Diapycnal mixing plays an important role in modifying water mass, transporting heat and maintaining ocean stratification. Understanding its spatial and temporal variations is key to the improvement of numerical model representation of large-scale ocean circulation and climate changes. Away from boundaries, diapycnal mixing occurs mainly through internal wave breaking. Winds and tides input a primary energy into the internal wave field (Wunsch and Ferrari, 2004). The heterogeneity of the energy sources leads to pronounced spatial variability of diapycnal mixing, which has been clearly revealed by measurements in the past three decades (e.g., Gregg, 1987; Kunze et al., 2006; Wu et al., 2011; Whalen et al., 2012; Waterhouse et al., 2014). An enhanced diapycnal diffusivity is found over the rough topography due to interactions of the bottom geostrophic or tidal flow with topography and under the storm tracks due to an energetic near-inertial energy flux into the ocean.

Strong internal tides are generated over the rough topography around the Luzon Strait and propagate westwards into the South China Sea (SCS) (Alford et al., 2015). Previous measurements along 21°N (Tian et al., 2009) revealed the enhanced diapycnal mixing in the deep ocean of the SCS and attributed it to the dissipation of energetic internal tides generated around the Luzon Strait. However, it remains unclear what are the energy sources responsible for the diapycnal mixing in the region of the SCS to the south of the Luzon Strait where the internal energy flux is much weaker (Alford et al., 2015). In particular, the wind-generated near-inertial internal waves may play a significant role in furnishing diapycnal mixing there, which remains poorly assessed.

In this paper, we analyze the variation of turbulent diapycnal mixing to the south of Luzon Strait and its controlling factors based on the CTD profiles collected along 18°N by the Key Laboratory of Tropical Marine Environment Dynamics, the South China Institute of Oceanology, the Chinese Academy of Sciences. The diapycnal diffusivity is inferred from a fine-scale parameterization following Kunze et al. (2006). The paper is organized as follows. Methodology and data are introduced in Section 2. Results and analysis are shown on Section 3. Finally, summary and some discussion are given in Section 4.

We first evaluated the time-mean diapycnal diffusivity K based on the CTD profiles via fine-scale parameterization method. There were two identified features. First, the diffusivity near the bottom is redder than that in upper column. In some longitude, the diffusivity near the bottom, which could reach 10^–2 m²/s, is two or three order of magnitude stronger than that in the upper column (e.g., around 18°N, 113°E, 119°E and 119.5°E, see Fig. 2). The diffusivity averaged over all segments could reach the magnitude of 10^–4 m²/s in the SCS (Fig. 3), which was enhanced compared with that in the open ocean (Kunze et al., 2006). In the upper ocean, the vertical averaged (first four segments) diffusivity was an order magnitude less than the full depth averaged diffusivity. A temporal-vertical averaged (all segments) diffusivity is various with the topography (Fig. 3). It is found that the diffusivity can exceed to 10^–4 m²/s and even reached 10^–2 m²/s near the rough topography (Fig. 3), which is probably due to the dissipation of internal tides from barotropic flow (Egbert and Ray, 2001; Wunsch and Ferrari, 2004; Tian et al., 2009). Second, over flat abyssal basins, where the internal energy flux is much weaker, the mean diffusivity of upper four segments can over 4.5×10^–5 m²/s (Fig. 2), which is stronger than that in the outside Pacific (10^–6–10^–5 m²/s).

In order to focus on the variation of diapycnal turbulence diffusivity, the regional averaged diffusivity based on the roughness of each segment in the upper ocean was analyzed respectively (Figs 4 and 5). We selected 114.5°–118.5°E as smooth region and 119°–120°E as rough region (Fig. 3). For the flat region, we sieved data from 2006 to 2010. While in the rough region, only the year of 2005, 2006, 2007 and 2010 could be selected. Missing year due to the lack of data in the selected area. All of the three segments (256–512, 513–768 and 769 –1 024) in the flat region had an increasing trend, even though it varied in each year (Fig. 4). The features in the rough region were more remarkable. All of four upper segments shown an increasing trend in the first 3 a despite the diffusivity of all four segments slight decrease in 2010. It could be identified that the turbulent diapycnal diffusivity has enhanced from 2005 to 2010 in this area (Fig. 5). So both regions had increased in the diapycnal diffusivity in the upper ocean from 2006 to 2010.

Wind-work input great amount of energy on ocean inertial motions (Gill et al., 1974; Alford, 2001), which can affect local diapycnal mixing (Jing et al., 2011; Jing and Wu, 2014; Li and Xu, 2014). Jing and Wu (2010) had found a significant correlation between the diapycnal diffusivity and the wind stress, meanwhile wind-induced downward-propagating near-inertial wave can input the energy for mixing (Jing et al., 2011; Jing and Wu, 2014; Li and Xu, 2014). So we compared the regional averaged diffusivity with flux from a local wind to inertial motions. We used 10 d averaged wind-induced near-inertial energy flux leading the observation of the CTD profile as the local wind flux. Like the variation of diffusivity, the wind flux was various every year, but a remarkable increasing trend could be identified in both regions (Figs 4 and 5). Yang and Wu (2012) find the energy transferred from wind to SCS has enhanced from 1959 to 2008. However, the variation of the spatial averaged diffusivities of flat and rough regions varied with depth (Fig. 6), this might contribute to the joint effort of the topography and the barotropic flow.

Comparing flat region with rough region, we can also find that topography play a considerable role for the turbulent diapycnal diffusivity (Figs 4 and 5). In the flat region, the maximum wind flux can nearly reach 0.8 mW/m², which was 50% larger than that in the rough region. However, comparing to diffusivity in the rough region, it was 30% to 50% less in the flat region. This resulted from the enhanced mixing around the topography (Tian et al., 2009; Wu et al., 2011; Jing and Wu, 2014). As shown in Fig. 4, with the propagating of the near-inertial wave, the energy can be dissipated by turbulence. While this process only appeared in the first two segments around the rough region (Figs 5 and 6). The diapycnal turbulence diffusivity decreased with depth in the flat region, contrasted with the rough region, where the diffusivity increased below 700 m. In flat region (Fig. 3), away from seabed (deeper than 2 600 m), the wind-stir played an important role in diapycnal mixing. However, in the rough region, the diapycnal mixing can be affected by both wind stress and internal tide. Even though the forth segment is away from seabed (more than 2 000 m, Fig. 6), the rapid change of topography in the east of selected rough region (Figs 2 and 3) may enhance the diapycnal mixing.

The fine-scale parameterization method was used to estimate the turbulent diapycnal diffusivity in the SCS along 18°N. We examined the influence of the wind-induced near-inertial energy flux and topographic roughness on diffusivity. It should be noted that, it was the first time discovering the enhanced turbulent diapycnal mixing trend resulting from the growth of the wind-induced near-inertial energy.

The temporal averaged diffusivity varied with depth. In the region far away from the bottom, the magnitude of the diapycnal diffusivity was 10^–5 m²/s, which was an order of magnitude less than the vertical-averaged diffusivity.

The enhanced turbulent diapycnal mixing trend had been found in the upper ocean at 18°N based on the CTD from 2005 to 2010, which resulted from the growth of energy from wind to near-inertial motions.

The total tidal dissipation rate is about 0.7–0.9 TW, which is one of the primary energy sources supporting diapycnal mixing, especially in the abyssal ocean (Munk and Wunsch, 1998). Tian et al. (2009) point out that the internal tide energy flux generated in the Luzon Strait supports diapycnal mixing and enhanced mixing in the SCS. This work also finds enhanced diapycnal mixing below 1 000 m near the Luzon Strait. However, the tidal dissipation rate may only contribute 50% of the energy required for mixing (Munk and Wunsch, 1998). Therefore, it has been argued that wind may supply a part of energy (Wunsch and Ferrari, 2004). Alford (2003) suggests that the total power from wind to near-inertial motions could can 0.47 TW. Jing et al. (2011) found a correlation between diapycnal diffusivity and wind stress on spatial-seasonal variations. The wind-induced near-inertial wave plays an important role on diapycnal mixing in the northwestern Pacific (Jing and Wu, 2014; Li and Xu, 2014), while there are seldom researches that study how it influence the SCS. In this work, we find the wind-induced near-inertial energy can enhance diapycnal mixing specially in the flat region where the internal energy flux is much weaker (Fig. 4), and the diffusivity decreases with the increase of depth (Fig. 6). It is different in the rough region that the diffusivity increases with depth below 700 m owing to the joint effort of the topography and the barotropic flow (Fig. 6). The signal of growth of the enhanced diapycnal mixing resulting from the wind-induced near-inertial energy can be found in the upper three segments. There is still an argument about the depth that near-inertial waves can influence. Zhai et al. (2009) pointed out nearly 70% of the energy from wind working on near-inertial motions is lost within the top 200 m. However, recent researches indicate that the wind-induced near-inertial wave can penetrate up to 1 000 m (Wu et al., 2011; Li and Xu, 2014). Our study here indicates the growth of wind can enhance the mixing with depth extending to more than 700 m.

The authors thank the Key Laboratory of Tropical Marine Environment Dynamics, the South China Sea Institute of Oceanology, Chinese Academy of Sciences for providing the CTD data. We are grateful for the suggestions from Yang Qingxuan.