Exploring the Green Tide Transport Mechanisms and Evaluating Leeway Coefficient Estimation via Moderate-Resolution Geostationary Images

Abstract

The recurring occurrence of green tides as an ecological disaster has been reported annually in the Yellow Sea. While remote sensing technology effectively tracks the scale, extent, and duration of green tide outbreaks, there is limited research on the underlying driving mechanisms of green tide drift transport and the determination of the leeway coefficient. This study investigates the green tide transport mechanism and evaluates the feasibility of estimating the leeway coefficient by analyzing green tide drift velocities obtained from Geostationary Ocean Color Imager-II (GOCI-II) images using the maximum cross-correlation (MCC) technique and leeway method across various time intervals alongside ocean current and wind speed data. The results reveal the following: (1) Significant spatial variations in green tide movement, with a distinct boundary at 34°40′N. (2) Short-term green tide transport is primarily influenced by tidal forces, while wind and ocean currents, especially the combined Ekman and geostrophic current component, predominantly govern net transport. (3) Compared to 1, 3, and 7 h intervals, estimating the leeway coefficient with a 25 h interval is feasible for moderate-resolution geostationary images, yielding values consistent with previous studies. This study offers new insights into exploring the transport mechanisms of green tides through remote sensing-driven velocity.

1. Introduction

The annual outbreak of the green tide, caused by Ulva prolifera, in the Yellow Sea has evolved into a recurring ecological disaster phenomenon over the past 17 consecutive years [1,2,3]. Large-scale green tides on the sea surface have a detrimental impact on aquaculture, tourism, and the local ecological ecosystems [4,5]. It is imperative to implement measures for the removal and mitigation of this ecological disaster [6]. The expeditious and precise acquisition of spatial distribution and drift status of green tides, coupled with a comprehensive understanding of the dynamic mechanisms governing the onset and drift of green tides, constitutes a formidable challenge in coastal disaster mitigation [7].

Currently, the rapid extraction of the spatial distribution of green tides is achievable through the utilization of optical, SAR, or unmanned aerial vehicle (UAV) imagery [8,9,10]. This process involves single- or multi-source data fusion methods [11], coupled with the application of spectral indices and machine learning techniques [1,12]. Simultaneously, a relationship model between spectral indices and biomass is established to determine the scale of green tide outbreaks by calculating the biomass of the green tide [13,14,15]. However, limitations imposed by cloud cover and the temporal resolution of satellites result in temporal gaps in the spatiotemporal distribution of monitored green tides [1]. Numerical simulation serves as an alternative tool for monitoring green tides, utilizing dynamics and environmental factors to simulate their drift trajectories [16]. Achieving high-precision simulation of green tide trajectories requires a comprehensive understanding of the mechanisms governing the drift transport.

In order to elucidate the driving mechanisms behind green tide drift, scholars have conducted extensive research using various methods, including teleconnection analysis, numerical simulations, and field investigation. Analysis of multi-year data on green tide disasters has revealed that interannual variations in the spatiotemporal distribution and drift paths of green tides are primarily influenced by changes in ocean currents induced by interannual variations in the wind field [17,18]. Numerical simulations further confirm that wind and ocean currents are the primary forces driving green tide drift. Lee et al. used the Regional Ocean Modeling System (ROMS) to investigate green tide transport along the Shandong Peninsula coast in 2008 [19]. Their findings showed southerly winds were the main force driving green tide drift along Jiangsu coast towards Shandong Peninsula in May, while easterly winds dominate in June. Wang et al. comprehensively examined green tide drift and dispersion in the Subei shoal using field experiments, remote sensing, and numerical simulations [7]. Their research demonstrated wind and ocean currents are the key factors influencing Ulva prolifera drift in the kelp farming region.

In recent years, the calculation of green tide drift velocity using multi-source remote sensing imagery has emerged as a novel approach to study the transport mechanisms of green tides. Studies typically rely on high-temporal-resolution imagery from satellites such as MODIS, GOCI, and GF-4 [20,21]. However, high-temporal-resolution imagery often comes with relatively lower spatial resolution, such as MODIS (250 m) and GOCI (500 m). This leads to the presence of numerous mixed pixels in the images, making the matching of green tide patches more challenging and consequently undermining the accuracy of drift velocity extraction [22]. To achieve more precise green tide drift velocity, researchers have turned to high-spatial-resolution images from polar-orbiting satellites, including COSMO-SkyMed SAR image pairs, HY-1 C/D image pairs, and Landsat 8 and Sentinel-2 image pairs, among others [22,23,24]. Amid its rapid advancement, UAV technology has been extensively used in assessing green tide drift velocity. Jiang et al. employed a UAV launched from research vessel platforms to collect data on the drift velocity of green tides at three distinct locations in the Yellow Sea [25]. The research revealed that the drift velocity of green tides over short time intervals is primarily influenced by the synergistic effects of wind and tidal currents, with speeds ranging from 0.26 to 0.44 m/s. Additionally, the drift direction undergoes significant changes within a single day.

The aforementioned studies primarily focused on conducting sensitivity analysis of numerical models and investigating teleconnections to elucidate green tide transport mechanisms [25,26]. There is, however, a deficiency in direct comparative studies of green tide drift velocity and dynamic factors across various time scales to analyze its transport mechanisms. Moreover, the leeway coefficient describes the force exerted by the wind on the exposed portion of objects floating on the ocean surface, significantly impacting the trajectories of green tide drift [27]. In numerical models, the leeway coefficient is usually set empirically or determined through sensitivity experiments [28]. With the growing availability of remote sensing data, the integration of this information with physical mechanism models offers great potential for accurately quantifying the leeway coefficient. Geostationary Ocean Color Imager-II (GOCI-II), succeeding GOCI, is a geostationary satellite sensor equipped with a spatial resolution of 250 m per hour, covering the time period from 7:16 a.m. to 4:16 p.m. Beijing time, making it an ideal source for monitoring green tide transport. Therefore, this study is based on the green tide drift velocities extracted from GOCI-II images at different time scales, in conjunction with oceanic currents and wind speeds, to investigate the green tide transport mechanisms and assess the feasibility of estimating leeway coefficient for moderate-resolution geostationary images.

2. Materials and Methods

2.1. Study Area

The study area in the South Yellow Sea, with geographic coordinates spanning from 32°N to 37°N in latitude and 119°E to 124°E in longitude, is regarded as the primary area for the origin, development, and outbreak of green algal blooms, as illustrated in Figure 1a. This area is characterized by the dominance of an amphidromic system associated with the M2 tide (Figure 1b), with an amphidromic point located at approximately 34°40′N [29,30]. Figure 1c,d depict the tidal current patterns at two specific times, as calculated using the Copernicus Marine Environment Monitoring Service (CMEMS) ocean current product, detailed in Section 2.3. The intensity of tides varies noticeably in space, with the intensity being significantly greater south of 34°40′N compared to the northern region. Furthermore, tidal currents flow in a north–south direction in the region south of 34°40′N, while they flow in an east–west direction in the northern region. Over the past decade, there has been a notable increase in the seawater temperature in this region, especially during the summer season, with most areas experiencing a temperature rise of over 1 °C [31]. Additionally, this area is characterized by a high degree of eutrophication, creating ideal conditions for the outbreak of algal blooms [32,33].

2.2. Image Acquisition and Processing

The green tide drift velocity data analyzed in this study were derived from GOCI-II images. High-quality GOCI-II Rayleigh-corrected reflectance ${\mathrm{R}}_{\mathrm{rc}}$ data from the periods of green tide outbreaks in 2021 and 2023 were acquired from the website (https://nosc.go.kr/opendap/GOCI-II, accessed on 25 May 2024). Detailed information about the acquired images is presented in Table 1. These images first undergo geometric correction using a geographic lookup table to ensure precise alignment of each pixel with its corresponding latitude and longitude coordinates [34]. Pixels contaminated by land and clouds are then eliminated through visual interpretation, incorporating the ultraviolet band at 380 nm [35]. Subsequently, the distribution of the green tide is derived using spectral indices. Finally, based on the extracted green tide, the maximum cross-correlation (MCC) method is employed to ascertain pixel displacement within green tide patches over various time intervals, thereby estimating their velocity [36].

2.2.1. Green Tide Distribution Extraction

Several spectral indices are used to extract green tide information, including the traditional vegetation index (NDVI, EVI, DVI, etc.), floating algae index (FAI), Virtual-Baseline Floating Macroalgae Height (VB-FAH), and alternative floating algae index (AFAI) [37,38,39,40]. Among these, the FAI is particularly effective for extraction green tides as it can tolerate aerosols, thin clouds, and stray light [1]. However, due to the absence of the shortwave infrared band on the GOCI-II satellite, this study utilized the AFAI index. Green tide distribution was determined by distinguishing algae and non-algae pixels based on a threshold derived from the spectral angle index (SAI) image [41]. The SAI helps mitigate false detections caused by environmental factors. The calculation of AFAI is as follows:

(1)$\begin{array}{l}\mathrm{AFAI}={\mathrm{R}}_{\mathrm{rc}}({\mathsf{\lambda }}_{745})-{\mathrm{R}}_{\mathrm{rc}}^{\prime }({\mathsf{\lambda }}_{745})\\ {\mathrm{R}}_{\mathrm{rc}}^{\prime }({\mathsf{\lambda }}_{745})={\mathrm{R}}_{\mathrm{rc}}({\mathsf{\lambda }}_{660})+{[\mathrm{R}}_{\mathrm{rc}}({\mathsf{\lambda }}_{865})-{\mathrm{R}}_{\mathrm{rc}}({\mathsf{\lambda }}_{745})]\times {\displaystyle \frac{{\mathsf{\lambda }}_{745}-{\mathsf{\lambda }}_{660}}{{\mathsf{\lambda }}_{865}-{\mathsf{\lambda }}_{660}}}\end{array}$

where ${\mathsf{\lambda }}_{660}$, ${\mathsf{\lambda }}_{745}$, and ${\mathsf{\lambda }}_{865}$ represent wavelengths at 660 nm, 745 nm, and 865 nm, respectively. Subsequently, ${\mathrm{SAI}}_{\mathrm{AFAI}}$ is derived from the AFAI image:

(2)${\mathrm{SAI}}_{\mathrm{AFAI}}=\mathrm{AFAI}-{\mathrm{Ref}}_{\mathrm{AFAI}}$

where ${\mathrm{Ref}}_{\mathrm{AFAI}}$ represents the median value obtained from the AFAI image after applying a 51 × 51 median filter window, following the criteria specified by Jin et al. (2018) [42]. Through visual interpretation, the threshold value for distinguishing algae pixels from non-algae pixels in the SAI image is determined. Pixels exceeding this threshold are identified as the green tide. Due to variations in seawater characteristics and the different stages of the green tide outbreak, this threshold value fluctuates between 0.01 and 0.05 [33].

2.2.2. Green Tide Drift Velocity Extraction

This study selects the MCC method to extract the green tide drift velocity due to its simplicity and widespread use in extracting sea ice and ocean current velocities [23,36,43]. The MCC method, a template matching technique, identifies the maximum correlation coefficient between two images to calculate the displacement of the target [36]. Assuming that there is a ($\mathrm{m},\mathrm{n}$) pixel displacement between two images, the cross-correlation coefficient ($\mathrm{k}$) can be calculated as:

(3)$\mathrm{k}={\displaystyle \frac{{\displaystyle {\sum }_{\mathrm{i}}{\displaystyle {\sum }_{\mathrm{j}}[\mathrm{f}(\mathrm{i},\mathrm{j})-\overline{\mathrm{f}}]}}[\mathrm{g}(\mathrm{i}+\mathrm{m},\mathrm{j}+\mathrm{n})-\overline{\mathrm{g}}]}{{\left[{\displaystyle {\sum }_{\mathrm{i}}{\displaystyle {\sum }_{\mathrm{j}}{[\mathrm{f}(\mathrm{i},\mathrm{j})-\overline{\mathrm{f}}]}^2{\displaystyle {\sum }_{\mathrm{i}}{\displaystyle {\sum }_{\mathrm{j}}{[\mathrm{g}(\mathrm{i}+\mathrm{m},\mathrm{j}+\mathrm{n})-\overline{\mathrm{g}}]}^2}}}}\right]}^{1/2}}}$

where $\mathrm{f}(\mathrm{i},\mathrm{j})$ and $\mathrm{g}(\mathrm{i}+\mathrm{m},\mathrm{j}+\mathrm{n})$ are the $\mathrm{SAI}$ image pixel values on the first and second image, respectively. $\overline{\mathrm{f}}$ and $\overline{\mathrm{g}}$ are the average values of the corresponding images. The template size is determined using green tide patch contour information obtained through morphological operations on the binary green tide image [24]. To ensure the reliability of the results and obtain sufficient matching patches, the threshold of the MCC method is set to 0.5 for intervals ranging from 1 to 9 h, and 0 for 25 h time intervals. Additionally, all extraction results undergo rigorous validation through visual interpretation to eliminate erroneous matches.

The extraction of green tide drift velocity typically involves high-spatial-resolution satellites, with intervals of less than 1 h, utilizing image pairs from satellites like Landsat 8 and Sentinel-2. Specifically, the time interval for morning and afternoon satellite image pairs, such as MODIS, is 3 h, while GOCI satellite data allows for a maximum time interval of 7 h within a single day. Moreover, continuous and high-quality satellite images can fulfill the requirements of a 25 h interval, covering two tidal cycles to mitigate the impact of tidal currents. Therefore, four intervals (1, 3, 7, and 25 h) are selected to analyze the driving mechanism of green tide transport and determine the leeway coefficient.

2.3. Dynamic Factor Data

Concurrently, ocean currents and wind speed are acquired to elucidate the driving mechanisms behind the drift of the green tide. Ocean currents are sourced from the Global Analysis and Forecast products (GLOBAL_ANALYSISFORECAST_PHY_001_024), which are available from the CMEMS website (https://marine.copernicus.eu/, accessed on 25 May 2024). This dataset includes Stokes drift, geostrophic plus Ekman currents, tidal currents, and the sum of these three currents in both meridional and zonal directions. It features a spatial resolution of 1/12°, a temporal resolution of 1 h, and covers depths ranging from 0 m to 5500 m across 50 discrete layers. Additional details, including information on the validation of CMEMS products as outlined in the quality document, are available on the CMEMS website and in the technical atlas [44].

Wind speed is sourced from the Fifth Generation of Global Climate and Weather Reanalysis product dataset, ERA5, available at the European Centre for Medium-Range Weather Forecasts website (https://cds.climate.copernicus.eu/, accessed on 25 May 2024). ERA5 offers an enhanced spatiotemporal resolution, with a spatial resolution of 0.25° and a temporal resolution of 1 h. Numerous studies have affirmed the applicability of ERA5 data to the study area [45,46]. Furthermore, ocean currents and wind speed are interpolated to the location and time of each green tide patch for subsequent analysis.

2.4. Leeway Model

The leeway model characterizes the movement of floating objects at sea under the influence of wind and ocean current [47]. In the case of green tides floating on the sea surface, three main factors influence their migration: near-surface currents, wave-induced Stokes drift, and wind-induced drift speed [47,48]. Consequently, the movement of floating Ulva prolifera can be expressed as following:

(4)${\mathsf{\nu }}_{\mathrm{u}}={\mathsf{\nu }}_{\mathrm{o}}+{\mathsf{\nu }}_{\mathrm{s}}+\mathsf{\alpha }\times {\mathsf{\nu }}_{10}$

where ${\mathsf{\nu }}_{\mathrm{u}}$ is the Ulva prolifera drift velocity, ${\mathsf{\nu }}_{\mathrm{o}}$ is the near-surface ocean current, ${\mathsf{\nu }}_{\mathrm{s}}$ is the Stokes drift, ${\mathsf{\nu }}_{10}$ is the 10 m height wind velocity, and $\mathsf{\alpha }$ is the leeway coefficient. When the velocities of ${\mathsf{\nu }}_{\mathrm{u}}$, ${\mathsf{\nu }}_{\mathrm{o}}$, ${\mathsf{\nu }}_{\mathrm{s}}$, and ${\mathsf{\nu }}_{10}$ are known, the leeway coefficient can be directly calculated as:

(5)$\mathsf{\alpha }=({\mathsf{\nu }}_{\mathrm{u}}-{\mathsf{\nu }}_{\mathrm{o}}-{\mathsf{\nu }}_{\mathrm{s}})/{\mathsf{\nu }}_{10}$

The leeway coefficient is calculated separately for the meridional and zonal directions. In the following, the contribution of Stokes drift to green tide drift is defined as the Stokes coefficient, where its value is determined by the ratio of ${\mathsf{\nu }}_{\mathrm{s}}/{\mathsf{\nu }}_{10}$.

3. Results

3.1. Characteristics of Green Tide Transport at Various Time Intervals

The distribution of green tides, extracted from the GOCI-II images at different time intervals, is shown in Figure 2. As the time intervals increase, the movement of the patches become more pronounced, accompanied by a decrease in similarity between these patches. The green tide drift velocities extracted using the MCC method at corresponding time intervals are depicted in Figure 3. However, the 25 h interval is not shown due to the limited number of matched patches. With increasing time intervals, the number of matched green tide patches noticeably decreases, emphasizing the differences in drift velocities between patches. This phenomenon primarily arises from the limited movement of patches over short time intervals. Furthermore, the spatial resolution of GOCI-II at 250 m means that portions with a movement distance less than one pixel are not displayed, resulting in similar drift velocities being extracted for patches with less noticeable movement differences. Longer time intervals magnify these differences, accentuating the varying drift velocities between different patches. However, increasing the time intervals results in changes in the shape of patches and increases the likelihood of cloud influence, consequently reducing the number of matched patches.

In the meridional direction, there is a noticeable difference in the drift velocity at 34°40′N, while in the zonal direction, drift velocities show an increasing trend with the latitude. To the region south of 34°40′N, the meridional drift velocity decreases with rising latitude, whereas in the region north of 34°40′N, the meridional drift velocity increases with latitude. Moreover, the smallest meridional drift velocity is observed near the 34°40′N line.

Furthermore, Figure 4 shows the rose diagram of all green tide drift velocities extracted at intervals of 1, 3, and 7 h, providing a more intuitive presentation of the drift direction of the green tide. As shown in Figure 4, the results further confirm significant regional variations in the direction of green tide drift. The predominant drift direction of green tides is meridional in the region south of 34°40′N and zonal in the region north of 34°40′N. Additionally, with an increase in the time interval, the drift velocity of the green tide tends to decrease.

3.2. Analysis of Driving Forces in Green Tide Drift Transport

To investigate the driving mechanisms behind the green tide drift transport, 50 patches were randomly selected from the drift velocity collected at 1, 3, and 7 h intervals in 2021 and 2023. Among these, only patches exhibiting evident movement within the 1 h interval was selected, and 25 patches were specifically chosen from both the northern and southern regions of 34°40′N. This analysis aims to elucidate the main factors influencing green tide migration by directly comparing the drift velocity of these patches with the ocean currents and wind contribution.

A cumulative bar chart illustrating green tide drift velocities under different ocean currents and wind speeds at 1, 3, and 7 h intervals is presented in Figure 5. The direct contribution of wind to the green tide is quantified by multiplying the wind speed by 2%, which represents the default leeway coefficient for the drift models [49]. The results reveal significant variations in the contribution of tidal forces. In the meridional direction, south of 34°40′N, the transport of green tide patches is predominantly controlled by the tide at 1, 3, and 7 h intervals. Conversely, in the north region of 34°40′N, where the predominant migration of green tides is westward and the drift velocities in the meridional direction are weak, no discernible dominant driving force in the meridional direction is observed. In the zonal direction, tides dominate the westward drift of green tide patches at 1 and 3 h intervals, while at 7 h intervals, the influence of wind and the Ekman plus geostrophic components becomes more pronounced. The Ekman plus geostrophic components exhibit significant spatiotemporal variations across different time intervals, with a greater contribution observed over longer durations and zonal direction. The contribution of Stokes drift is negligible, with minimal values that are not noticeable in the effects at 1, 3, and 7 h intervals. A notable contrast exists between the cumulative sum of each oceanic current component and the wind speed multiplied by 2%, as compared to the green tide drift velocity. Further discussion on this phenomenon will be provided in the Section 4.

Furthermore, Figure 6 illustrates the contributions of ocean currents and wind speed to the green tide drift velocity over a 25 h interval. As illustrated in Figure 6, the average value of the tidal current in a 25 h interval, often referred to as the tidal residual current, is not zero [50]. For most samples, the tidal residual current is less than 0.01 m/s, although some areas exceed this threshold, as seen in samples 14, 15, 16, and 19 in Figure 6b. Compared to the Stokes drift, the combined effects of Ekman and geostrophic currents, and wind forces, which are typically several 0.01 m/s, the impact of the tidal residual current is minimal. For instance, in sample 19 from Figure 6b, which has the largest tidal residual current, the tidal residual current is 0.016 m/s, whereas the combined force of the other currents and wind is 0.176 m/s. This indicates that the tidal residual current accounts for less than 1/10 of the total force, and in most cases, the contribution of the tidal residual current to the total driving force is significantly less than this value. Therefore, the net transport of the green tide over the 25 h interval is primarily driven by currents and wind speeds other than the tidal residual current. Additionally, compared with the results shown in Figure 5, the drift speed of the green tide over a 25 h interval aligns more closely with the cumulative sum of each current component and the wind speed multiplied by 2%, compared to the 1, 3, and 7 h intervals. For instance, in Figure 6a, samples 9, 16, and 25, and in Figure 6b, samples 18, 19, and 22, show that the values of the green tide drift speed are consistent with the cumulative sum of each current component and the wind speed multiplied by 2%. This result underscores the justification for selecting a 2% leeway coefficient to represent the direct force of wind on the green tide.

3.3. Estimation of Leeway Coefficient

From the above analysis, it is evident that the green tide drift velocity collected at 25 h intervals is suitable for estimating the leeway coefficient because its drift velocity closely approximates the combined contribution of ocean current and wind speed multiplied by 2%. Additionally, the Yellow Sea experiences semidiurnal tides, with a tidal cycle lasting approximately 12.4 h. Therefore, a 25 h period corresponds almost exactly to two tidal cycles. Being a multiple of the tidal period, the 25 h interval is optimal for analyzing green tide drift velocity, as it accounts for the balancing effects of tidal flood and ebb currents. The contribution of Stokes drift and wind to the green tide drift at the 25 h interval is depicted in Figure 7. According to Figure 7, it is apparent that the contribution of Stokes drift to green tide drift is closely linked to the wind speed. The Stokes drift coefficient becomes positive when the wind speed exceeds 1 m/s and exhibits an increasing trend with higher wind speeds. It fluctuates around 1% when the wind speed surpasses 2 m/s. The leeway coefficient exhibits significant fluctuations, with smaller values at higher wind speeds and vice versa. Notably, the fluctuation is more pronounced in the meridional direction compared to the zonal direction.

The analysis is further refined by excluding samples with lower wind speeds. Specifically, a wind speed threshold of 2 m/s or higher is imposed for 1 h, 3 h, and 7 h intervals, while a threshold of 1 m/s or higher is applied for the 25 h interval. Figure 8 and Table 2 present the contribution of Stokes drift and wind to the green tide drift at different time intervals. The results indicate significant variations and fluctuations in the contribution of Stokes drift and wind at different time intervals. Regarding the Stokes coefficient, distinct directional disparities are observed within the time intervals of 1 h, 3 h, and 7 h, wherein the zonal direction demonstrates a higher value than the meridional direction. Conversely, for the leeway coefficient, significant directional disparities are absent except within the 7 h interval. Furthermore, as the time intervals increase, both the estimated value threshold range and deviation of both Stokes and leeway coefficients decrease significantly.

Compared with the preceding time intervals, the outcomes obtained during the 25 h interval demonstrate substantial enhancement. The estimated Stokes coefficient exhibits minimal variability, with values of estimated coefficients in both meridional and zonal directions showing similarity. In the meridional direction, the Stokes coefficient primarily falls between 0.9% and 1.2%, while in the zonal direction, it mainly ranges from 0.8% to 1.05%. This discrepancy may be linked to variations in wind speed magnitude and the location of green tide patches. The variations in the estimated leeway coefficient are significantly reduced, and its meridional and zonal direction values are similar. The average value of the leeway coefficient is 1.8%, consistent with previous studies [28,49,50], while the average value of the Stokes drift coefficient is 0.85%, slightly lower than the default parameter of 1.2% in the drift model. This finding indicates that extracting green tide drift velocity at the 25 h interval is feasible for calculating the leeway coefficient.

4. Discussion

4.1. Characteristics and Driving Forces of Green Tide Transport across Different Time Intervals

According to the results depicted in Figure 3 and Figure 4, a distinct demarcation line at 34°40′N is evident in the extracted green tide drift velocity at time intervals of 1, 3, and 7 h. South of this boundary, patches predominantly drift in a meridional direction, whereas north of it, the primary drift direction is zonal. This corresponds to the observed directional disparity between the northern and southern regions of the tidal current at 34°40′N (Figure 1c,d). Notably, 34°40′N coincides with the latitude of the amphidromic point of the M2 tide (Figure 1b). Hence, to investigate the underlying reasons for this phenomenon, Figure 9 presents the rose diagram of the tidal currents corresponding to the patches depicted in Figure 4. The results indicate a strong alignment between the direction of green tide drift and the direction of the tidal current at these time intervals, suggesting that fluctuations in tidal currents induce variations in the transport direction of the green tide. Moreover, this indirectly implies that tidal currents predominantly govern the short-term drift of green tide.

By directly comparing the green tide drift velocity with the contribution of each dynamic factor (Figure 5), it becomes evident that tides exert a significant influence on green tide transport at time intervals of 1 and 3 h. At 7 h intervals, tidal currents continue to play a crucial role in the meridional direction south of 34°40′N. In contrast, in the region north of 34°40′N, the dominant factors are the wind and ocean currents. Combining the results from Figure 4 and Figure 9, it can be inferred that the drift transport of green tide is primarily governed by tidal currents in the short term.

As shown in Figure 10, the drift direction of the green tide generally aligns with the geostrophic and Ekman current components over the 25 h interval, albeit with slightly stronger drift speeds attributed to wind effects. This finding supports previous studies indicating that the transport of Yellow Sea green tide disturbances can be simulated using only geostrophic and Ekman current components [51]. Consequently, over longer time intervals with weakened tidal effects (Figure 6), the net transport of green tide is predominantly influenced by wind and ocean currents. This is consistent with previous research results indicating that green tides are mainly influenced by tidal currents on short time scales, while they are affected by wind and local ocean currents on time scales greater than 12 h [26].

4.2. Leeway Coefficient Error Assessment

The estimation of the leeway coefficient is subject to various sources of error, including the inaccuracies in estimating green tide drift velocity and errors in the wind and oceanic currents [52]. To evaluate the accuracy of GOCI-II in extracting green tide drift velocity, Figure 11 illustrates the comparison between green tide velocity extraction using high-spatial-resolution satellite images and the results obtained from GOCI-II. The results indicate that, at both 1 h and 3 h intervals, the error in extracting the displacement of green tide patches using the MCC method based on GOCI-II images is approximately 1 pixel. Moreover, the accuracy of green tide drift velocity extracted at 3 h intervals notably surpasses that at 1 h intervals. Consequently, increasing the time interval used to calculate green tide drift velocity patches can notably mitigate the extraction error based on GOCI-II images, despite presenting a heightened challenge in accurately matching green tide drift patches when longer time intervals are adopted. In the context of the 25 h interval, the error in green tide drift velocity is approximately 0.003 m/s, a value deemed negligible for estimating the leeway coefficient. Furthermore, the distribution characteristics of green tide patches in high-spatial-resolution images at 25 h intervals are depicted in Figure 12, confirming the reliability of the green tide drift velocity extracted at 25 h intervals.

At the time intervals of 1 h, 3 h, and 7 h, a noticeable disparity exists between the green tide drift velocity and the cumulative sum of the ocean currents and wind speed multiplied by 2%, as depicted in Figure 5. This discrepancy results in significant uncertainty in the estimated leeway coefficient (Figure 8 and Table 2). Furthermore, regardless of the stringency of the filtering conditions, the estimated leeway coefficients within these time intervals predominantly fall within the range of −6% to 10%, accompanied by considerable standard deviation values. However, as the influence of tides decreases, the difference between the green tide drift velocity and the cumulative sum of the ocean currents and wind speed multiplied by 2% is significantly reduced (Figure 6). Consequently, there is a notable enhancement in the estimated leeway coefficient, demonstrating consistent directionality. This implies that employing green tide drift velocity and corresponding oceanic currents and wind data at tidal periodic intervals can mitigate the uncertainty in the leeway coefficient estimation. Furthermore, the estimated leeway coefficient shows a closer agreement with previous studies, affirming the effectiveness of this approach in estimating the leeway coefficient [28,49,52]. As demonstrated earlier, the drift transport of green tide is primarily dominated by tides over short time intervals, whereas the net transport of green tides over tidal periodic intervals is mainly affected by currents and wind, apart from tides. Hence, it can be inferred that the error in tidal current and green tide drift velocity are the primary contributing factors to this disparity.

In addition, temporal and spatial interpolation of ocean currents and wind speed to match the location and time of green tide patches, coupled with the oversight of dynamic changes in the position of patches, may lead to discrepancies in oceanic currents and wind speed beyond inherent data errors, thus also contributing to the error in the estimation of the leeway coefficient [52]. The leeway coefficient estimated based on green tide drift velocity, wind speed, and ocean currents at the 25 h interval closely aligns with previous findings. Moreover, the estimated Stokes coefficient derived from this approach is slightly lower than the prescribed value in drift models, indicating the reliability of the wind speed and ocean currents utilized in this study. Furthermore, it also suggests that errors in spatiotemporal interpolation may have minimal impact.

5. Conclusions

This study employs multiple remote sensing images from the GOCI-II satellite to investigate the driving mechanisms of green tide drift and evaluate the feasibility of estimating the leeway coefficient from moderate-resolution images at various time intervals. The results indicate that, at short time intervals, the movement of green tide patches is primarily influenced by the tidal currents. Conversely, over tidal periodic intervals, the net movement of patches is dominated by the wind and ocean currents, particularly Ekman plus geostrophic currents. Notably, a distinct difference in the drift directions of green tides at short time intervals is observed between the northern and southern regions of 34°40′N. The difference is primarily attributed to variations in tidal currents, with an amphidromic point near 34°40′N leading to the smallest drift velocity in the meridional direction of this area.

The leeway coefficient estimated with 1, 3, and 7 h intervals exhibits notable variation and significant directional dependence. In contrast, the leeway coefficient estimated with a 25 h interval shows minimal variation and no apparent directional difference, which is consistently around 1.8%. This demonstrates the feasibility of estimating leeway coefficients at tidal periodic intervals for moderate-resolution images. Moreover, the coefficient of Stokes drift is found to be smaller than the typical values used in the drift models, with wind serving as a proxy for Stokes drift with a coefficient of about 0.85%. In summary, this study enhances our comprehension of the transport mechanisms of green tides across diverse temporal scales and provides a basis for the selection of time intervals for estimating the leeway coefficient from moderate-resolution images.

Author Contributions

Conceptualization, M.J. and C.Z.; methodology, M.J.; software, M.J. and X.D.; validation, M.J.; writing—original draft preparation, M.J.; writing—review and editing, C.Z., X.D. and J.Z.; visualization, M.J. and X.D.; supervision, C.Z. and J.Z.; project administration, C.Z.; funding acquisition, C.Z. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

Publicly available datasets were analyzed in this study. These data can be found here: [https://nosc.go.kr/opendap/GOCI-II/, accessed on 25 May 2024; https://marine.copernicus.eu/, accessed on 25 May 2024; https://cds.climate.copernicus.eu/, accessed on 25 May 2024].

Acknowledgments

The authors appreciate the Korea Ocean Satellite Center (KOSC) for providing GOCI-II L2 data, the Copernicus Marine Environment Monitoring Service (CMEMS) for providing Global Ocean Physics Analysis and Forecast data, the European Centre for Medium-Range Weather Forecasts (ECMWF) for providing ERA5 reanalysis data.

Conflicts of Interest

The authors declare no conflicts of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. The map delineates the study area in the Yellow Sea where green tides reoccur. (a,b) The location of the study area and the M2 cotidal graph (the values of the contour lines represent the phase in degrees), respectively. (c,d) The average tidal current patterns from 10:30 to 13:30 local time during the periods of 4–10 June and 11–19 June 2021, respectively. The blue dashed line, positioned at 34°40′N, serves as the boundary for the drift velocity of the green tide.

Figure 2. The distribution of green tides extracted at various time intervals: (a) 1 h interval, (b) 3 h interval, (c) 7 h interval, and (d) 25 h interval.

Figure 3. Green tide drift velocities extracted at different time intervals is depicted in subplots (a–f). Subplots (a–d) illustrate the result for 1 h intervals, (b–e) for 3 h intervals, and (c–f) for 7 h intervals. The top panel represents the meridional direction, while the bottom panel represents the zonal direction. The red dashed line indicates 34°40′N, and the blue line is the trend line. North and east direction values are positive.

Figure 4. The rose diagram of green tide drift velocities at 1, 3, and 7 h time intervals is depicted in subplots (a–f). The upper panel corresponds to the area south of 34°40′N, while the bottom panel corresponds to the region north of 34°40′N. Colors represent the speed, while the length of the bars indicates the frequency.

Figure 5. Cumulative bar chart of green tide drift velocities under varying ocean currents and wind speeds at 1, 3, and 7 h intervals. Ulva represents the green tide drift velocity, while the names of ocean current components (including tides, Stokes, and Ekman plus geostrophic) represent their velocities in either the zonal or meridional directions. The wind multiplied by 2% represents the direct contribution of wind to the green tide. Blue triangles indicate the central latitude of the green tide patches. The left panel (a,c,e) and right panel (b,d,f) represent the meridional and zonal directions, respectively. North and east direction values are positive.

Figure 6. Cumulative bar chart of green tide drift velocities under varying ocean currents and wind speeds at 25 h intervals. The left panel (a) and right panel (b) represent the meridional and zonal directions, respectively. The legend in the figure is consistent with that in Figure 5. N represents the sample size.

Figure 7. Stokes and leeway coefficients relative to wind calculated based on green tide drift velocities extracted at 25 h intervals. Mn represents mean value and SD represents standard deviation. The top panel (a) represents the meridional direction, while the bottom panel (b) represents the zonal direction.

Figure 8. Box plots of the estimated leeway and Stokes coefficients at various time intervals. The yellow box represents the meridional direction, while the green box represents the zonal direction. The top panel shows the contribution of Stokes drift, while the bottom panel shows the contribution of wind.

Figure 9. The rose diagram of tidal currents at 1, 3, and 7 h time intervals is depicted in subplots (a–f). The top panel corresponds to the area south of 34°40′N, while the bottom panel corresponds to the region north of 34°40′N. Colors represent the speed, while the length of the bars indicates the frequency.

Figure 10. Wind rose of velocities for the green tide (a), Ekman plus geostrophic current (b), and wind (c) at 25 h intervals. Colors represent the speed, while the length of the bars indicates the frequency.

Figure 11. Comparison of green tide drift velocity extracted from high-spatial-resolution images with GOCI-II images. (a,b) Comparison of green tide drift velocity extracted at 1 h intervals by GOCI-II and the Landsat 8 and Sentinel-2 image pair on 7 June 2021. (c,d) Comparison of green tide drift velocity extracted at 3 h intervals by GOCI-II and HY-1C/D on 6 June 2021. South and east direction values are positive.

Figure 12. The distribution of green tides at 25 h intervals extracted from various satellite images. The left panel displays the GOCI-II image, while the right panel shows images from Landsat 8 and Sentinel-2.

Table 1

Details of the dates of GOCI-II images used in this study.

Year	Date
2021	4 June, 5 June, 6 June, 7 June, 19 June, 20 June, 22 June, 23 June, 1 July, 9 July, 10 July
2023	3 June, 6 June, 8 June, 9 June, 11 June, 12 June, 13 June, 14 June, 15 June, 22 June, 24 June, 27 June, 5 July, 9 July, 10 July

Table 2

Mean and standard deviation of Stokes drift and leeway coefficients calculated at different time intervals.

Coefficient	1 h		3 h		7 h		25 h
	Meridional	Zonal	Meridional	Zonal	Meridional	Zonal	Meridional	Zonal
Stokes (%)	0.38 ± 0.41	0.76 ± 0.38	0.47 ± 0.40	0.77 ± 0.33	0.37 ± 0.39	0.74 ± 0.42	0.88 ± 0.38	0.85 ± 0.24
Leeway (%)	2.22 ± 6.25	2.45 ± 6.25	1.96 ± 3.62	2.25 ± 4.42	0.86 ± 3.60	3.49 ± 3.24	1.80 ± 1.31	1.80 ± 0.64
Sample	1003	873	319	357	151	81	22	23

References

1. Hu, C.; Qi, L.; Hu, L.; Cui, T.; Xing, Q.; He, M.; Wang, N.; Xiao, Y.; Sun, D.; Lu, Y. et al. Mapping Ulva prolifera green tides from space: A revisit on algorithm design and data products. Int. J. Appl. Earth Obs. Geoinf.; 2023; 116, 103173. [DOI: https://dx.doi.org/10.1016/j.jag.2022.103173]

2. Liu, D.; Keesing, J.K.; He, P.; Wang, Z.; Shi, Y.; Wang, Y. The world’s largest macroalgal bloom in the Yellow Sea, China: Formation and implications. Estuar. Coast. Shelf Sci.; 2013; 129, pp. 2-10. [DOI: https://dx.doi.org/10.1016/j.ecss.2013.05.021]

3. Zhang, Y.; He, P.; Li, H.; Li, G.; Liu, J.; Jiao, F.; Zhang, J.; Huo, Y.; Shi, X.; Sun, R. et al. Ulva prolifera green-tide outbreaks and their environmental impact in the Yellow Sea, China. Natl. Sci. Rev.; 2019; 6, pp. 825-838. [DOI: https://dx.doi.org/10.1093/nsr/nwz026] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34691936]

4. Li, H.; Zhang, Y.; Tang, H.; Shi, X.; Rivkin, R.B.; Legendre, L. Spatiotemporal variations of inorganic nutrients along the Jiangsu coast, China, and the occurrence of macroalgal blooms (green tides) in the southern Yellow Sea. Harmful Algae; 2017; 63, pp. 164-172. [DOI: https://dx.doi.org/10.1016/j.hal.2017.02.006] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28366391]

5. Xing, Q.; Tosi, L.; Braga, F.; Gao, X.; Gao, M. Interpreting the progressive eutrophication behind the world’s largest macroalgal blooms with water quality and ocean color data. Nat. Hazards; 2015; 78, pp. 7-21. [DOI: https://dx.doi.org/10.1007/s11069-015-1694-x]

6. Xia, Z.; Yuan, H.; Liu, J.; Sun, Y.; Tong, Y.; Zhao, S.; Xia, J.; Li, S.; Hu, M.; Cao, J. et al. A review of physical, chemical, and biological green tide prevention methods in the Southern Yellow Sea. Mar. Pollut. Bull.; 2022; 180, 113772. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2022.113772] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35623218]

7. Wang, Z.; Xiao, J.; Yuan, C.; Miao, X.; Fan, S.; Fu, M.; Xia, T.; Zhang, X. The drifting and spreading mechanism of floating Ulva mass in the waterways of Subei shoal, the Yellow Sea of China–Application for abating the world’s largest green tides. Mar. Pollut. Bull.; 2023; 190, 114789. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2023.114789]

8. Jiang, X.; Gao, M.; Gao, Z. A novel index to detect green-tide using UAV-based RGB imagery. Estuar. Coast. Shelf Sci.; 2022; 245, 106943. [DOI: https://dx.doi.org/10.1016/j.ecss.2020.106943]

9. Li, D.; Gao, Z.; Wang, Z. Analysis of the reasons for the outbreak of Yellow Sea green tide in 2021 based on long-term multi-source data. Mar. Environ. Res.; 2022; 178, 105649. [DOI: https://dx.doi.org/10.1016/j.marenvres.2022.105649]

10. Qi, L.; Wang, M.; Hu, C.; Holt, B. On the capacity of Sentinel-1 synthetic aperture radar in detecting floating macroalgae and other floating matters. Remote Sens. Environ.; 2022; 280, 113188. [DOI: https://dx.doi.org/10.1016/j.rse.2022.113188]

11. Xu, S.; Yu, T.; Xu, J.; Pan, X.; Shao, W.; Zuo, J.; Yu, Y. Monitoring and Forecasting Green Tide in the Yellow Sea Using Satellite Imagery. Remote Sens.; 2023; 15, 2196. [DOI: https://dx.doi.org/10.3390/rs15082196]

12. Cui, B.; Liu, M.; Chen, R.; Zhang, H.; Zhang, X. Anisotropic Green Tide Patch Information Extraction Based on Deformable Convolution. Remote Sens.; 2024; 16, 1162. [DOI: https://dx.doi.org/10.3390/rs16071162]

13. Hu, L.; Hu, C.; HE, M. Remote estimation of biomass of Ulva prolifera macroalgae in the Yellow Sea. Remote Sens. Environ.; 2017; 192, pp. 217-227. [DOI: https://dx.doi.org/10.1016/j.rse.2017.01.037]

14. Tian, W.; Wang, J.; Zhang, F.; Liu, X.; Yang, J.; Yuan, J.; Mi, X.; Shao, Y. The detection of green tide biomass by remote sensing images and in situ measurement in the Yellow Sea of China. Remote Sens.; 2023; 15, 3625. [DOI: https://dx.doi.org/10.3390/rs15143625]

15. Xing, Q.; Wu, L.; Tian, L.; Cui, T.; Li, L.; Kong, F.; Gao, X.; Wu, M. Remote sensing of early-stage green tide in the Yellow Sea for floating-macroalgae collecting campaign. Mar. Pollut. Bull.; 2018; 133, pp. 150-156. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2018.05.035] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30041302]

16. Qi, L.; Cheng, P.; Wang, M.; Hu, C.; Xie, Y.; Mao, K. Where does floating Sargassum in the East China Sea come from?. Harmful Algae; 2023; 129, 102523. [DOI: https://dx.doi.org/10.1016/j.hal.2023.102523] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37951622]

17. Zhang, G.; Wu, M.; Sun, X.; Zhao, D.; Xing, Q.; Liang, F. The inter-annual drift and driven force of Ulva prolifera bloom in the southern Yellow Sea. Oceanol. Limn. Sin.; 2018; 49, pp. 1084-1093. [DOI: https://dx.doi.org/10.11693/hyhz20180400093]

18. Keesing, J.K.; Liu, D.; Fearns, P.; Garcia, R. Inter-and intra-annual patterns of Ulva prolifera green tides in the Yellow Sea during 2007–2009, their origin and relationship to the expansion of coastal seaweed aquaculture in China. Mar. Pollut. Bull.; 2011; 62, pp. 1169-1182. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2011.03.040] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21497856]

19. Lee, J.H.; Pang, I.C.; Moon, I.J.; Ryu, J.H. On physical factors that controlled the massive green tide occurrence along the southern coast of the Shandong Peninsula in 2008: A numerical study using a particle-tracking experiment. J. Geophys. Res. Ocean.; 2011; 116, [DOI: https://dx.doi.org/10.1029/2011JC007512]

20. Chen, X.; Zhang, J.; Cui, T.; Song, P. Extraction of the green tide drift velocity in the Yellow Sea based on GF-4. Acta Oceanol. Sin.; 2018; 40, pp. 29-38. [DOI: https://dx.doi.org/10.3969/j.issn.0253-4193.2018.01.004]

21. Podlejski, W.; Berline, L.; Nerini, D.; Doglioli, A.; Lett, C. A new Sargassum drift model derived from features tracking in MODIS images. Mar. Pollut. Bull.; 2023; 188, 114629. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2023.114629] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36860021]

22. Liu, H.; Xing, Q.; Ding, J.; Liu, J.; Zheng, X.; Wu, L.; Li, L.; Lu, Y. High-resolution remote sensing of the transportation of floating macroalgae: Case studies with the Ulva prolifera green tide. Natl. Remote Sens. Bull.; 2023; 27, pp. 187-196. [DOI: https://dx.doi.org/10.11834/jrs.20235001]

23. Ciappa, A.; Pietranera, L.; Coletta, A.; Jiang, X. Surface transport detected by pairs of COSMO-SkyMed ScanSAR images in the Qingdao region (Yellow Sea) during a macro-algal bloom in July 2008. J. Mar. Syst.; 2010; 80, pp. 135-142. [DOI: https://dx.doi.org/10.1016/j.jmarsys.2009.02.010]

24. Ji, M.; Zhao, C.; Dou, X.; Wang, C.; Zhou, D.; Zhu, J. Identification and assessment of the drift velocity of green tides using the maximum cross-correlation method in the Yellow Sea. Mar. Pollut. Bull.; 2023; 194, 115420. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2023.115420]

25. Jiang, X.; Gao, Z.; Wu, X.; Wang, Y.; Ning, J. Estimation and analysis of the green-tide drift velocity using ship-borne UAV. Acta Oceanol. Sin.; 2021; 43, pp. 96-105. [DOI: https://dx.doi.org/10.12284/hyxb2021054]

26. Son, Y.B.; Choi, B.J.; Kim, Y.H.; Park, Y.G. Tracing floating green algae blooms in the Yellow Sea and the East China Sea using GOCI satellite data and Lagrangian transport simulations. Remote Sens. Environ.; 2015; 156, pp. 21-33. [DOI: https://dx.doi.org/10.1016/j.rse.2014.09.024]

27. Putman, N.F.; Goni, G.J.; Gramer, L.J.; Hu, C.; Johns, E.M.; Trinanes, J.; Wang, M. Simulating transport pathways of pelagic Sargassum from the Equatorial Atlantic into the Caribbean Sea. Prog. Oceanogr.; 2018; 165, pp. 205-214. [DOI: https://dx.doi.org/10.1016/j.pocean.2018.06.009]

28. Zhou, F.; Ge, J.; Liu, D.; Ding, P.; Chen, C.; Wei, X. The Lagrangian-based Floating Macroalgal Growth and Drift Model (FMGDM v1.0): Application to the Yellow Sea green tide. Geosci. Model Dev.; 2021; 14, pp. 6049-6070. [DOI: https://dx.doi.org/10.5194/gmd-14-6049-2021]

29. Chen, S.; Jiang, L.; Cheng, X.; Liao, G.; Gerkema, T. A physical perspective of recurrent water quality degradation: A case study in the Jiangsu coastal waters, China. J. Geophys. Res. Ocean.; 2023; 128, e2022JC019607. [DOI: https://dx.doi.org/10.1029/2022JC019607]

30. Fang, G. Tide and tidal current charts for the marginal seas adjacent to China. Chin. J. Oceanol. Limnol.; 1986; 4, pp. 1-16. [DOI: https://dx.doi.org/10.1007/BF02850393]

31. Xu, Y.; Xu, T. An evolving marine environment and its driving forces of algal blooms in the Southern Yellow Sea of China. Mar. Environ. Res.; 2022; 178, 105635. [DOI: https://dx.doi.org/10.1016/j.marenvres.2022.105635] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35644075]

32. Zhang, H.; Su, R.; Shi, X.; Zhang, C.; Yin, H.; Zhou, Y.; Wang, G. Role of nutrients in the development of floating green tides in the Southern Yellow Sea, China, in 2017. Mar. Pollut. Bull.; 2020; 156, 111197. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2020.111197] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32510359]

33. Xue, M.; Wu, M.; Zheng, L.; Liu, J.; Liu, L.; Zhu, S.; Liu, S.; Liu, L. Multi-Factors Synthetically Contribute to Ulva prolifera Outbreaks in the South Yellow Sea of China. Remote Sens.; 2023; 15, 5151. [DOI: https://dx.doi.org/10.3390/rs15215151]

34. Du, C.G.; Li, Y.M.; Wang, Q.; Liu, G.; Zheng, Z.B.; Mu, M.; Li, Y. Tempo-spatial dynamics of water quality and its response to river flow in estuary of taihu lake based on GOCI imagery. Environ. Sci. Pollut. Res.; 2017; 24, pp. 28079-28101. [DOI: https://dx.doi.org/10.1007/s11356-017-0305-7] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28994019]

35. Ding, N.; Shao, J.; Yan, C.; Zhang, J.; Qiao, Y.; Pan, Y.; Yuan, J.; Dong, Y.; Yu, B. Near-ultraviolet to near-infrared band thresholds cloud detection algorithm for TANSAT-CAPI. Remote Sens.; 2021; 13, 1906. [DOI: https://dx.doi.org/10.3390/rs13101906]

36. Yang, H.; Arnone, R.; Jolliff, J. Estimating advective near-surface currents from ocean color satellite images. Remote Sens. Environ.; 2015; 158, pp. 1-14. [DOI: https://dx.doi.org/10.1016/j.rse.2014.11.010]

37. Cui, T.; Zhang, J.; Sun, L.; Jia, Y.; Zhao, W.; Wang, Z.; Meng, J. Satellite monitoring of massive green macroalgae bloom (GMB): Imaging ability comparison of multi-source data and drifting velocity estimation. Int. J. Remote Sens.; 2012; 33, pp. 5513-5527. [DOI: https://dx.doi.org/10.1080/01431161.2012.663112]

38. Hu, C. A novel ocean color index to detect floating algae in the global oceans. Remote Sens. Environ.; 2009; 113, pp. 2118-2129. [DOI: https://dx.doi.org/10.1016/j.rse.2009.05.012]

39. Xing, Q.; Hu, C. Mapping macroalgal blooms in the Yellow Sea and East China Sea using HJ-1 and Landsat data: Application of a virtual baseline reflectance height technique. Remote Sens. Environ.; 2016; 178, pp. 113-126. [DOI: https://dx.doi.org/10.1016/j.rse.2016.02.065]

40. Qi, L.; Hu, C.; Barnes, B.B.; Lapointe, B.E.; Chen, Y.; Xie, Y.; Wang, M. Climate and anthropogenic controls of seaweed expansions in the East China Sea and Yellow Sea. Geophys. Res. Lett.; 2022; 49, e2022GL098185. [DOI: https://dx.doi.org/10.1029/2022GL098185]

41. Garcia, R.A.; Fearns, P.; Keesing, J.K.; Liu, D. Quantification of floating macroalgae blooms using the scaled algae index. J. Geophys. Res. Ocean.; 2013; 118, pp. 26-42. [DOI: https://dx.doi.org/10.1029/2012JC008292]

42. Jin, S.; Liu, Y.; Sun, C.; Wei, X.; Li, H.; Han, Z. A study of the environmental factors influencing the growth phases of Ulva prolifera in the southern Yellow Sea, China. Mar. Pollut. Bull.; 2018; 135, pp. 1016-1025. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2018.08.035]

43. Ciappa, A.; Pietranera, L.; Battazza, F. Perito Moreno Glacier (Argentina) flow estimation by COSMO SkyMed sequence of high-resolution SAR-X imagery. Remote Sens. Environ.; 2010; 114, pp. 2088-2096. [DOI: https://dx.doi.org/10.1016/j.rse.2010.04.014]

44. Dobler, D.; Maes, C.; Martinez, E.; Rahmania, R.; Gautama, B.G.; Farhan, A.R.; Dounias, E. On the fate of floating marine debris carried to the sea through the main rivers of Indonesia. J. Mar. Sci. Eng.; 2022; 10, 1009. [DOI: https://dx.doi.org/10.3390/jmse10081009]

45. Geng, S.; Shi, X.; Yue, X. Applicability analysis of the sea surface pressure and wind speed of ERA5 reanalysis data in the Bohai Sea and the Northern Huanghai Seas. Mar. Sci. Bull.; 2022; 24, pp. 1-16. [DOI: https://dx.doi.org/10.11840/j.issn.1001-6392.2023.02.005]

46. Li, J.; Zhu, Y.; Xu, J.; Yao, Y. A comparative study on the applicability of ERA-Interim and ERA5 reanalysis wind data in the coastal waters of China. Mar. Sci. Bull.; 2023; 42, pp. 260-271. [DOI: https://dx.doi.org/10.11840/j.issn.1001-6392.2023.03.003]

47. Dagestad, K.-F.; Röhrs, J. Prediction of ocean surface trajectories using satellite derived vs. modeled ocean currents. Remote Sens. Environ.; 2019; 223, pp. 130-142. [DOI: https://dx.doi.org/10.1016/j.rse.2019.01.001]

48. Wu, J.; Cheng, L.; Chu, S. Modeling the leeway drift characteristics of persons-in-water at a sea-area scale in the seas of China. Ocean Eng.; 2023; 270, 113444. [DOI: https://dx.doi.org/10.1016/j.oceaneng.2022.113444]

49. Jones, C.E.; Dagestad, K.-F.; Breivik, Ø.; Holt, B.; Röhrs, J.; Christensen, K.H.; Espeseth, M.; Brekke, C.; Skrunes, S. Measurement and Modeling of Oil Slick Transport. J. Geophys. Res. Ocean.; 2016; 121, pp. 7759-7775. [DOI: https://dx.doi.org/10.1002/2016JC012113]

50. Lee, S.H.; Beardsley, R.C. Influence of stratification on residual tidal currents in the Yellow Sea. J. Geophys. Res.; 1999; 104, pp. 15679-15701. [DOI: https://dx.doi.org/10.1029/1999JC900108]

51. Choi, J.G.; Kim, D.; Shin, J.; Jang, S.W.; Lippmann, T.C.; Jo, Y.H.; Cho, S.W. New diagnostic sea surface current fields to trace floating algae in the Yellow Sea. Mar. Pollut. Bull.; 2023; 195, 115494. [DOI: https://dx.doi.org/10.1016/j.marpolbul.2023.115494] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37703632]

52. Sutherland, G.; Soontiens, N.; Davidson, F.; Smith, G.C.; Bernier, N.; Blanken, H.; Schillinger, D.; Marcotte, G.; Röhrs, J.; Dagestad, K.-F. et al. Evaluating the leeway coefficient of ocean drifters using operational marine environmental prediction systems. J. Atmos. Ocean. Technol.; 2020; 37, pp. 1943-1954. [DOI: https://dx.doi.org/10.1175/JTECH-D-20-0013.1]