The South China Sea is an oligotrophic marginal sea located in the tropical-subtropical Northwestern Pacific Ocean. Under the influences of monsoon winds, both the physical and biogeochemical processes exhibit distinct seasonal variability in the upper waters. In order to study the seasonal variations of surface phytoplankton, a one-dimensional coupled physical-biogeochemical model was developed and applied to the deep basins of the Northern South China Sea, away from the coastal upwelling regions. Forced under real-time surface monsoon winds and heat flux, the model reproduced the mixed layer depth, sea surface temperature and surface chlorophyll-a compared with satellite observations and previous reported values. In seasonal mean, the mixed layer depth was highest in winter (∼61.62 m) and lowest in spring (∼12.07 m). The sea surface temperature was lowest (∼25.05°C) in winter and highest (∼29.20°C) in summer. Furthermore, conspicuous phytoplankton blooms occurred in winter with the highest chlorophyll-a concentration up to ∼0.21 mg m−3. In other seasons, the concentration remained relatively low, especially in summer (∼0.05 mg m−3). The spatial distributions of phytoplankton were closely related with patterns of surface nutrient availability, as well as mixed layer depth and sea surface temperature. These relationships indicate that surface phytoplankton primary production was mainly controlled by nutrient availability, which was dominated by vertical turbulent diffusion in the deep basin of the northern Sea which is away from the coastal upwelling regions. Overall, our model results indicated that the seasonal variability of surface phytoplankton was modulated by coupled effects of physical and biogeochemical processes in the Northern South China Sea.
It is believed that marine phytoplankton play important roles in global carbon cycle and global climate change issues (Fasham, 2003). In the upper euphotic zone, the growth of phytoplankton is modulated by several physical (e.g. light, temperature, wind mixing, etc.), biogeochemical (e.g. nutrient availability of nitrate, phosphate, silicate, and iron, natural mortality and zooplankton grazing, etc.) and physiological (e.g. photo-adaptation) factors (Mann and Lazier, 1996). Essentially, phytoplankton need light and nutrients to carry out photosynthesis; therefore these two factors play predominant roles in modulating oceanic primary productions as well as the phytoplankton biomass. In most temperate coastal waters and northern North Atlantic, phytoplankton standing stock always experiences a characteristic seasonal cycle, which includes a strong spring increase known as the spring blooms (Sverdrup, 1953). Because of different physical and biogeochemical conditions in various open oceans, this seasonal pattern is not always the case, especially in the subtropical gyres, the equatorial zone, and upwelling regions (Miller, 2004). Does the South China Sea (SCS), as an oligotrophic marginal sea located in the tropical and subtropical Northwestern Pacific Ocean (NPO), experience a spring bloom in the phytoplankton standing stock in the upper water column?
It is known that the SCS is a semi-enclosed marginal sea, with a total area of about 3.5 × 106 km2 and with maximum depth more than 5500 m in the center deep basin (Chen et al., 2001). Situated between the western Pacific warm pool and the Tibetan Plateau, the SCS is under the persistent influence of South East Asian Monsoon wind which prevails in the northeastward in winter and reverses to southwestward in summer, spring and autumn being transitional periods (Wong et al., 2007). And the seasonal reversing monsoon winds play important roles in regulating the upper ocean physical and biological processes in the SCS (Liu et al., 2002; Gan et al., 2006).
Based on two cruises carried out in 1998, Ning et al. (2004) proposed that the spatial distributions of phytoplankton and primary production were closely related to the coupled processes (e.g. coastal upwelling, basin scale to mesoscale eddies) primarily forced under the monsoon winds. Tang et al. (1999) and Wang et al. (2010) investigated the winter phytoplankton bloom phenomena occurred off the southwest of Luzon Strait in winter using both satellite derived data and shipboard measurements. They suggested that the phytoplankton bloom was mainly attributed to the winter upwelling occurring there (Shaw and Chao, 1994; Shaw et al., 1996). Using a three dimensional coupled physical-biogeochemical model, Liu et al. (2002, 2007) examined the spatial and temporal distributions of chlorophyll-a and primary production under the influence of monsoon winds in the SCS. Their model successfully reproduced the patches of high chlorophyll-a in the upwelling regions both in winter and summer.
Although many previous observations and numerical modeling studies have been conducted to examine the basin scale ecosystem dynamics in the SCS (Liu et al., 2002, 2007; Ning et al., 2004; Wong et al., 2007; Lu et al., 2010), we still do not have a quantitative understanding on the seasonal variability and the associated driving mechanisms of surface phytoplankton in the area away from the upwelling regions of the SCS. In present research, a vertical one-dimensional coupled physical-biogeochemical model has been constructed and utilized. The coupled model was forced under real-time surface forcing and was verified against satellite observations and previous reported results.
Study area and the one-dimensional coupled model
The study area is shown in Figure 1. The domain is located between 114°52′ E and 116°32′ E in longitude, and between 17°15′ N and 18°47′ N in latitude, just within the deep basin of the Northern SCS. Obviously, the study area is away from the previously reported coastal upwelling regions in the SCS (Shaw et al., 1996; Jing et al., 2009; Gan et al., 2009) but partly includes the South East Asia Time-Series Station (SEATS station, Wong et al., 2007). The bottom topography is relatively uniform with domain averaged depth >3500 m.
The hydrodynamic model used is a vertical one-dimensional water column mixed layer model (Mellor and Durbin, 1975; Mellor, 2001). The mixed layer model is modified based on Princeton Ocean Model (POM) (Blumberg and Mellor, 1987) which is a three-dimensional, primitive equation, ocean general circulation model with free surface in the upper layer. The mixed layer model uses the Mellor-Yamada Level 2.5 turbulence closure scheme to calculate the vertical turbulent viscosity and diffusivity coefficients, KM and KH, for momentum and scalar, respectively (Mellor and Yamada, 1982). For detailed model description and modification, the readers can be referred to previous research (Mellor, 2001; Wang, 2007).
The biogeochemical model used in present research was developed and tested by Liu et al. (2002, 2007). The model includes four state variables (Figure 2), i.e. dissolved inorganic nitrogen (N), phytoplankton (P), zooplankton (Z), and detritus (D). The time-dependent change of each state variable (C) is governed by the following equation:2002).
The coupled physical-biogeochemical model only resolved the upper 200 m of water column. And there were 101 levels in the vertical with logarithmic distribution near surface and bottom in order to precisely reproduce the boundary layer processes.
Initial profiles for temperature, salinity and nitrate in winter, used for water column mixed layer model and biogeochemical model, were derived from World Ocean Atlas 2001 (WOA01, Conkright et al., 2002). Chlorophyll-a profile was obtained by using climatological winter mean. Sea-viewing Wide Field-of-view Sensor (SeaWiFS, NASA) satellite observed chlorophyll-a combined with Eumeli Profiles (French JGOFS, http://www.obs-vlfr.fr/cd_rom_dmtt/eu_main.htm) (Liu et al., 2002). In order to avoid the excess deviations of modeled profiles from the real profiles, nudging terms were added for all state variables during the model run, the modeled profiles were relaxed to the climatological values.
All the surface forcing fields, i.e. surface wind components at 10 m, air temperature at 2 m, relative humidity, sea level pressure, total cloud cover, and net shortwave radiation, were obtained from ERA-Interim archives of European Center for Medium-Range Weather Forecasts (ECMWF; Berrisford et al., 2009) reanalysis data with a temporal resolution of 6 h. Furthermore, the photosynthetically active radiation (PAR) was directly calculated from net shortwave radiation based on a constant ratio of 0.45 (Fennel et al., 2006).
The time step for the water column mixed layer model is 60 s. The biogeochemical model uses the same time step as the physical model. Firstly, the coupled model was run from rest for one year forced under climatological winter averaged surface forcing. After the spin-up periods, another model run was carried out from January 1996 to July 2002 for more than 6 years forced by real-time surface forcing. The model outputs from January 1997 to July 2002 were archived and the climatological monthly results were used for analysis in the following sections.
Results and Discussions
Mixed layer depth and sea surface temperature
Under the influence of monsoon winds and air-sea heat flux, the model simulated mixed layer depth (MLD) demonstrated distinct seasonal evolutions in the Northern SCS (Figure 3). The simulated MLD was deepest in winter (∼61.62 m) and reached to a minimum value (∼12.07 m) in spring (Table 1). The MLD was deepening slowly with the decreased surface heating during summer. When the southwest monsoon was diminishing and the stronger northeast monsoon winds began appearing in the autumn, the MLD deepened rapidly again and reached the annual maximum till boreal winter. The observed MLD in SEATS station was ∼20 ± 6 and ∼67 ± 26 m in summer and winter, respectively (Wong et al., 2007; Tseng et al., 2009), which agreed reasonably with our model results.
As shown in Figure 4a, the model simulated sea surface temperature (SST) also experienced obvious seasonal trends with the lowest temperature (∼25.05°C) occurred in winter and the highest temperature (∼29.20°C) in summer (Table 1). The SST was increasing during the spring when the wind induced mixing was weakened in the water column (with minimum MLD) and solar radiation was intensified above the sea surface. In contrast, the stronger northeast monsoon winds and relative low air temperature led to significant depression in SST in the autumn. Then the model simulated SST was compared with satellite observations from advanced very high resolution radiometer data (AVHRR, NOAA) and conspicuous agreement was obtained between them (Figure 4a).
Sea surface chlorophyll-a and phytoplankton biomass
The model simulated sea surface chlorophyll-a concentration was highest (∼0.21 mg m−3) in winter (Table 1) which was consistent with the shipboard observations carried out at SEATS station (∼0.3 mg m−3, Tseng et al., 2005). The surface chlorophyll-a appeared very low in the rest of the year, especially in summer with the concentration only ∼0.05 mg m−3 (Table 1).
When the model simulated surface chlorophyll-a was compared to satellite observations from SeaWiFS (from January 1998 to January 2002), reasonable agreements were found between them (Figure 4b). Both the model results and SeaWiFS data had shown the persistent phytoplankton blooms in winter. The biomass became relatively low in the rest periods of the year. However, the model had overestimated the winter chlorophyll-a concentration and underestimated those in other seasons.
The model simulated surface phytoplankton biomass (Figure 5) shared the same seasonal trends with the surface chlorophyll-a (Table 1). The phytoplankton blooms occurred in the winter and the biomass decreased to a relative low level in other seasons, especially during summer (∼0.05 mmol N m−3; Table 1). Furthermore, the seasonal patterns of phytoplankton were consistent with filed observations (Tseng et al., 2005) and previous modeling results (Liu et al., 2002; Wang, 2007).
Factors influence the seasonal variability of surface phytoplankton
Nutrient enrichment experiments in the SCS have revealed that the primary productivity in the euphotic zone is limited by the availability of nutrients (nitrate or phosphate) (Chen, 2005; Ning et al., 2004). Our model simulated surface phytoplankton distributions closely followed the surface nutrient abundance in the Northern SCS (Figure 5a and b). The maximum phytoplankton biomass (∼0.21 mmol N m−3) occurred when the nutrient was replete as the northeast monsoon winds were prevailing in the winter (Table 1). In other seasons, the nutrient concentration was too low to support the phytoplankton growth resulting in significant decrease in phytoplankton biomass, especially in summer. Then the question arises: which processes are responsible for the nutrient supply to the upper euphotic zone in the Northern SCS away from coastal upwelling regions? Our model results demonstrated that both the phytoplankton biomass and nutrient concentration were positively correlated with the MLD and negatively correlated with SST (Figures 3 and 4). The relationships implied the importance of vertical turbulent mixing processes on regulating the surface phytoplankton evolutions in the Northern SCS (Tseng et al., 2005).
As shown in Figure 5c, the maximum PAR occurred in the spring and the minimum value occurred in autumn, both of them were found in the monsoon transitional seasons. The seasonal pattern coincided with the distributions of net solar radiation. In seasonal mean, the PAR was ∼92 W m−2 and ∼59 W m−2 in spring and autumn, respectively, and the values were nearly the same during the northeast and southwest monsoon seasons (∼72 W m−2; Table 1). The small seasonal variability of PAR, especially in summer and winter, indicated that the light might not be the primary driving factor accounting for the highest surface phytoplankton biomass in winter as well as their seasonal variation in the Northern SCS.
A one-dimensional coupled physical-biogeo- chemical model was constructed and used in the Northern SCS away from coastal upwelling regions. The coupled model was forced with real time surface forcing and model results compared reasonably well with satellite observations, as well as with previous field measurements.
Based on our model results, there were distinct seasonal patterns in model simulated surface phytoplankton biomass. The spatial distribution of surface phytoplankton was positively and negatively correlated with MLD and SST, respectively and closely followed the spatial distributions of surface nutrient concentration. It is believed that the deep nutrient-rich water was brought to the euphotic zone mainly by vertical turbulent diffusion processes. The seasonal variability of surface phytoplankton was predominantly controlled by nutrient availability in the deep basin of the Northern SCS away from land runoff or coastal upwelling regions. Overall, our results support the hypothesis that the seasonal variability of surface phytoplankton is attributed to the coupled effects of physical and biogeochemical processes in the Northern South China Sea.
We would like to thank Dr. Kon-Kee Liu for sharing the NPZD model code with us. We also want to thank NASA, NOAA and ECMWF for providing the SeaWiFS, AVHRR and ERA-Interim data for present research. This research was supported by the Open Research Program from State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences (CAS), as well as financially supported by the 908 Project of Guangdong (GD908-02-03) and MOST of China (Grant no.2011CB403504).