels of near-inertial energy compared to deeper areas, with significant NICs mainly observed in the Xiamen–Penghu Depression located to the north of the Taiwan Bank. The maximum NIKE can reach 1.45 J/m³, and the average maximum Nike can reach 0.45 J/m³ during the set 1 period. Different monsoon onset events led to varying intensities of NIOs, indicating that rapid changes in wind fields were not the only factor influencing near-inertial energy. The study also found that a decrease in SST enhanced NICs, and the lack of significant SST variations during monsoon onset prevented the generation of NICs. Although the two-month observational dataset had limitations, future research is likely to focus on long-term observations of monsoon-induced NIOs.