Multi-Year Winter Variations in Suspended Sediment Flux through the Bohai Strait

: The Bohai Strait is the only channel that allows material exchanges between the Bohai Sea and the Yellow Sea. It is also the only channel for the transportation of materials from the Yellow River to the Yellow Sea and the East China Sea. The supply of suspended sediment from the Bohai Sea plays a decisive role in the evolution of the mud area in the northern Yellow Sea and even the muddy area in the southern Yellow Sea. Previous studies have demonstrated that sediment exchange through the Bohai Strait occurs mainly in winter, but due to the lack of long-term observational data, changes in the sediment ﬂux over multiple years have not been studied. In this paper, based on L1B data from the MODIS (Moderate Resolution Imaging Spectroradiometer) -Aqua satellite, an interannual time series of the suspended sediment concentration (SSC) at each depth layers in the Bohai Strait in winter was established through 16 cruises that beneﬁted from the complete vertical mixing water in the strait in winter. The numerical model FVCOM, (Finite-Volume Community Ocean Model) which is forced by the hourly averaged wind ﬁeld, reﬂected the e ﬀ ect of winter gales. With the model simulated winter current from 2002 to the present and the SSC at each layer, multi-year winter suspended sediment ﬂux data were obtained for the Bohai Strait. This study found that in the winter, the suspended sediment output from the Bohai Sea to the Yellow Sea through the southern part of the Bohai Strait, while the suspended sediment input from the Yellow Sea to the Bohai Sea is through the northern part. In terms of long-term changes, the net ﬂux ranged between 1.22 to 2.70 million tons in winter and showed a weak downward trend. The output ﬂux and input ﬂux both showed an upward trend, but the increase rate of the input ﬂux was 51,100 tons / year, which was higher than the increase of the output ﬂux rate (46,100 tons / year). These changes were mainly controlled by the increasing strength of east component of winter wind. And the weak decrease in net ﬂux is controlled by the di ﬀ erence of output and input ﬂux.


Introduction
Research on the suspended sediment fluxes of the Bohai Strait is relatively extensive, with a variety of technical methods. Seasonal variation in the flux and the differences in the suspended sediment concentration (SSC) between different water layers are mainly controlled by the advance and retreat of the cold water mass in the northern Yellow Sea and the temperature and salinity changes in the Bohai Strait [1][2][3]. Suspended sediment transport though the Bohai Strait is concentrated mainly on the south side of the Bohai Strait and occurs mainly in winter. This is driven by the eastward current in the southern Bohai Strait and the effects of winter wind [4][5][6], and the scope is mainly restricted

Measured Hydrological Parameters
The measured data in this study mainly include SSC, water turbidity, temperature and salinity by SeaBird 911 conductivity-temperature-depth system (CTD). These in situ data were mainly used for the retrieval of remote sensing data and the verification of the retrieved results. The resolution of the temperature, salinity, and turbidity data was 1 m, and the resolution of the SSC data was based on the number of layers in the different water depths (generally 3-6 layers). To ensure a sufficient data volume, this study used measured data from different seasons for retrieval in order to obtain a retrieval model that can be applied to all seasons. The first dataset is from the joint voyage of the National Natural Science Foundation of China (11 cruises in total, see Figure 1), undertaken mainly by the "Dongfanghong 2" scientific research ship of Ocean University of China. The second dataset is from the joint voyage of the Qingdao Marine Science and Technology Pilot National Laboratory "Transparent Ocean" plan undertaken by the "Innovation 1" scientific research vessel of the Yantai Institute of Coastal Zone Research, Chinese Academy of Sciences in 2018 (3 cruises in total, see Table 1).

Measured Hydrological Parameters
The measured data in this study mainly include SSC, water turbidity, temperature and salinity by SeaBird 911 conductivity-temperature-depth system (CTD). These in situ data were mainly used for the retrieval of remote sensing data and the verification of the retrieved results. The resolution of the temperature, salinity, and turbidity data was 1 m, and the resolution of the SSC data was based on the number of layers in the different water depths (generally 3-6 layers). To ensure a sufficient data volume, this study used measured data from different seasons for retrieval in order to obtain a retrieval model that can be applied to all seasons. The first dataset is from the joint voyage of the National Natural Science Foundation of China (11 cruises in total, see Figure 1), undertaken mainly by the "Dongfanghong 2" scientific research ship of Ocean University of China. The second dataset is from the joint voyage of the Qingdao Marine Science and Technology Pilot National Laboratory "Transparent Ocean" plan undertaken by the "Innovation 1" scientific research vessel of the Yantai Institute of Coastal Zone Research, Chinese Academy of Sciences in 2018 (3 cruises in total, see Table 1).

Fitting Methods for Each Water Layer
Due to the influence of stronger winter winds in the Bohai Sea and the Bohai Strait, the water bodies become fully mixed ( Figure 2), with the water turbidity in the upper and lower layers relatively uniform. Therefore, the winter turbidity data over many years can be used to establish a fitting relationship between the surface layer and deeper layers in order to translate the surface remote sensing retrieval results into SSC in the middle and bottom layers.

Fitting Methods for Each Water Layer
Due to the influence of stronger winter winds in the Bohai Sea and the Bohai Strait, the water bodies become fully mixed ( Figure 2), with the water turbidity in the upper and lower layers relatively uniform. Therefore, the winter turbidity data over many years can be used to establish a fitting relationship between the surface layer and deeper layers in order to translate the surface remote sensing retrieval results into SSC in the middle and bottom layers. According to the previous studies, the SSC in seawater had a significant correlation with the distribution of seawater turbidity. Thus, turbidity values can be converted into SSC values [3,7]. Since surface SSC retrievals use measured SSC data from multiple cruises, it is necessary to convert the turbidity of each layer into the SSC before fitting the relationship between the surface turbidity and the turbidity of each layer. The conversion formula used in this article is the conversion method established by Liu et al. The specific method is as follows (Figure 3). According to the previous studies, the SSC in seawater had a significant correlation with the distribution of seawater turbidity. Thus, turbidity values can be converted into SSC values [3,7]. Since surface SSC retrievals use measured SSC data from multiple cruises, it is necessary to convert the turbidity of each layer into the SSC before fitting the relationship between the surface turbidity and the turbidity of each layer. The conversion formula used in this article is the conversion method established by Liu et al. The specific method is as follows (Figure 3). Figure 4 shows that the surface SSC has obvious linear relationships with the SSCs of the other layers. The correlation coefficient (R 2 ) gradually decreases from the surface layer to the bottom layer, but the lowest value is still above 0.84. The root mean square error (RMSE) gradually increases from the surface to the bottom and is nearly always below 3.3 mg/L ( Table 2), which indicates an acceptable fit for this article.  Figure 4 shows that the surface SSC has obvious linear relationships with the SSCs of the other layers. The correlation coefficient (R 2 ) gradually decreases from the surface layer to the bottom layer, but the lowest value is still above 0.84. The root mean square error (RMSE) gradually increases from the surface to the bottom and is nearly always below 3.3 mg/L ( Table 2), which indicates an acceptable fit for this article.

Remote Sensing Data
The remote sensing image data used were Moderate Resolution Imaging Spectroradiometer (MODIS)-Aqua satellite data downloaded from the NASA website (https://www.nasa.gov). The Level-1B (L1B) data for December, January, and February of each year from 2002 to 2018 (the winter of 16 years) were obtained. The images had a spatial resolution of 1 km.
The remote sensing data in this study were processed and transformed mainly using the SeaWiFS Data Analysis System (SeaDAS) software officially provided by NASA for atmospheric correction, cloud removal and other preliminary processing. The L1B remote sensing data were batched into Level-2 remote sensing reflectance (Rrs). When establishing a remote sensing retrieval

Remote Sensing Data
The remote sensing image data used were Moderate Resolution Imaging Spectroradiometer (MODIS)-Aqua satellite data downloaded from the NASA website (https://www.nasa.gov). The Level-1B (L1B) data for December, January, and February of each year from 2002 to 2018 (the winter of 16 years) were obtained. The images had a spatial resolution of 1 km.
The remote sensing data in this study were processed and transformed mainly using the SeaWiFS Data Analysis System (SeaDAS) software officially provided by NASA for atmospheric correction, cloud removal and other preliminary processing. The L1B remote sensing data were batched into Level-2 remote sensing reflectance (Rrs). When establishing a remote sensing retrieval model, the Rrs that best matched the measured data in time and space was selected. A time control period within 3 h was effective for our data. A fitting relationship between the remote sensing data and the measured data was established. In addition, because there is visible sea ice in the Bohai Sea and the Yellow Sea in winter, sea ice has a great influence on SSC retrieval. Therefore, this study used methods described in previous studies to remove the influence of sea ice [14,15].
With the above method, this study used the measured surface SSC at 28 stations to fit the remote sensing data of the Rrs555 band and establish a remote sensing retrieval model. The retrieval fitting function is as follows: The measured surface SSC had a high degree of fit with Rrs555; the R 2 was 0.9522 (Figure 5a), indicating that the established retrieval model had good accuracy and met the requirements of this study.
Remote Sens. 2020, 12, x FOR PEER REVIEW 7 of 16 With the above method, this study used the measured surface SSC at 28 stations to fit the remote sensing data of the Rrs555 band and establish a remote sensing retrieval model. The retrieval fitting function is as follows: The measured surface SSC had a high degree of fit with Rrs555; the R 2 was 0.9522 (Figure 5a), indicating that the established retrieval model had good accuracy and met the requirements of this study. Figure 2 shows that the area with high SSC values in winter is mainly concentrated in the southern part of the Bohai Strait; this may indicate that suspended sediment transport in the Bohai Strait occurs mainly in this area. The difference between the acquisition times of the satellite image and the in situ data was less than 3 h, the spatial difference was less than 500 m, and the remaining data after the sea ice was removed were considered as valid data values. To control the quality of the data, when calculating the average SSC in winter, if there were fewer valid points in the southern Bohai Strait and nearby areas, the SSC value in that year was considered to be an unreliable value. Figure 6 shows that the number of valid points was low in 2008 and 2009, and the number of valid points in the southern Bohai Strait was extremely low in 2009 ( Figure 6). Therefore, the data from 2009 were considered unreliable. In the following analysis, when analyzing the overall trends, unreliable results were not included.   Figure 2 shows that the area with high SSC values in winter is mainly concentrated in the southern part of the Bohai Strait; this may indicate that suspended sediment transport in the Bohai Strait occurs mainly in this area. The difference between the acquisition times of the satellite image and the in situ data was less than 3 h, the spatial difference was less than 500 m, and the remaining data after the sea ice Remote Sens. 2020, 12, 4066 7 of 14 was removed were considered as valid data values. To control the quality of the data, when calculating the average SSC in winter, if there were fewer valid points in the southern Bohai Strait and nearby areas, the SSC value in that year was considered to be an unreliable value. Figure 6 shows that the number of valid points was low in 2008 and 2009, and the number of valid points in the southern Bohai Strait was extremely low in 2009 ( Figure 6). Therefore, the data from 2009 were considered unreliable. In the following analysis, when analyzing the overall trends, unreliable results were not included.

Other Data
In this paper, the runoff and sediment transport data for the Yellow River were derived from the statistics from Lijin Station in the "Chinese River Sediment Bulletin". In addition, this study used cross-calibrated multi-platform (CCMP) wind field data to analyze the effect of wind on the transport of suspended sediment. The CCMP wind field data are vector data derived from remote sensing systems (RSS); their temporal resolution is 6 h, and their spatial resolution is 0.25° × 0.25° (http://www.remss.com/measurements/ccmp).

Suspended Sediment Flux Calculation Method
Using the current data simulated by the Finite-Volume Community Ocean Model (FVCOM) and the predicted data for the suspended sediments in various layers of the Bohai Strait in winter over many years, the output and input fluxes of the suspended sediments were calculated quantitatively. To study the trend of the suspended sediment transport in the Bohai Strait in winter over many years, we first used the equation [22,23] (1): The single-width fluxes at various points across the Bohai Strait were calculated, where Fs represents the single-width flux at a given point, H represents the water depth (unit: m), C represents predicted SSC (unit: mg/L) fitted at each layer, and U represents the seasonal average velocity (unit: m/s). Then, Equation (2) was used to calculate the suspended sediment flux across the entire Bohai Strait: (2) where S represents the distance between the first and last points in the section.
Considering that the satellite image and the measured data were fitted in the form of a seasonal average, and considering the quality of the satellite image, the seasonal average result was used for the flux calculation. The established FVCOM applied hourly wind data from 2012 to 2018.

Other Data
In this paper, the runoff and sediment transport data for the Yellow River were derived from the statistics from Lijin Station in the "Chinese River Sediment Bulletin". In addition, this study used cross-calibrated multi-platform (CCMP) wind field data to analyze the effect of wind on the transport of suspended sediment. The CCMP wind field data are vector data derived from remote sensing systems (RSS); their temporal resolution is 6 h, and their spatial resolution is 0.25 • × 0.25 • (http://www.remss.com/measurements/ccmp).

Suspended Sediment Flux Calculation Method
Using the current data simulated by the Finite-Volume Community Ocean Model (FVCOM) and the predicted data for the suspended sediments in various layers of the Bohai Strait in winter over many years, the output and input fluxes of the suspended sediments were calculated quantitatively. To study the trend of the suspended sediment transport in the Bohai Strait in winter over many years, we first used the equation [22,23] (1): The single-width fluxes at various points across the Bohai Strait were calculated, where F S represents the single-width flux at a given point, H represents the water depth (unit: m), C represents predicted SSC (unit: mg/L) fitted at each layer, and U represents the seasonal average velocity (unit: m/s). Then, Equation (2) was used to calculate the suspended sediment flux across the entire Bohai Strait: where S represents the distance between the first and last points in the section. Considering that the satellite image and the measured data were fitted in the form of a seasonal average, and considering the quality of the satellite image, the seasonal average result was used for the Remote Sens. 2020, 12, 4066 8 of 14 flux calculation. The established FVCOM applied hourly wind data from 2012 to 2018. The model also considered the effects of strong winter winds and different wind speed and direction conditions. The model has been verified in detail to prove its reliability [24].

Multi-Year Variations in the Surface SSC in the Bohai Strait
The surface SSC of the Bohai Strait exhibits obvious temporal and special variations. The interannual variation in the surface SSC in the northern Bohai Strait is small, with the difference between the highest and the lowest value is only approximately 5 mg/L (Figure 7). The surface SSC in the southern Strait varies greatly, and the difference between the highest and lowest values is approximately 28 mg/L. The southern part of the Bohai Strait is bounded by the northern islands of the Shandong Peninsula. The magnitude of change in the western strait is greater than that in the eastern. The difference between the highest and lowest value over these years in the eastern part of the strait is less than 10 mg/L. This maybe caused by the barrier effect of the island [17].

Multi-Year Variations in the Surface SSC in the Bohai Strait
The surface SSC of the Bohai Strait exhibits obvious temporal and special variations. The interannual variation in the surface SSC in the northern Bohai Strait is small, with the difference between the highest and the lowest value is only approximately 5 mg/L (Figure 7). The surface SSC in the southern Strait varies greatly, and the difference between the highest and lowest values is approximately 28 mg/L. The southern part of the Bohai Strait is bounded by the northern islands of the Shandong Peninsula. The magnitude of change in the western strait is greater than that in the eastern. The difference between the highest and lowest value over these years in the eastern part of the strait is less than 10 mg/L. This maybe caused by the barrier effect of the island [17].

Multi-Year Variations in SSC Flux in the Bohai Strait in Winter
Due to the north-side inflow and the south-side outflow current, the suspended sediment will be transported in different directions between the southern and northern Bohai Strait (Figure 8) which directed from the Bohai Sea to the Yellow Sea in the south, and from the Yellow Sea to the Bohai Sea in the north. However, as along with the water depth deepen, the location of the boundary between waters with opposite flux directions changes significantly. The surface boundary is relatively stable throughout the year, mainly located near 38.56 • N. There is a clear interannual change in the mid-level shear front, and the boundary generally exists between 38.47 • N and 38.53 • N. The position of the bottom boundary is also relatively stable near 38.4 • N. The location of the boundary gradually moves southwardly with increasing depth. This is mainly because the surface currents of the Bohai Strait is mainly controlled by the east component of the winter wind, which clearly exhibited a stronger output flux corresponds to the heavy northwesterly wind (Figure 8a). The bottom of the Bohai Strait is invaded mainly by a warm current from the Yellow Sea. The intensity of the Yellow Sea Warm Current gradually weakens from the bottom to the top, so the location of the boundary gradually moves south from the surface to the bottom [1,24,25].  To better study the vertical differences in the multi-year variations of the suspended sediment flux in the Bohai Strait in winter, fluxes at two stations located in the southern and northern Bohai Strait respectively were compared (Figure 9). Most of the output flux occurred at the surface layer in the southern Bohai Strait, which gradually decrease downward. This is mainly because the southern part of the Bohai Strait is affected by strong northwesterly winds in winter, and the SSC mixes well vertically, while the surface current velocity is significantly higher than the bottom current. The multi-year variations of suspended sediment flux in the northern Bohai Strait were more complicated than those in the southern part. The middle and lower layers in the northern part of the strait were mainly represented by the westward input flux. This process is close related to the Yellow Sea Warm Current, which is a upwind current carrying a large amount of suspended sediment through the Yellow Sea into the Bohai Strait, and finally, into the Bohai Sea.

Multi-Year Variations in the Output Flux of Suspended Sediment in the Bohai Strait in Winter
The output and input flux represents the sediment transported from the Bohai Sea to the Yellow Sea and from the Yellow Sea to the Bohai Sea, respectively. The sum vector of the output and input flux is the net flux. To further analyze the relationship between the fluxes of the Bohai Strait and various influencing factors, this section calculates the sediment volume of the Yellow River, the average wind speed in winter each year, the east component of wind

Multi-Year Variations in the Input Flux of Suspended Sediment in the Bohai Strait in Winter
From 2002 to 2017, the input flux of suspended sediment in the Bohai Strait in winter ranged between 2.233 and 4.047 million tons. The magnitude of the change in the input flux was much smaller than that in the output flux. The overall input flux also showed an upward trend, and the increase rate was higher than that of the output flux, reaching 51,100 tons/year (Figure 10b). The change in the input flux depended mainly on the Yellow Sea Warm Current flowing westward into the Bohai Sea, and its main channel varied within the strait. The Yellow Sea Warm Current is an upwind positive-pressure flow originating from the southern Yellow Sea [27], which is also related to the strength of east component winter wind (Figure 10f). Its velocity is small, and the SSC in the Yellow Sea is lower than that in the Bohai Sea, thus the input flux of suspended sediment is relatively lower than output in winter. Thus both of the input and output flux are induced by the east component of winter wind.  [26]. A large quantity of the water escape from Bohai Sea to Yellow Sea through Bohai Strait. Meanwhile, the relatively high sea surface height is distributed along the northern coast of the Shandong Peninsula, and the Northern Shandong Coastal Current is markedly enhanced because of geostrophic balance. Furthermore, intensified waves facilitate the rapid increase of SSC in the shallow water of southern Bohai Sea, which is then carried out of Bohai Sea by the Northern Shandong Coastal Current. This period is the main transport stage of the sediments from Bohai Sea [16,17].

Multi-Year Variations in the Input Flux of Suspended Sediment in the Bohai Strait in Winter
From 2002 to 2017, the input flux of suspended sediment in the Bohai Strait in winter ranged between 2.233 and 4.047 million tons. The magnitude of the change in the input flux was much smaller than that in the output flux. The overall input flux also showed an upward trend, and the increase rate was higher than that of the output flux, reaching 51,100 tons/year (Figure 10b). The change in the input flux depended mainly on the Yellow Sea Warm Current flowing westward into the Bohai Sea, and its main channel varied within the strait. The Yellow Sea Warm Current is an upwind positive-pressure flow originating from the southern Yellow Sea [27], which is also related to the strength of east component winter wind (Figure 10f). Its velocity is small, and the SSC in the Yellow Sea is lower than that in the Bohai Sea, thus the input flux of suspended sediment is relatively lower than output in winter. Thus both of the input and output flux are induced by the east component of winter wind.

Multi-Year Variations in the Net Flux of Suspended Sediment in the Bohai Strait in Winter
Overall, the net flux in the Bohai Strait ranged from 1.22 to 2.70 million tons. The net flux changed steadily over the years, showing a very slow downward trend, with a decline of only 4960 tons/year (Figure 10a). The net flux is controlled by the difference between the input and output. In the years when the east component was larger, the net flux in the Bohai Strait was also larger. In the years when the east component was above 1.6 m/s, the net flux in the Bohai Strait exceeded 2.5 million tons. The net flux in 2013, which corresponded to the smallest east component of winter wind, was the lowest, approximately 1.22 million tons. The net flux in the Bohai Strait has little to do with the sediment transport volume of the Yellow River, which indicates that the sediment in the Bohai Strait is mainly derived from the resuspension of sediments in the Bohai Sea.

Conclusions
Considering the vertical well mixed water in the Bohai Strait in winter, remote sensing data from MODIS-Aqua in combination with in situ measured water turbidity and SSC were used to establish a remote sensing model in order to translate the surface remote sensing retrieval results into SSC in each water layers in winter. Then, by incorporating current data simulated by FVCOM, the suspended sediment flux in the Bohai Strait in winter of 2002 to 2018 was obtained.
The output suspended sediment flux occurs in the southern part of the Bohai Strait in winter from the Bohai Sea to the northern Yellow Sea, and the sediment flux gradually decreases from the surface downward. This is mainly because of the stronger wind-driven current at the surface layer than that at the bottom, while the input sediment flux directing from the northern Yellow Sea to the Bohai Sea occurs in the middle and lower layers of the northern part of the strait. This process is mainly affected by the Yellow Sea Warm Current which carries suspended sediment flowing into the Bohai Strait.
In terms of long-term changes, the net sediment flux to the Yellow Sea ranged between 1.22 and 2.70 million tons in winter, showing a slight downward trend. That is because of a relatively smaller increase rate of output flux by 46,100 tons/year comparing to the increase rate of input flux by 51,100 tons/yr. The multi-year winter variations in the suspended sediment flux in the strait are mainly controlled by the east component of the winter wind. This is necessary for the accurate study of sediment transport in different types of straits in the world.