2. Study Area and Methods
The main object of this study is the Third Nuclear Power Plant (marked as NPP3 in
Figure 1) located on the northwestern coast of Nanwan Bay at about 121° E and 22° N where the Taiwan Strait, the South China Sea (SCS), and the West Philippine Sea (WPS) meet. The cooling water inlet is inside the power plant compound, but the thermal discharge outlet is a few kilometers south of the power plant. The bay has an area of about 38 km
2, a steep bottom topography reaching 100 m in depth, and only small creeks with seasonal outflow. The bay is part of Kenting National Park and is home to extensive coral coverage and abundant marine life. The park attracts several million visitors each year, so the potential pollution emitted by the power plant is of great concern. The plant’s two reactors started began operating in May 1984 and May 1985, respectively, and the Environmental Impact Assessment (EIA) study began in July 1979. We took over the field surveys in July 1993. The hydrology team took CTD (conductivity, temperature, and depth) measurements every season using a Seabird SBE 19 at 19 stations with GPS positioning (shown as dots in
Figure 1). A drifter equipped with a Pace Scientific Inc. XR5-T temperature logger (Mooresville, NC, USA) and a GPS was deployed at the cooling water outlet and followed to track the plume on the same day as the CTD deployments. The precision is 0.02 °C. A NorTek AS Aquadopp (NorTek, Vangkroken, Norway) current meter equipped with a temperature logger was used to continuously record the current at a water depth of 20 m, 500 m south of the outlet. In addition, temperature loggers (HOBO, Water Temp Pro v2; Tinytag, TG-4100) (HOBO, Woburn, MA, USA) were placed near the cooling water inlet (8 m in water depth), outlet (3 and 9 m in water depth), and in Moubitou (9 m in water depth).
The biogeochemical team took CTD measurements at ten stations with GPS positioning (shown as “+” in
Figure 1) four times each year. In addition, we collected seawater samples at various depths at seven stations (St. 12, 18, 20, 21, 22, 23, and 24). We measured temperature and salinity using an Autosal 8400B salinometer (Sea-Bird Scientific, Bellevue, WA, USA) calibrated with the International Association of Physical Sciences of the Oceans standard seawater with a precision of ±0.004. Seawater samples were collected in a Niskin bottle (Ocean Scientific International, Avant, UK), subdivided into dark bottles, and stored in a cooler after being poisoned with HgCl
2. Samples were analyzed in a laboratory. The dissolved oxygen was measured using Winkler titration with a precision of 0.5%. Nitrate (NO
3) and nitrite (NO
2) were measured using a cadmium reduction + azo dye colorimetric method with an OI Analytical FS 3100 Colorimeter at a precision of ±0.035 μmol/kg for NO
3 and ±0.02 μmol/kg for NO
2.
Phosphate (PO
4) was measured using an ascorbic acid reduction–calorimetric method with a precision of 0.01 μmol/kg. Silicate (SiO
2) was measured by the ascorbic acid/oxalate reduction–colorimetric method with a precision of ±0.1 μmol/kg. The colorimeter mentioned above was used for PO
4 and SiO
2 measurements. The nutrient standards were purchased from WAKO, Japan. Data reports are available at URL:
https://npp3.nsysu.edu.tw (accessed on 30 May 2023).
4. Results and Discussion
As an example of the internal tides,
Figure 2 shows the continuous temperature records measured at 3 m and 9 m in water depth from the outlet and at 9 m away from Moubitou south of the power plant (
Figure 1) between 6 May and 14 August 2021. There are low-frequency variations lasting a week or two, apparently related to the tidal movements, which are also shown in
Figure 2. The water temperature sometimes reaches as high as 32 °C, which is far too high for the healthy growth of corals. Fortunately, there are daily temperature variations of a few °C; for a few hours every day, cold waters that temporarily cool the environment are pumped up by internal tides.
The temperatures at 3 m near the outlet can be around 2 °C higher than at 9 m, showing the influence of warmer, lighter thermal discharge. It is important to note that, at 9 m, the temperature near the outlet is almost always lower than that in Moubitou by up to 2 °C (
Figure 2). The fluctuating effect caused by stratification likely causes a lower temperature near the outlet [
11]. When the plume spreads near the surface, it entrains some subsurface water, thus lowering the temperature of the thermal plume. Low-temperature offshore subsurface waters must continuously flow onshore to replace the outflowing entrained water. As a result, beneath the plume near the outlet, the temperature is lower than it would have been otherwise, thus helping to relieve corals from thermal stress in summer [
12].
The daily temperature decreases last a few hours and then return to the pre-internal tide value, except when typhoons are nearby. It has been reported that typhoons affect Nanwan Bay. During the period shown in
Figure 2, the tropical storms, Choi-wan (around 5 June) and Lupit (around 23 July), as well as typhoon In-fa (around 5 August) occurred. The seawater temperature continuously dropped from a few days to at least two weeks following consecutive typhoons. In the case of In-fa and Lupit (
Figure 2), the cumulative temperature drop was as high as 10 °C. Seemingly, storms pushed offshore waters towards the coast, causing the upwelling of cold, subsurface waters. Heavy rains after a storm also generate stratification and buoyancy, which helps bring up cold, subsurface waters [
13]. Other distant oceanic phenomena, such as eddies and the meandering of the Kuroshio Current and its branches, could also replace the waters of Nanwan Bay, thus diminishing any local effects caused by the power plant.
Comparisons between temperatures at the background station 1500 m away from the outlet and temperatures at three stations (not shown in
Figure 1) 500 m away from the outlet are shown in
Figure 3. The local law dictates that temperatures at 500 m away from the outlet must not exceed 4 °C higher than those at the background stations. It is clear that the temperature difference was mostly less than 2 °C, although occasionally, it reached 3.5 °C. The average temperature increases were 0.75 and 0.82 °C north and east of the outlet (marked as Northern sta. and Central sta. in
Figure 3), respectively. South of the outlet, the temperature was 1.2 °C higher than at the background station because of the prevailing southward flowing current, which is described later. Hydrological and drifter data indicate that a thermal plume with a temperature 1 °C higher than t the background station temperature extends for a maximum of 1 km from the outlet. Note that the temperatures at these three stations are occasionally lower than at the background station. This is perhaps because of the upwelled cold water induced by the entrainment.
Figure 4 shows the typical horizontal distribution of temperature, salinity, pH, DO, nitrate + nitrate, phosphate, and silicate at a water depth of 10 m in Nanwan Bay. An upwelling center of subsurface waters occurs in the middle of the bay, with a temperature of more than 1 °C below that close to the coast. Furthermore, higher salinity and nutrients but lower pH and DO, typical of subsurface waters in the tropical NW Pacific Ocean are also found near the upwelling center. Naturally, the upwelled waters replace and push surface waters offshore, carrying with them the thermal plume from the power plant. Partly due to its poor resolution, the thermal plume is not apparent in
Figure 4. The waters near the outlet with a temperature above 28 °C shown in
Figure 4 are 1.5 °C warmer than the water at the upwelling center but still cooler than waters in the eastern bay far from the nuclear power plant.
The potential temperature (theta) vs. salinity plot is frequently used to identify water masses, such as Kuroshio and the SCS waters [
14]. In the upper water column, the WPS waters are generally more saline than the SCS water when compared at the same potential temperature.
Figure 5 shows the typical theta/S plots for the WPS (broken line) and the SCS (thin line), with the former located to the right of the SCS curve. Additionally, plotted in
Figure 5 are temperature and salinity data for all bottled samples collected between 1993 and 2022. Except for rare cases when coastal water reaches Nanwan Bay in winter following an extreme NE monsoon, our study area comprises only waters from the WPS and the SCS. Mixing waters from these two seas creates water with temperature and salinity located between the two trends, with a higher contribution to the WPS, as shown by data skewed to the right. Temperature and salinity data in our study area mostly lie between these two trends. Notably, during the cold phase of the Pacific Decadal Oscillation (PDO), waters in Nanwan Bay tend to contain fresher SCS water (
Figure 5a), while during the warm PDO phase, they contain more saline WPS (
Figure 5b).
The El Niño–Southern Oscillation (ENSO) phenomenon also affects the source of waters in Nanwan Bay, with more intrusion from the Kuroshio Current during El Niño when the salinity and temperature move significantly further to the right. In other words, waters in Nanwan Bay are closer to the typical potential temperature vs. salinity line of the Kuroshio Current. During a La Niña event, Kuroshio Current intrusion is reduced, and the salinity and temperature signature move to the left and become closer to the typical SCS line (
Figure 5c,d). Furthermore, during an El Niño event, monsoons weaken, and basinwide upwelling in the SCS is reduced, thus raising the surface water temperature compared to the normal periods [
15]. On the other hand, during a La Niña event, there are stronger monsoons, and upwelling in the SCS increases, lowering the surface temperature of the water.
Figure 6 shows the average temperature anomaly for the surface samples (excluding St. 24) and the Oceanic Niño Index; ONI) since 1986 (data before 1993 were collected from the study by Hung et al., 1993 [
16]). The El Niño events with a positive ONI anomaly (shown in red in
Figure 6a) generally correlate with cooler temperatures, as discussed above. On the other hand, during the La Niña years (negative ONI anomalies are shown in blue in
Figure 6a), there were generally positive temperature anomalies, i.e., there is an inverse relationship between the ONI and the temperature anomaly. Based on this set of temperature records, the rate of temperature increase has been reported to be a highly statistically significant at 0.026 °C/yr (
p = 0.004; 1986–2022;
Table 1), compared with the global temperature increase of 0.018 °C/yr that occurred between 1981 and 2022 [
17]. Giuliani et al. [
18] reported an increase of 0.028 °C/yr between 1960 and 2011, similar to that reported by Bai et al. [
19]: around 0.03 °C/yr between 2003 and 2014 in low-latitude Eurasian marginal seas. The results of this study are consistent with those values.
We cannot evaluate whether the ENSO or the PDO correlates better with the temperature anomaly. However, during most phases, including the current PDO cold phase (negative values shown in blue; cooler than normal in the eastern North Pacific yet warmer than usual in the western North Pacific), there is generally a positive temperature anomaly (
Figure 6b). In contrast, during the PDO warm phase (positive anomalies are shown in red in
Figure 6b; warmer than normal in the eastern North Pacific but cooler than usual in the western North Pacific). Our temperature pattern agrees with the PDO signals in the western North Pacific Ocean. Note that the temperature fluctuations reach as high as 5 °C between El Niño and La Niña events or between PDO cold and warm phases. These temperature variabilities are several times larger than the temperature variations of the thermal plume in Nanwan Bay.
Statistically, the PDO correlates well with salinity anomalies in data collected from every season between 1986 and 2022 (
Figure 7), except for summer data, where the correlation has a “
p” value of 0.06. In other seasons, “
p” values are 0.05 or lower. On interannual and longer timescales, a recent study [
22] showed that the PDO’s potential impact should be considered when examining the inter-annual variability of flow patterns in the low-latitude northwest Pacific. An anomalous anticyclonic wind field appears over the Philippine Sea during a warm PDO phase. The southerly anomalous wind off the Philippines reduces the Kuroshio Current speed near Luzon. A slower-flowing Kuroshio Current means it can change direction more easily, causing increased salty Kuroshio Current intrusion into the SCS. In other words, Nanwan Bay, which connects the Kuroshio Current and the SCS is more affected by the Kuroshio, and its salinity increases, as shown in
Figure 7.
Across the Taiwan Strait and on almost the same latitude is Daya Bay (
Figure 1;
Table 1), with an area of about 600 km
2. It is relatively shallow, with an average depth of only 11 m and a maximum of 21 m, compared to 120 m in Nanwan Bay. Nuclear power plants with two 844 MWe reactors began operating in August 1993 and February 1994, respectively [
23]. In 1993, there was essentially no thermal pollution, and in 1994, a thermal plume with a temperature 5 °C higher than the ambient temperature occupied an area of only 0.01 km
2. In 1999, this area increased to 0.08 km
2, and seawater temperature 7 °C higher than the ambient temperature was recorded for the first time, occupying an area of 0.02 km
2. In 2001, areas with temperatures 5 and 7 °C higher than the ambient temperature expanded to 0.55 and 0.07 km
2, respectively. In 1999, a thermal plume with a temperature 1 °C higher than the ambient temperature extended as far as 10 km south of the plant (Tang et al. [
24];
Table 1). This is comparable to findings for Nanwan Bay, where the temperature increase is much lower, and the thermal plume with a temperature 1 °C higher than the background extends only 1 km (
Table 1). The rate of temperature increase recorded in Daya Bay between 1982 and 2004 was 0.079 °C/yr (
p = 0.033), based on the data of Wang et al. [
20], a much higher figure than that found in Nanwan Bay (0.026 °C/yr) with power units that were half the size.
In 2003, 2 × 950 + 2 × 1007 MWe nuclear reactors began operating, resulting in even more enormous thermal plumes, reaching 0.36 km2 for waters with temperatures higher than 7 °C higher than the ambient temperature. Unlike Nanwan Bay, where internal waves or upwelling can suck up subsurface waters 10 °C lower in temperature, there are no internal waves or upwelling in Daya Bay. Furthermore, because of the shallowness of the bay, the buoyant plume does not entrain any subsurface water that is more than a couple of degrees C cooler. The vase-like shape of Daya Bay, unlike the bowl-like shape of Nanwan Bay, also limits the exchange of water with the outside water. As a result, the heat in the thermal plume in Daya Bay is not dissipated quickly enough compared to Nanwan Bay.
Remote sensing images show that thermal plumes accumulate near nuclear power plants in Daya Bay [
24,
25,
26]. On the other hand, the thermal plume in our study area frequently travels southwestward and leaves the bay.
Figure 8 shows a rose diagram for currents measured between 24 December 2020 and 10 November 2022. On average, for almost two years, 57% of the time, the currents flow toward the southwest, i.e., leaving the bay. The second most dominant flow direction is toward the west–southwest, also leaving the bay. The northeastward currents, moving inside the bay, occur about as frequently as the south–southwest flowing currents. In short, most currents flow towards the open ocean, carrying with them the heat discharged by the nuclear power station.
The dynamics for the dominance of the southwestward flow is attributed to the cyclonic recirculation created by the eastward-flowing ebb current sweeping off the bowl-shaped embayment (
Figure 1), resulting in the tidal currents mostly moving southwestward in the northwest of the bay during both ebb and flood tides [
8]. We compared the temperatures recorded during the normal operation and maintenance of the nuclear power plants at Nanwan Bay (
Table 2). No maintenance occurred in summer, but in the the other three seasons, the results are mixed, and the data have no anomalies except for those recorded at St. 24 near the outlet, where the average temperature of 28.7 °C during normal operation in autumn is 1.5 °C higher than the average temperature of 27.2 °C when the plants are under maintenance. During spring and winter, the average temperatures at St. 24 are 0.7 °C higher than those recording during maintenance periods. Note that autumn is the second busiest season for the NPP3; therefore, more heat is released during normal operation, causing s slightly larger difference between normal operation and under-maintenance operation compared with spring and winter.