Detection and Characterization of Marine Ecotones Using Satellite-Derived Environmental Indicators

Abstract

The delimitation of an ecotone is an important reference for ecosystem conservation; however, the assessment of a marine ecotone from an ecological point of view represents a knowledge gap. The Yellow River Estuary (YRE) serves as both spawning and feeding grounds for numerous economically important organisms. Delineating the boundary of YRE and assessing the boundary change have great importance in maintaining its ecosystem health. This study attempts to apply a Moving Split Window (MSW) to determine marine boundary in YRE. Level 2 remote sensing satellite data spanning from 2012 to 2020 sourced from the Geostationary Ocean Color Imager (GOCI) were utilized. Chlorophyll-a, Chromophoric Dissolved Organic Matter (CDOM), and Total Suspended Solids (TSS) were employed as variables, with Squared Euclidean Distance (SED) serving as the determinant for identifying the marine ecological ecotone within the Yellow Estuary and its adjacent waters. Results indicate the following: (1) SED values exhibit distinct peaks and valleys, facilitating the accurate identification of marine ecotones via MSW. (2) Evident ecotones are observable in both the gate and coastal regions. (3) The influence range of TSS on the gate spans between 10 km and 14 km. In synthesis, the ensuing conclusions are drawn: MSW proves to be a reliable method for quantitatively determining ecotones in marine environments. Furthermore, MSW introduces a novel approach to the delineation of marine ecotones.

1. Introduction

Estuaries represent highly dynamic natural complexes where the confluence of rivers and oceans occurs. They serve as the most sensitive and active arenas for the interaction among the hydrosphere, lithosphere, biosphere, atmosphere, and anthroposphere [1,2,3,4]. As the second longest river in China, the Yellow River is the main founder of the geological changes in the Bohai Sea. Substantial quantities of nutrients are transported into the estuary and its surrounding marine ecosystem via Yellow River runoff. The discharge of organic carbon from the Yellow River runoff profoundly influences Lanzhou Bay and the northwest coast of the Yellow River Estuary (YRE) [5]. The temperate monsoon climate fosters interactions between the YRE and its adjacent waters, facilitating the formation of a stable ecosystem [6,7]. Nonetheless, accelerated climate change and escalating human demands have engendered significant environmental changes in the YRE. Consequently, the Yellow River and the delta region present formidable challenges, necessitating a thorough examination of the ecosystem in YRE.

The concept of ecotone was first introduced by Clements [8], referring to the transitional zone between adjacent communities, which aligns with early ecological studies focusing on terrestrial communities. With the rise of landscape ecology, research on the interaction processes and relationships between ecosystems has gained prominence [9]. Consequently, an ecotone was further defined as the transitional area between neighboring ecosystems. This refined concept emphasizes spatiotemporal scales and interactions between ecosystems, expanding the scope of study from terrestrial communities to transitional zones between land and water, sea and land, and marine systems. Additionally, the research focus has also shifted from the community level to the ecosystem level.

Ecotones perform diverse functions within ecosystems: (1) serving as corridors between ecosystems [10]; (2) acting as filters for components moving between ecosystems [10]; (3) providing sources of matter, energy, and organisms for adjacent ecosystems [10]; (4) absorbing or accumulating certain components of ecosystems [10]; and (5) offering habitats for edge species [10]. Given these critical roles, ecotones are increasingly recognized as vital indicators of ecosystem health. Internal ecological indicators within ecotones undergo substantial changes, fostering rich species and genetic diversity [11,12,13,14]. Research on ecotones would help to understand the trends of ecosystems [15].

Ecotones manifest across various scales, from eco-regions to biota areas and community scales [16]. The multiple ecotone scale determines the diversification of quantitative research technology [17]. Commonly employed methods for land boundary analysis include sorting methods [18], wavelet methods [19], and the wombling method [20]. The results of ecotone detection are also affected by different methods [20]. Meanwhile, for analyzing one-dimensional sample data, the Moving Split-Window (MSW) technique effectively delineates regions. MSW is widely used for researching the width and type of an ecotone. Whittaker first began researching MSW in 1960 [21]. However, compared with the current MSW, the distance coefficient of that time was similarity rather than dissimilarity; on the other hand, the results of the analysis were not reflected in real position. Mature MSW methods were first applied to soil science [22,23,24]. Since the 1980s, MSW has been increasingly used in vegetation science and later in animal ecology [25,26,27,28]. At present, MSW is mainly used to analyze satellite images and determine an ecotone according to the reflectivity of satellite images [29]. Based on MSW, Stanisci found that the forest line growing shrubs might move to higher altitudes over time, while without shrubs, the forest moved to lower altitudes [30]. Ibanez found that in the ecotone of tropical rainforests and sparse tree steppe, the boundary was alternating between stable and unstable [31]. Devi Choesin comparatively analyzed MSW and Detrended Correspondence Analysis (DCA) methods on the determination of ecotones [32]. Compared with DCA, MSW makes it easier to detect the position and width of an ecotone [32]. Körmöczi analyzed the boundary types of alkaline grassland communities in Hungary based on MSW and found that MSW can determine the type of ecotone well [33].

Marine ecotones, the transitional zones between adjacent marine ecosystems, exhibit higher primary productivity, biomass, and biodiversity, making them highly valuable for the ecosystem. However, a considerable gap exists in the current determination of marine ecotones, particularly in China, where the demarcation of marine ecosystems relies predominantly on functional attributes rather than ecological methods.

Marine ecosystems are based on seawater, which is constantly influenced by factors such as wind, waves, tides, and currents [34]. Unlike terrestrial ecosystems, the marine ecosystem’s perpetual and vigorous movement distinguishes it from terrestrial soil. Although field sampling is widely employed for identifying terrestrial ecotones, its application in oceanic environments is significantly hindered by the inherently dynamic and unstable environment of marine ecosystems. Moreover, the scarcity of long-term stable data poses a significant hurdle to marine ecotone research. Accordingly, this paper explores the application of the MSW method and satellite data for quantitatively studying marine ecotones within YRE, offering a novel research methodology and paradigm for marine ecological delineation.

The Geostationary Ocean Color Imager (GOCI) serves as the principal sensor aboard the Korean COMS satellite, inaugurated successfully in 2010, thereby marking a milestone as the pioneer geostationary sea color satellite [35]. The monitoring scope of GOCI covers the Bohai Sea, the Yellow Sea, and segments of the East China Sea. Capable of capturing eight images daily, GOCI’s coverage extends over a width of 2500 km × 2500 km, stationed at an orbital height of 35,837 km, and offers a spatial resolution of 500 m. Encompassing a spectral range of 0.412–0.865 μm, GOCI incorporates six visible bands and two near-infrared bands (

Table 1. Satellite band parameter information of GOCI.

Band	Wavelength (nm)	Band Width (nm)	Type of Band	Applications
B1	412	20	Visible	CDOM, turbidity
B2	443	20	Visible	Maximum absorption of Chlorophyll-a
B3	488	20	Visible	Chlorophyll and other pigments
B4	555	20	Visible	Turbidity and suspended sediment
B5	660	20	Visible	Fluorescence signal and suspended sediment
B6	680	10	Visible	Atmospheric correction and fluorescence signal
B7	745	20	Near-infrared	Atmospheric correction and fluorescence signal
B8	865	40	Near-infrared	Aerosol thickness

). Its primary data find extensive application in studies pertaining to red tide phenomena [36,37,38] and air pollution monitoring [39].

2. Materials and Methods

2.1. Data Source

In this study, GOCI satellite remote sensing data in the vicinity of the Yellow River Estuary (YRE) spanning from 2012 to 2020 were selected and subjected to preprocessing through GOCI data processing system software (Version 2.4). ROI tool within the ENVI Classic software (Version 5.6) was employed to demarcate the study area.

2.2. Study Area

YRE was selected as our study area due to its unique location influenced by the Yellow River, the landmass of Shandong Province, and the Bohai Sea, making it an ideal site for marine ecotone research. The red line in Figure 1, starting from the gate (black point in Figure 1), where the Yellow River meets the sea, indicates the location of sample points. The sample belt (red line in Figure 1), spanning a total length of 111 km, includes 222 sampling points. Each sample point is situated at a latitude of 37°47′54″ N, with a longitudinal interval of 0.005° between adjacent sample points. Data extraction from each sample point along the belt is conducted using MATLAB (Version 9.10). Subsequently, the bord-ER software (Version 3.01) is employed for data analysis, enabling the determination of the ecotone’s location, width, and classification based on the obtained analysis results.

2.3. MSW Method

The MSW method determines the location and width of ecotones by calculating dissimilarity coefficients along a sample belt through a moving window approach. Figure 2 illustrates the MSW method using 12 sample points with a window width of 6 (window width is the number of sample points in one window), while the actual study involves 222 sample points in the red sample belt shown in Figure 1.

First, the window (represented by the black box in Figure 2a) is equally divided into two half-windows, and the dissimilarity coefficient between the two half-windows is calculated. Then, the window is moved to the right (Figure 2b), and the dissimilarity coefficient between the two half-windows is recalculated. This process is repeated until the right endpoint of the window reaches the final sample point; thereby, the dissimilarity coefficient across the entire sample belt is obtained.

Next, the dissimilarity coefficients in the sample belt are plotted. By analyzing the steepness of the curve, the type, location, and width of the ecotone can be determined. Specifically, the interval of a steep peak indicates the location of rapid ecotones, while the interval of a gentle peak corresponds to transitional ecotones. Excessively low peaks are considered non-ecotones. Additionally, the width of a peak represents the width of the ecotone.

Since the window moves one sample point at a time and the window width is always greater than two, overlapping windows are generated during the movement, with each overlapping window corresponding to a distance coefficient. Variations in window width result in changes to the overlapping windows. The number of overlapping windows is governed by Equation (1):

$$S=m-q+1$$

(1)

where $S$ is the number of overlap windows, $m$ is the total number of points, $q$ is the window width, and $q$ is a non-zero even number (Equation (1)).

The position of the dissimilarity coefficient of each overlapping window is at the center of the overlapping window. For instance, if r₁ is the dissimilarity coefficient within the first overlapping window and r₂ is a dissimilarity coefficient within the second overlapping window, with a window width of 2, then the abscissa of r₁ is 1.5, and the abscissa of r₂ is the abscissa of r₁ + 1, which is 2.5. The location of the dissimilarity coefficient on the abscissa for different window widths is summarized in

Table 2. Position of the dissimilarity coefficient values on the abscissa under different window widths.

Window Width	Abscissa of r1	Abscissa of r2	Abscissa of r3	Abscissa of r4	Abscissa of r5	Abscissa of rn
2	1.5	2.5	3.5	4.5	5.5	2/2 + (n − 0.5)
4	2.5	3.5	4.5	5.5	6.5	4/2 + (n − 0.5)
6	3.5	4.5	5.5	6.5	7.5	6/2 + (n − 0.5)
8	4.5	5.5	6.5	7.5	8.5	8/2 + (n − 0.5)
q	q/2 + 0.5	q/2 + 1.5	q/2 + 2.5	q/2 + 3.5	q/2 + 4.5	q/2 + (n − 0.5)

Notes: “Window width” represents the number of sample points included in one window. The “abscissa of r” represents the location of the dissimilarity coefficient in our sample belt (red line in Figure 1); for example, 1.5 indicates the midpoint between the first and second sample points. n: total number of sample points. r: value of dissimilarity coefficient.

.

2.4. Selection of Dissimilarity Coefficient

The dissimilarity coefficient quantifies the difference between the two half-windows within an overlapping window. A higher coefficient indicates a greater disparity between the two half windows, while a lower coefficient suggests a higher degree of similarity. Commonly utilized dissimilarity coefficients encompass Squared Euclidean Distance (SED), Relative Euclidean Distance (RED), and Bray–Curtis Distance (BCD). SED is better able to reflect the steepness of the environmental gradient, providing more intuitive dissimilarity coefficient plots of sample belts [40]. Our analysis reveals distinct wave crest patterns and a wide value range using SED as the dissimilarity coefficient, which means we effectively mitigate the impact of low-value noise on the ecotone determination. Furthermore, we enhance the SED algorithm, presenting the specific formula in Equation (2):

$$SED=\sum _{i=1}^a{\left(X_{iA}-X_{iB}\right)}^2$$

(2)

$SED$ is the value of the squared Euclidean distance in a window; $a$ is the total number of sample points in a half window, $X_{iA}$ is the value of variation in half window A when the sequence of the sample points is $i$, $X_{iB}$ is the value of variation in half window B when the sequence of sample points is $i$. Equation (2) and Figure 2 together provide a clearer understanding (Equation (2)).

2.5. Selection of Variable

This study utilized Chlorophyll-a, CDOM, and TSS as analytical variables in MSW. Chlorophyll-a, a vital component of phytoplankton photosynthesis, exhibits significant temporal and spatial variations in the Yellow River estuary and adjacent water [41]. By incorporating Chlorophyll-a as a variable, we can effectively identify areas of pronounced variation in ecosystem primary productivity within the study area.

CDOM, an integral aspect of the Dissolved Organic Matter (DOM) and the primary constituent of the Dissolved Organic Carbon (DOC) in aquatic environments, serves a dual role of safeguarding water ecosystems and providing essential nutrients for phytoplankton. Therefore, its presence influences the primary productivity of aquatic ecosystems. Including CDOM as a variable allows for a better indication of ecosystem changes.

TSS encompasses both organic and inorganic suspended matter [42] and stands as a pivotal parameter in water quality assessment and environmental monitoring [43]. In the Yellow River estuary, the substantial sediment transported by the river’s discharge exerts a significant influence on estuarine bottom topography. Runoff channels numerous elements from the surface to the bottom layers and from the river to the ocean, with TSS acting as a crucial carrier. Making TSS involved as a variable allows for the quantification of sediment transport dynamics in the YRE.

3. Results

3.1. MSW Based on Chlorophyll-a

The selection of window width profoundly impacts the precision of ecotone determination. Specifically, a narrow window width, which includes fewer sample points per window, often results in multiple wave crests, thereby complicating the delineation of ecotones. While widening the window, increasing the number of sample points encompassed within a window reduces the likelihood of noise interference and makes the peak more discernible. Upon continuous widening of the window width to a certain threshold, whereby the trend of peak variation approaches consistency, the minimum window width emerges as the optimal choice for accurate ecotone determination.

In the following text, “win”, followed by a number, is used to denote the window width, which represents the number of sample points included in a window. For example, win2 means the window contains two sample points (window width is 2). Under win2 and win4, Figure 3 depicts the SED value with Chlorophyll-a as the variable. The analysis reveals significant noise present in the curve under win2 and win4. The noise is particularly noticeable within the SED range of 0 to 0.01. To mitigate the noise on ecotone determination, it becomes imperative to augment the window width. Upon expanding the window width to 12, 14, and 16, the lines of different window widths (win12, win14, win16) show similar trends (Figure 3). According to the method for window width selection, the optimal window width is determined to be 14 (the minimum window width when the trend of peak variation approaches consistency).

The optimal window widths for Chlorophyll-a during the period of 2012 to 2020 are synthesized in

Table 3. The optimum window width of each year.

Year	Window Width
2012	14
2013	14
2014	14
2015	12
2016	16
2017	12
2018	12
2019	22
2020	10

Note: window width is the number of sample points in one window.

. Subsequently, leveraging the optimum window width in different years, the outcomes of the MSW analysis, with Chlorophyll-a as the variable, are depicted in Figure 4.

The Moving-Split Window (MSW) method avoids noise induced by merely calculating the dissimilarity coefficient between adjacent samples through a moving process. Based on the MSW method, the position and width of the ecotone are determined objectively [44]. Specifically, the location of a sharp peak in the “SED-sample sequence tape” figure is the location of the rapid ecotone, and the width of the sharp peak interval is the width of the rapid ecotones [17]. The position of a low peak in the “SED-sample sequence tape” figure is the location of the transitional ecotone, and the width of a low peak is the width of the transitional ecotone.

After establishing the optimal window width, the position, width, and type of the ecotone are determined from 2012 to 2020. Utilizing the results in 2020 as an illustration, five distinct peaks are observed in the belt sequence (Figure 5).

The first peak emerges between sequence positions 10 and 23, spanning a distance of 5.0 km to 11.5 km from the gate, with a width of 6.5 km and exhibiting a high SED value. Characterized by a narrow interval, it typifies a rapid ecotone. The second ecotone occurs between sequence positions 42 and 55, featuring a width of 22.5 km and a low SED value, indicative of a transitional ecotone. The third ecotone surfaces between sequence positions 101 and 108, situated at a distance of 50.5 km to 54.0 km from the gate, with a width of 3.5 km and a high SED value, denoting another rapid ecotone due to its narrow interval. The fifth peak manifests between sequence positions 135 and 141, located 67.5 km to 70.5 km from the gate, with a width of 3.0 km and a low SED value, representing a typical transitional ecotone. Lastly, the final peak arises between sequence positions 206 and 213, spanning a distance of 103.0 km to 106.0 km from the gate, with a width of 6.5 km and a high SED value. Its narrow interval again classifies it as a rapid ecotone.

Considering the figure made from 2012 to 2020, it is evident that multiple ecotones are present in the vicinity of the gate and its adjacent coastal area. Within a 30.0 km distance from the gate, ecotones are consistently detected in all analyses. According to the decision rule, the ecotones within this range are predominantly of the rapid ecotone type. Conversely, fewer peaks are observed between 30.0 km and 90.0 km from the gate, indicating a relatively stable ecosystem in this area. However, at a distance of 90.0 km from the gate, several peaks are identified as rapid ecotones. Through MSW analysis, it is revealed that the estuary and its adjacent coast exhibit a dense concentration of ecotones, with the majority being of the rapid variation type. Conversely, the inner bay predominantly features transitional ecotones.

3.2. MSW Based on CDOM

The quarterly CDOM data from 2012 to the third quarter of 2020 were subjected to MSW analysis (Figure 6). As a result, it was observed that while the range of SED values was narrow, distinct peaks were evident in the sample belt sequence each year. This suggests that the range of SED values does not significantly impact the determination of ecotone boundaries.

Within 30.0 km from the gate, two to three rapid ecotones are consistently identified each year, except for 2015, according to the analysis results. For instance, the analysis of 2013 reveals two peaks within the 30.0 km range from the gate. The first peak occurs between sequence positions 9 and 20, spanning a distance of 4.5 km to 10.0 km from the gate, with a width of 5.5 km and a standard deviation of CDOM content of 5.27%. The second peak appears between sequence positions 39 and 59, situated 14.5 km to 29 km from the gate, with a width of 9.5 km and a standard deviation of CDOM content of 7.88%. Between these two peaks lies a region characterized by stable CDOM changes, spanning from 10 km to 19.5 km from the gate with a width of 9.5 km. Overall, the MSW results utilize the CDOM variable to indicate variations in the number of ecotones across different years. The detailed results of the CDOM analysis are in

Table 4. Results of MSW analyses based on CDOM between 2012 and 2020.

Year	Position of Peak (in Sequence)	Distance from the Gate (km)	Width (km)	Type of Ecotone	Window Width
2012	10–21	5.0–10.5	5.5	Rapid ecotone	14
	35–46	17.5–23.0	5.5	Transitional ecotone
	149–161	74.5–80.5	6.0	Transitional ecotone
	169–199	84.5–99.5	15.0	Transitional variation
	206–215	103.0–107.5	4.5	Transitional ecotone
2013	9–20	4.5–10.0	5.5	Rapid ecotone	14
	39–58	19.5–29.0	9.5	Rapid ecotone
	136–146	68.0–73.0	5.0	Transitional ecotone
	158–174	79.0–87.0	8.0	Rapid ecotone
	193–207	96.5–103.5	7.0	Rapid ecotone
2014	7–27	3.5–13.5	10.0	Transitional ecotone	14
	35–61	17.5–30.5	13.0	Rapid ecotone
	91–102	45.5–51.0	5.5	Transitional ecotone
	114–123	57.0–61.5	4.5	Transitional ecotone
	176–188	88.0–94.0	6.0	Transitional ecotone
	198–206	99.0–103.0	4.0	Transitional ecotone
2015	7–18	3.5–9.0	5.5	Transitional ecotone	14
	23–32	11.5–16.0	4.5	Transitional ecotone
	40–62	20.0–31.0	11.0	Transitional ecotone
	124–137	62.0–68.5	6.5	Transitional ecotone
	144–155	72.0–75.5	5.5	Transitional ecotone
	193–215	96.5–107.5	11.0	Transitional ecotone
2016	18–31	9.0–15.5	6.5	Rapid ecotone	10
	102–111	51.0–55.5	4.5	Rapid ecotone
	149–160	74.5–80.0	5.5	Rapid ecotone
	203–211	101.5–105.5	4.0	Transitional ecotone
2017	21–33	10.5–16.5	6.0	Transitional ecotone	18
	36–46	18.0–23.0	5.0	Transitional ecotone
	48–67	24.0–33.5	9.5	Transitional ecotone
	123–140	61.5–70.0	8.5	Rapid ecotone
	140–154	70.0–77.0	7.0	Transitional ecotone
	175–189	87.5–94.5	7.0	Transitional ecotone
2018	31–49	15.5–24.5	9.0	Rapid ecotone	22
	58–74	29.0–37.0	8.0	Rapid ecotone
	164–204	82.0–102.0	20.0	Transitional ecotone
2019	22–39	11.0–19.5	8.5	Rapid ecotone	16
	48–70	24.0–35.0	11.0	Rapid ecotone
	121–131	60.5–65.5	5.0	Rapid ecotone
	154–163	77.0–81.5	4.5	Transitional ecotone
	178–194	89.0–97.0	8.0	Transitional ecotone
	202–213	101.0–106.5	5.5	Transitional ecotone
2020	10–24	5.0–12.0	7.0	Rapid ecotone	20
	26–54	13.0–27.0	14.0	Rapid ecotone
	59–72	29.5–36.0	6.5	Rapid ecotone
	83–97	41.5–48.5	7.0	Rapid ecotone

Note: The position of the peak is related to the location of the sample points. For example, an abscissa value of 1 represents the location of sample point 1.

.

3.3. The Results of MSW Based on TSS

The Total Suspended Solids (TSS) data were subjected to MSW analysis (Figure 7). Except for the years 2013, 2016, and 2017, only one ecotone is discerned in MSW analysis based on TSS, with the singular peak situated at the gate area (Figure 7). The detailed results of the TSS analysis are in

Table 5. Results of MSW analyses based on TSS between 2012 and 2020.

Year	Position of Peak (in Sequence)	Distance from the Gate (km)	Width (km)	Type of Ecotone	Window Width
2012	0–21.0	0–10.5	10.5	Rapid ecotone	10
2013	0–20.0	0–10.0	10.0	Rapid ecotone	12
	156.0–166.0	78.0–83.0	5.0	Rapid ecotone
	210.0–222.0	105.0–111.0	6.0	Transitional ecotone
2014	0–14.0	0–7.0	7.0	Rapid ecotone	14
2015	0–18.0	0–9.0	9.0	Rapid ecotone	6
2016	0–17.0	0–8.5	8.5	Rapid ecotone	10
	49.0–59.0	24.5–29.5	5.0	Transitional ecotone
	200.0–211.0	100.0–105.5	5.5	Transitional ecotone
2017	0–12.0	0–6.0	6.0	Rapid ecotone	8
	39.0–48.0	19.5–24.0	4.5	Transitional ecotone
2018	0–16.0	0–8.0	8.0	Rapid ecotone	10
2019	0–28.0	0–14.0	14.0	Rapid ecotone	14
2020	0–26.0	0–13.0	13.0	Rapid ecotone	10

.

The peak observed at the gate exhibits a high SED value. While the precise starting point of most peaks cannot be definitively determined, it is inferred that the peak originates from the gate. TSS serves as the primary source of sediment, with the YRE conveying a significant amount of sediment into the sea, thereby exerting a substantial influence on the distribution of TSS at the gate. The MSW analysis reveals the formation of rapid ecotones at the gate, indicative of sediment impacts spanning a range of 6.0 km to 14.0 km within the estuary region.

MSW analysis based on TSS data accurately delineates ecotones within the gate area. Taking the data from 2020 as an example, a rapid variation ecotone spanning between sequence positions 0 and 26 is identified, exhibiting a width of 13.0 km. Within this area, the maximum TSS value recorded is 11.67 mg/m³, the minimum is 3.98 mg/m³, and the standard deviation is 239%. Conversely, from sequence position 27 to the end of the sample belt, the maximum TSS value is 4.05 mg/m³, with a standard deviation of 42.95%. In other words, the TSS valuation at the gate is significantly higher than that at the inner bay.

4. Discussion

4.1. Defect of MSW Method and Data

The MSW methodology inherently presents certain limitations and uncertainties. Primarily, the absence of a predefined threshold to identify the primary peak poses a challenge. In the analytical results, the existence of small peaks within narrow intervals indicates the presence of considerable noise in ecotone determination. In an effort to mitigate this noise, the threshold for the distance coefficient is subjectively established. Only peaks exceeding this threshold are deemed to represent ecotones. While noise can be reduced by expanding the number of sample sequences contained within a window, the consequence of this approach may lead to inaccuracies in determining the interval width and could result in the merging of multiple peaks into a single peak, thereby compromising the accuracy of ecotone determination.

The accuracy of data in certain sample points presents an additional challenge. In this study, data were sourced from the Korean Geostationary Ocean Color Imager (GOCI) satellite, which provides level-2 data with a spatial resolution of 500 m. During the extraction of sample points using MATLAB (Version 9.10), a subset of points exhibited identical values, indicating potential errors. This lack of accuracy is likely attributable to cloud cover [45]. This type of issue could be addressed in the future by attempting to obtain marine sampling data and combining it with remote sensing data.

The research variables employed in this study are limited. Specifically, Chlorophyll-a, Chromophoric, Dissolved Organic Matter (CDOM), and Total Suspended Solids (TSS) serve as the primary variables. The spectral bands of Chlorophyll-a and CDOM are closely aligned, contributing to the consistency of the analysis results. Also, the utilization of only three variables poses challenges in discerning the influence of different variables on ecotone determination. Given the complexity of marine ecosystems, which are influenced by a multitude of biological and physical factors, the inclusion of only three physical variables may not adequately capture the ecosystem’s dynamics. Hence, future research endeavors should seek to identify additional variables that offer a more comprehensive representation of the ecosystem.

In terrestrial studies, MSW analysis accounts for the comprehensive effects of various variables and assigns weight values to variables based on the magnitude of their impact. A similar approach could be adopted in marine research. This method requires future researchers to further enhance the understanding of variables in marine ecosystems and obtain more precise variable data.

4.2. Strength of GOCI Data and Position of Sample Belt

MSW methodology, rooted in the sample belt research approach, provides an effective and objective means to determine the location and width of ecotones in terrestrial environments [40]. In terrestrial MSW analysis, variables pertaining to terrestrial plants, such as height, biomass, and coverage, are predominantly utilized as research variables [40]. The stability exhibited by terrestrial plants facilitates the accurate delineation of ecotones. Conversely, in marine research, ecotone boundary investigations are centered around organisms such as ascidians [46], with data derived from bibliographical sources. However, the recorded data span various regions and may pose challenges for utilization in small-scale research endeavors like YRE. It is inherent that the mobility of marine organisms complicates fixed-point data acquisition. In contrast, satellite data provides a long-term dataset suited for MSW analysis in marine environments, where complex interactions between water masses and ocean currents prevail. Additionally, high-resolution data from the Korean Geostationary Ocean Color Imager (GOCI) offers sufficient accuracy for small-scale research within YRE.

The placement and orientation of the sample belt are advantageous for our research. The chosen sample belt, oriented parallel to the latitude line, is primarily aimed at assessing the impact of Yellow River runoff on the local ecosystem. As the Yellow River discharges into the ocean at gate area, its influence gradually attenuates across the sampled belt. Aligning the sample belt parallel to the latitude line provides a more practical framework for analyzing ecosystem dynamics. Additionally, the sample belt is strategically positioned at the entrance of Laizhou Bay, thereby subjecting it to influences from the surrounding regions, including land in Shandong province, Laizhou Bay, and the Bohai Sea. The environmental conditions in this area are characterized by complexity, which inherently aids in the determination of ecotones. Previous studies on the boundaries of the Yellow River Estuary have primarily focused on salinity, finding that these boundaries, which also lie within the estuarine gate area, are consistent with the findings presented in this paper [47].

4.3. Practical Applications of Result

In all three analyses, the variables exhibit sharp changes in the gate area, indicating the presence of a rapid ecotone. The ecotone situated around the water area demonstrates significant buffering capacity [48]. CDOM variation indicates the presence of phytoplankton. Phytoplankton, which are widely present in the Bohai Sea [49], can effectively degrade organic matter and purify pollutants, thereby reducing the influx of pollutants from the Yellow River into the Bohai Sea. Additionally, the primary productivity in the ecotone area is higher, thereby leading to an abundance of prey organisms and the abundance of prey organisms, making the estuary a crucial spawning and feeding ground for various marine species, such as the oyster (Crassostrea rivularis) [50] and the haarder (Liza haematocheila) [51]. Therefore, it is imperative to enhance the monitoring and assessment of the ecotone area to protect its ecosystem and biodiversity.

5. Conclusions

MSW analysis is a valuable tool for determining ecotones in marine environments. With the advancement of remote sensing technology, MSW’s application has expanded beyond vegetation habitat demarcation. This study employs MSW to identify ecotones in marine environments, utilizing level 2 data from GOCI satellites. The results indicate distinct peaks, allowing for the determination of the location, width, and type of ecotones. Multiple ecotones are identified in the Yellow River gate and coastal areas. Within 30 km of the gate, there are rapid ecotones from 2012 to 2020. Conversely, in the inner bay area, located 30 km to 90 km from the gate, transitional ecotones are detected, indicating a more stable ecosystem. Beyond 90 km from the gate, several peaks are identified as rapid ecotones. In conclusion, rapid ecotones predominantly occur in the gate and coastal areas, while transitional ecotones are more common in the inner bay.

The influence of Total Suspended Solids (TSS) on the Yellow River gate extends between 10 km and 14 km. Within this range, a pronounced peak is observed, indicating a rapid ecotone. The emergence of these ecotones is due to the substantial sediment load discharged from the Yellow River gate into the sea. Annual analysis consistently identifies ecotones within this area, signifying that the Yellow River runoff significantly impacts the region within 14 km of the gate.

MSW conducts quantitative analysis of marine ecotones from an ecological perspective, clearly delineating interactions between ecosystems and their changing trends. It provides a long-term series analysis of the dynamic changes within ecotones, aiding in the understanding of marine ecosystem dynamics and facilitating informed planning to ensure smooth agricultural operations. This study demonstrates the efficacy, feasibility, and scientific validity of MSW for ecotone determination in marine environments, offering a novel research perspective for identifying ecosystem boundaries in marine systems.