An Improved Method for Estimating Blue Carbon Storage in Coastal Salt Marsh Wetlands: Considering the Heterogeneity of Soil Thickness

Abstract

Coastal wetlands are vital ecosystems at the land–sea interface. They intercept land-based pollutants, regulate microclimates, and mediate carbon cycles. They play a significant role in enhancing carbon sequestration capacity and maintaining ecological structure and functioning. This study proposes an improved method for estimating blue carbon storage in coastal salt marsh wetlands, considering soil thickness, by utilizing an enhanced Soil Land Inference Model (SoLIM) to estimate soil thickness in coastal wetlands with a restricted number of sample points. The wetland soil thickness index is integrated into the Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) blue carbon storage estimation model, ultimately enabling the estimation and visualization of blue carbon storage in the Liaohe Estuary coastal wetland. Results indicate the following: (1) The studied area’s soil thickness shows a spatial distribution pattern that becomes progressively thinner from north to south. Soil thickness is more significant in the salt marsh vegetation areas and more minor in the coastal tidal flat areas, with 52% of the region having soil thickness between 40 and 60 cm. (2) In 2023, the blue carbon stock in the study area is estimated at 389.85 × 10⁶ t, with high-value areas concentrated in the northern natural landscapes, and low-value areas in the southern coastal zone, characterized by flat terrain and human influence. The coupled soil thickness–blue carbon storage estimation model provides methodological support for refining the estimation of blue carbon storage in coastal wetlands. It also offers technical support for formulating policies on the ecological restoration, compensation, protection, and management of coastal wetlands.

1. Introduction

Since the Industrial Revolution, greenhouse gas emissions from human activities have led to a 0.6 °C increase in global average temperature, causing ecological impacts such as species migration to higher latitudes and disruptions in phenological cycles. Meanwhile, the decline in biodiversity has reduced resource use efficiency, posing threats to agricultural productivity and increasing the risk of disease transmission. The dual feedback between climate and biological systems is further exacerbating the environmental crisis [1,2,3]. Extreme weather events have occurred in many regions worldwide, severely damaging terrestrial, freshwater, and marine ecosystems, bringing widespread adverse effects to both human society and natural ecosystems [4]. “Carbon peak” refers to the process in which the carbon dioxide emissions of a specific entity reach a historical peak, enter a plateau phase, and then gradually decline, signifying the decoupling of carbon emissions from economic growth. “Carbon neutrality” is achieved by offsetting carbon dioxide emissions generated within a specific period through human-induced carbon sequestration, ultimately resulting in net-zero emissions [5]. As a unique zone of land–sea interaction, the coastal zone serves as a crucial functional carrier for advancing the coordinated goals of “carbon peak” and “carbon neutrality”. Additionally, the coastal zone provides unique habitat conditions for coastal wetlands, making it a critical area for functioning wetland ecosystems [6]. The Ramsar Convention defines coastal wetlands as coastal lowlands, intertidal flats, and shallow waters regularly inundated by static or flowing water due to land–sea interactions. These wetlands are recognized as some of the most valuable ecosystems on the planet [7,8]. The ecological health of coastal wetlands is not only influenced by natural factors such as tidal dynamics and salinity gradients but also directly subjected to multiple pressures from human activities. Direct pressures such as agricultural expansion, urbanization, and industrial pollution alter hydrological connectivity, increase pollutant inputs, and fragment habitats, leading to the degradation of wetland ecosystem services and threatening critical ecological benefits, including climate regulation and disaster mitigation [9]. Coastal wetlands form a complex and dynamic ecosystem under the interaction of natural and human factors, providing a variety of ecosystem services. Specifically, these services include provisioning services (such as fishery resources and freshwater supply), regulating services (such as climate regulation, water quality purification, and blue carbon storage), and cultural services (such as ecotourism and spiritual value) [10,11,12]. Blue carbon denotes the organic carbon sequestered and retained within vegetated coastal ecosystems, including salt marshes, mangroves, and seagrass meadows [13]. Mangrove, salt marsh, and seagrass bed wetlands bury more than half of the annual organic carbon stored in marine sediments [14]. Salt marsh wetlands exhibit exceptional carbon sequestration efficiency in tidal-driven, water-saturated, and low-oxygen sedimentary environments. Their carbon accumulation rate can reach 64.70 tons ${\mathrm{C}\mathrm{O}}_2$e per hectare per year, significantly surpassing that of forest and marine ecosystems [15]. Compared to other ecosystems, their soil carbon burial rate is approximately 40 times higher [16,17]. Investigating the dynamic changes, formation mechanisms, and influencing factors of “blue carbon” in coastal wetlands is crucial for mitigating climate change and promoting the balance of the carbon cycle.

Since the 1970s, scholars have begun to conduct in-depth studies on the carbon storage of wetland ecosystems, mainly focusing on wetland vegetation biomass, carbon emissions and decomposition, and the transformation of sedimentary substances [18]. In the 21st century, research has gradually focused on topics such as the mechanisms of wetland carbon sources and sinks, the migration and transformation processes of carbon, and the spatiotemporal distribution patterns of carbon storage, aiming to explore the changing patterns and causes of carbon storage in wetland ecosystems. Key studies have focused on the characteristics of wetland carbon sources and sinks, the key factors influencing soil carbon sequestration rates, and the exploration of carbon storage changes by integrating various parameter data from wetland vegetation, soil, and biota [19,20]. Macreadie et al. estimated Australian tidal wetlands’ blue carbon storage and accumulation rate and explored the key factors influencing their carbon storage [21].The Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) model is employed to evaluate and measure ecosystem services. By utilizing spatial modeling, it examines how land-use changes influence these services, providing decision-makers with insights into their spatial and temporal dynamics [22]. Richa Sharma et al. combined InVEST and GEE-RF to estimate carbon storage in the Noida region for 2011 and 2019 and simulated land cover changes for 2027 [23]. Pengcheng Li et al. combined the PLUS and InVEST models to assess carbon storage in Liaoning Province from 2000 to 2020, examine its spatial distribution patterns, and project future land-use and land-cover changes (LUCC) for 2050 [24]. The blue carbon module within InVEST is a sophisticated tool utilizing spatially explicit models to offer scientific insights into blue carbon storage and its potential for emission reduction. It integrates wetland vegetation distribution and density data, vegetation biomass, habitat-specific carbon pool data, and local soil carbon storage and carbon fixation data to evaluate the storage of blue carbon in different land-use categories [25,26]. The commonly used InVEST carbon storage estimation model fails to adequately account for the differences in soil thickness and carbon density among different land use categories [27,28], resulting in a less accurate estimation of wetland soil carbon storage. Consequently, it is essential to improve the assessment of soil carbon density by accounting for differences in soil types and thicknesses to enhance the precision of wetland blue carbon storage estimates.

The Third Law of Geography can be stated as follows: At a specific spatial–temporal scale, the spatial heterogeneity of geographic phenomena is made predictable by integrating multi-source data and analyzing geographic configuration similarity. That is, regions with similar geographic and environmental combinations (spatial associations and hierarchical structures of natural elements) exhibit significant similarities in their target geographic variable characteristics or processes [29]. In 2005, Axing Zhu and Amanda C. Moore, among others, proposed the Soil and Land Inference Model (SoLIM) based on the third law of geography. The model seeks to address the shortcomings of conventional spatial prediction techniques and has found extensive use in a variety of geographic spatial forecasting applications [30]. In contrast to conventional digital soil mapping techniques, the SoLIM approach does not depend on a large number of sampling points as prior data. This significantly enhances the efficiency of soil surveys while also reducing both time and financial costs. At the same time, SoLIM outperforms traditional methods in terms of mapping accuracy and soil attribute extraction. It can rapidly generate regional soil databases, providing valuable references for soil surveys [31]. Zhu et al. input 10 m resolution SoLIM data into the Soil and Water Assessment Tool (SWAT) for the Brewery Creek watershed in the United States. They confirmed that runoff simulations in the watershed were potentially influenced by the contributing area, providing preliminary guidance for selecting soil data in hydrological simulations for small to medium-sized watersheds [32]. Qin et al. proposed a new method, SoLIM-IDW, to address the issue of disregarding spatial distance information when predicting soil attributes. In situations with limited sample points, this method alleviates the correlation among environmental variables through principal component analysis, enhancing the connections between interest points and sample points and thereby improving the accuracy of soil attribute predictions [33]. Studies have demonstrated that the SoLIM model can overcome the limitations of traditional digital soil mapping and accurately reflect the spatial distribution of soil attributes. However, the applicability of the SoLIM model in coastal wetlands still requires further validation and research.

The Liaohe Estuary wetland is the most extensive coastal Phragmites australis (P. australis) wetland in the high-latitude areas of China, with underwater delta shallow coastal wetlands, tidal flats, Suaeda salsa (S. salsa) lands, and P. australis marshes distributed from the sea to the land. These features directly show the coastal wetland evolution process [34]. Wetland ecosystems, with their unique surface cover patterns and hydrological connectivity, form a highly heterogeneous environmental foundation where the synergistic effects of multiple factors influence soil formation and development processes. Wetland plants regulate soil organic matter dynamics through litter input and root exudates, and their spatial distribution characteristics further control the material transfer process within the soil profile. Periodic hydrological fluctuations alter soil moisture transfer pathways, driving the migration of minerals and the reorganization of organic matter, ultimately shaping the spatial differentiation pattern of soil structure. The coupling mechanisms of these multi-scale environmental factors constitute an important driving force in forming soil spatial heterogeneity in wetlands. This study constructed an environmental factor dataset based on the 2023 Advanced Land Observing Satellite (ALOS) elevation data and Sentinel-2 remote sensing imagery, coupling the SoLIM-InVEST model to refine the estimation of blue carbon stock (BCS) in the Liaohe Estuary coastal wetland for 2023. The coupling of the SoLIM-InVEST model innovatively integrates the environmental response mechanism of SoLIM in soil property spatial differentiation analysis with the quantitative representation advantages of InVEST in blue carbon dynamic simulation [35]. This approach reveals the influence mechanism of environmental factor interactions on soil thickness. Based on this, a soil thickness–carbon density coupling equation is developed, and high-precision soil parameters are spatially assimilated with the InVEST model. This effectively addresses the issue of an insufficient representation of carbon density spatial heterogeneity in traditional BCS estimation, which arises from neglecting differences in soil thickness. The following outlines the specific objectives: (1) to predict wetland soil thickness with a small number of sample points, exploring the fundamental causes of soil thickness differences, and (2) to combine the spatiotemporal distribution of soil thickness with the spatial heterogeneity of carbon density, using high-resolution remote sensing data and measured soil samples to establish an accurate soil carbon density model, thereby enabling a refined estimation of wetland blue carbon storage. The findings of this study offer methodological guidance for estimating coastal wetland soil thickness and offer theoretical references for blue carbon storage accounting and carbon-negative emission planning in coastal wetlands.

2. Study Area and Data Sources

2.1. Formatting of Mathematical Components

The Liaohe Estuary National Nature Reserve (121°28′ E–121°58′ E, 40°45′ N–41°05′ N) is located in the northern Liaohe Plain of Panjin City, Liaoning Province, and the central region of the Liaohe Delta. The Liaohe Estuary study area is situated in the middle latitudes, characterized by a semi-humid monsoon climate typical of the northern temperate zone, with a total area of 128,685 hectares. The region experiences an average annual temperature of 8.4 °C and receives an annual precipitation of 623.2 mm on average, mainly concentrated in summer. The total annual sunshine duration is 2768.5 h, exceeding the provincial average for Liaoning. Influenced by the Bohai Sea, wind speed and direction remain relatively stable, with southwest winds prevailing year-round and an average wind speed of 4.3 m/s. The Liaohe Estuary wetland comprises six types: P. australis marshes, tidal flats, shallow sea areas, rivers, reservoirs, and rice paddies. The soils consist of tidal salt soils, coastal salt soils, meadow salt soils, and marsh salt soils. Approximately 217 plant species belong to 40 families [36].

As a typical representative of coastal wetlands in China, the Liaohe Estuary wetland has undergone significant changes in landscape patterns over the past 40 years. Remote sensing monitoring data show that the area of natural wetlands in the study area has decreased by a total of 270.12 km², with 15.64% of the natural wetlands converted to artificial wetland types and 6.20% of the wetland resources completely losing their ecological function due to the expansion of construction land and encroachment by oil field facilities, resulting in a significant decline in wetland ecosystem service capacity [37]. Therefore, implementing ecological compensation projects and promoting policies such as converting cropland back to wetlands and retiring aquaculture to restore tidal flats is urgent for wetland restoration [38]. The Liaohe Estuary wetland exhibits a typical plant community distribution controlled by topography. Areas with slightly higher elevations and shorter inundation periods are dominated by meadow P. australis, while areas with seasonal or prolonged inundation feature P. australis communities interspersed with plants such as Scots pine (Pinus sylvestris) and S. salsa [39]. Additionally, the salinity regulation by both riverine and seawater influences the dynamic changes in the vegetation distribution pattern of the Liaohe Estuary wetland. Increased freshwater input from rivers pushes the P. australis growth range outward while rising seawater salinity promotes the expansion of salt-tolerant plants like S. salsa [40]. These wetlands are primarily dominated by P. australis and salt-tolerant S. salsa, which play a vital role in carbon sequestration and storage. The Liaohekou was selected as the research location for predicting soil thickness and assessing BCS, providing a solid scientific foundation for wetland degradation monitoring and ecological restoration. Figure 1 provides an overview of the study area:

2.2. Data Sources

The elevation data used in this study are from the 2023 ALOS DEM (12.5 m). SAGA GIS 9.0.3 software was utilized to process terrain factors, including the topographic wetness index, plan curvature, slope, and profile curvature, for analyzing the topographical characteristics, geomorphological changes, surface coverage, and hydrological connectivity conditions of the study area. This study utilizes Sentinel-2 remote sensing imagery with a 10 m resolution. The Sentinel-2 satellites have advanced MSI instruments, providing high-resolution, hyperspectral observation data across 13 spectral bands. The Sentinel-2 data used in this study are the Level-2A surface reflectance products, which have undergone radiometric calibration and geometric correction. The subsequent data processing steps include using SNAP (Sentinel Application Platform) for preprocessing the downloaded L2A products, primarily involving atmospheric correction and resampling; ENVI 5.3 for mosaicking two images within the study area, selecting the seamless mosaic tool in ENVI, setting the data ignore value to 0, and performing color balancing before outputting the result as a TIFF file; and using ArcMap to mask and extract the mosaicked data according to the experimental area boundaries to obtain satellite remote sensing data for the study area.

2.3. Soil Thickness Sampling Data and Processing

The soil samples analyzed in this study were obtained from two field campaigns conducted from 20 to 25 May 2023 and from 23 to 26 October 2023, yielding 40 sample points. To ensure that the sample points were more representative and typical, this study selected characteristic areas of the Liaohekou wetland, such as reservoirs, river mouths, extensive tidal flats, and areas far from water sources. The geographic coordinates (latitude and longitude) and elevation of each sample point were accurately measured using an RTK instrument, with the coordinate reference system set to WGS-1984. Detailed information was recorded for each sample point, including soil thickness, topographic features, vegetation distribution, land use, and slope position. From top to bottom, soil sampling is divided into three distinct layers: the upper humus layer, the middle soil layer, and the lower sandy soil layer. The stratification is based on field morphological observations (such as texture, color, and structural characteristics). The upper humus layer is typically darker (grayish-black), has a loose texture and granular structure, and contains plant roots. The middle soil layer is usually brown, with a compact texture and contains clay particles. The lower muddy, sandy soil layer is the lightest in color. To measure the soil thickness at the sampling points, soil cores with a maximum depth of 1 m and a diameter of 50 mm were used to drill vertically to the contact surface of the bedrock layer. Soil samples were taken to a depth of 1 m at each sampling point. After the research team’s on-site assessment, no signs of soil layers were found beyond 80 cm; thus, the maximum soil thickness in the study area was determined to be 80 cm. Soil cores collected by the auger were divided according to soil horizons; samples were collected at depths of 0–20 cm, 20–40 cm, 40–60 cm, and 60–80 cm, and these samples were then placed into sample bags. To minimize measurement errors caused by small-scale variations in soil thickness, a secondary validation point (P1) was set 1 m apart along a random horizontal direction after the initial measurement at the sampling point (P0). If the soil thickness measurements at P0 and P1 fell within the same layer, the value of that layer was recorded as the soil thickness. If the two measurements fell into different layers, a third validation point (P2) was added 1 m vertically from P1. When any two points were within the same layer, the repeatedly occurring layer was selected as the final soil thickness value for the sampling point to ensure the accuracy of the thickness layer information.

3. Research Methodology

3.1. Methodological Pathways

This study used ALOS and Sentinel-2 data to extract topographic factors, hydrological factors, and remote sensing factor to construct an environmental factor dataset for wetland soil formation. The dataset was subsequently entered into the SoLIM model for analyzing and predicting soil thickness. The accuracy of soil thickness estimation in the SoLIM model is greatly influenced by the choice of environmental factors and the accuracy of the required data sources. The measured soil thickness data from the sampling points were used as the predicted soil thickness values for accuracy validation. Additionally, the InVEST model was used to overlay the 2023 land use type map with the soil thickness prediction map, and incorporating the observed carbon density values, estimate the BCS of the Liaohekou wetland in 2023. This study’s technical roadmap is presented in Figure 2.

3.2. Extraction of Environmental Factors in Coastal Wetlands

3.2.1. Terrain Factors Extraction

Elevation, slope, plan curvature, profile curvature, and the topographic wetness index are key terrain factors affecting soil thickness. Elevation influences climate conditions and vegetation distribution, determining soil type and thickness. Slope affects surface runoff speed and erosion intensity; soils are easily eroded and thin on steep slopes, while on gentle slopes or flatlands, soil accumulation is thicker. Curvature affects the convergence and dispersion of surface runoff and variations in flow velocity, thereby influencing soil erosion and deposition processes. The topographic wetness index reflects the impact of terrain on water retention capacity, with regions of higher humidity generally exhibiting thicker soil layers. The spatial differences in soil thickness can be more accurately reflected by selecting these five terrain factors as the primary factors for simulating soil thickness [41]. The terrain factors in this study were derived using the Terrain Analysis function and the Basic Terrain Analysis module under the Compound Analyses feature in SAGA GIS 9.0.3 software, as well as the TWI-TOPMODEL module.

(1)

Topographic wetness index (TWI)

The TWI is defined as the ratio of the convergent area per unit contour length to the local slope at a specific location; this index quantitatively characterizes the balance between surface runoff convergence and gravitational drainage, comprehensively reflecting how topographic factors control hydrological processes. It plays a crucial role in indicating the drainage, water retention, and soil erosion functions of the soil [42,43]. The mathematical formula for TWI is as follows:

$$w=s/\tan \alpha$$

(1)

where s represents the catchment area along the contour line, while $\tan \alpha$ represents the slope. The darker the color, the higher the value, indicating greater surface moisture content. Areas with high terrain moisture values are primarily distributed around estuaries, reservoirs, ponds, and similar locations.

(2)

Plan curvature (Plan C)

Plan curvature, also known as contour curvature, is the curve formed by the intersection of the horizontal plane at a point on the surface and the surface itself, representing the curvature at that point [44]. Plan curvature and profile curvature can reflect changes in surface morphological structures, including cross-sectional shapes in different directions and surface undulations, affecting the distribution of wetland organic carbon, microorganisms, and other components in the soil. The expression is as follows:

$$C_c={\displaystyle \frac{f_y^2{f}_{xx}-2f_x{f}_y{f}_{xy}+f_x^2{f}_{yy}}{(f_x^2+f_y^2)(1+f_x^2+f_y^2{)}^{3/2}}}$$

(2)

where $C_c$ represents plan curvature, $C_c$ > 0 indicates a convex contour shape, and $C_c$ < 0 indicates a concave contour shape. $f_x$ and $f_y$ represent the first derivatives of the elevation surface along the x and y axes, while $f_{xx}$ and $f_{yy}$ represent the second derivatives along the x and y axes.

As shown in Figure 3, the plan curvature of the Liaohe Estuary study area ranges from −0.1 to 0.008 ${\mathrm{m}}^{-1}$, with a relatively uniform spatial distribution and slight variation in curvature values. This indicates that a relatively flat surface is suitable for agricultural production, infrastructure construction, etc.

(3)

Profile curvature (Profile C)

Profile curvature represents the changes in the surface slope. It is the curve formed by the intersection of the standard section at a point on the surface and the terrain surface, representing the curvature at that point. Profile curvature influences the rate of migration, erosion, and the deposition of surface materials [45].

$$C_p={\displaystyle \frac{f_x^2{f}_{xx}-2f_x{f}_y{f}_{xy}+f_y^2{f}_{yy}}{(f_x^2+f_y^2)(1+f_x^2+f_y^2{)}^{3/2}}}$$

(3)

where $C_p$ represents profile curvature, $C_p$ > 0 indicates a convex profile shape, and $C_p$ < 0 indicates a concave profile shape. $f_x$ and $f_y$ represent the first derivatives of the elevation surface in the x and y directions, respectively, while $f_{xx}$ and $f_{yy}$ represent the second derivatives in the x and y directions.

The profile curvature obtained in this study is shown in Figure 3, with the profile curvature of the Liaohe Estuary study area ranging from 0 to 6.281 ${\mathrm{m}}^{-1}$. The profile curvature is relatively stable, with small curvature values, indicating gentle slope changes, suitable for certain production activities that require specific terrain conditions.

3.2.2. Remote Sensing Factor Extraction

The remote sensing factor selected in this study is the Normalized Difference Vegetation Index (NDVI), an important indicator used to quantify vegetation coverage based on remote sensing image processing. The NDVI obtained in this section is shown in Figure 4, derived from Sentinel-2 remote sensing imagery processed using eCognition 8.9.0 software. The NDVI range in the Liaohe Estuary study area is between −0.34 and 0.62, with most areas having positive NDVI values, indicating good vegetation growth and high vegetation coverage, consistent with the distribution characteristics dominated by P. australis and S. salsa.

3.2.3. Hydrologic Factors Extraction

The IIC and PCI were calculated using the Conefor Sensinode 2.6 software and the Conefor plugin in ArcGIS 10.8. The IIC reveals the intrinsic patterns and regional characteristics of wetland hydrological processes, aiding in analyzing water resources’ spatial and temporal distribution and regulation mechanisms. The PCI assesses the potential for water resource development, supporting water resource management and planning [46,47]. The formula for IIC and PCI is as follows:

$$IIC={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n\left[(a_i{a}_j)/(1+n{l}_{ij})\right]}}}{A_L^2}}$$

(4)

where n denotes the total number of surface water body patches within the wetland landscape; $a_i$ denotes the size of patch i; $a_j$ denotes the size of patch j; and ${nl}_{ij}$ represents the number of direct shortest path links between patch i and patch j. $A_L^2$ represents the total surface area of the wetland aquatic landscape patches.

$$PCI={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{a}_i{a}_j{P}_{il}^{*}}}}{A_L^2}}$$

(5)

where n represents the total count of water body patches in the landscape; P_il* represents the product of the highest probability of a potential link between patches i and j; $a_i$ represents the size of the patch i;${ a}_j$ represents the size of the patch j; and $A_L^2$ represents the sum of the areas squared for the water body landscape patches.

Lower values of IIC and PCI indicate weaker connectivity between patches, while higher values represent stronger connectivity. As shown in Figure 4, the darker the color, the better the hydrological connectivity in the study area.

3.3. SoLIM Digital Soil Mapping Principles

The soil–landscape model theory posits that specific soil types are formed under the combined influence of certain environmental factors. By utilizing geographically relevant environmental data and soil property data and based on the autocorrelation characteristics of the spatial distribution of soil properties, geostatistical methods can be employed to infer the spatial arrangement of soils. Based on this theory, SoLIM integrates geographic information system (GIS) technology, artificial intelligence under fuzzy logic, mathematical statistics methods, and expert knowledge. It uses GIS to represent the environmental conditions of soil formation. It combines geographic and environmental data and soil property data related to the mechanisms of soil spatial distribution to establish a similarity-based inference engine. This model links acquired knowledge with typical soil formation environmental conditions and, based on the autocorrelation characteristics of soil property spatial distribution, employs fuzzy inference and other geostatistical methods to infer the spatial distribution patterns of soil. This enables the digital representation and mapping of soil properties [31].

This study’s wetland soil thickness prediction was implemented using the SoLIM software. It assigns weights to pixels of different soil types, representing regional soils as a pixel array containing relevant soil property information. A soil similarity vector characterizes each soil pixel. The similarity of soil in the region is determined by linking the environmental conditions affecting soil characteristics with those present at specific locations. This merges the GIS database with the connections between soil environments and soil classification libraries, calculating the similarity values between each pixel and the soil environment. The soil similarity vectors for each point are determined, enabling the expression of soil similarity across the entire region [33]. The environmental factors and data source accuracy required by the SoLIM model directly determine the quality of digital soil mapping. A fuzzy map, which combines multiple pixels with their soil type similarity values, represents the continuous classification of soil types within the entire soil-environment landscape. The principle is shown in Figure 5.

3.4. Carbon Stock Estimation in Wetlands Based on Improved SoLIM-InVEST Modeling

The InVEST BCS is an advanced tool based on a spatially explicit model designed to assess the role of marine ecosystems in the carbon cycle and provide scientific evidence on BCS and emission reduction potential for policymakers and managers. The blue carbon module typically assumes a constant soil depth (1 m), meaning it does not set different depth values across space and does not assume uniform carbon density at all depths. However, this assumption fails to fully account for the differences in carbon stocks resulting from variations in soil profile depth, potentially affecting the accuracy and reliability of BCS estimates. To improve estimation accuracy, this study integrates the SoLIM-InVEST model, combining land use data with soil thickness data and incorporating field-sampled carbon density data into the InVEST model to provide a more refined assessment of wetland BCS. Carbon densities differ based on soil thickness and land use type, with specific values in

Table 1. Carbon storage values for various land cover categories and soil depths (unit: t/ha).

Code	Land Cover Type and Soil Depth	Soil Carbon Density	Biomass Carbon Density
11	P. australis 0 –20 cm	57.27	3.83
12	P. australis 20–40 cm	52.55
13	P. australis 40–60 cm	45.39
14	P. australis 60–80 cm	45.21
21	S. salsa 0 –20 cm	48.26	1.64
22	S. salsa 20–40 cm	47.39
23	S. salsa 40–60 cm	40.94
24	S. salsa 60–80 cm	40.23
31	Tidal flat 0–20 cm	46.58	0
32	Tidal flat 20–40 cm	44.82
33	Tidal flat 40–60 cm	44.56
34	Tidal flat 60–80 cm	39.60
41	Cropland 0–20 cm	0	7.92
42	Cropland 20–40 cm	0
43	Cropland 40–60 cm	0
44	Cropland 60–80 cm	0
51	Other land 0–20 cm	0	0
52	Other land 20–40 cm	0
53	Other land 40–60 cm	0
54	Other land 60–80 cm	0

. The blue carbon module considers four carbon pools: aboveground biomass (carbon in living vegetation), belowground biomass (carbon in roots and rhizomes), sediment/soil (organic carbon in soil), and dead/litter biomass (carbon in litter and dead vegetation). This study merges aboveground and belowground biomass into a single biomass carbon pool, excluding the litter pool due to data limitations [48]. The formula used to calculate carbon stock is presented below:

$$\begin{array}{c}{C}_i=C_{i-above}+C_{i-below}+C_{i-dead}+C_{i-soil}\\ C_{i-total}=C_i\times A_i\end{array}$$

(6)

where $C_i$ represents the overall carbon density associated with the i land cover category; $C_{i-soil}$, $C_{i-dead}$, $C_{i-above}$, and $C_{i-below}$ represent the soil organic content, dead organic content, aboveground, and belowground carbon density of the i land cover type; $C_{i-total}$ represents the overall carbon storage within the i land use category; and $A_i$ represents the area corresponding to the i land cover type.

The carbon pool information is primarily derived from on-site observations, with additional support from data gathered in the literature for comparative purposes, including studies by Rongxin Tong, Guangshuai Zhang, and others on the carbon pools in the study area [49]. The soil carbon pool was determined by following the guidelines outlined in the “Soil Sample Collection, Preparation, and Preservation Technical Regulations” [50] and “Ecological survey and Monitoring Standards” [51]. Soil samples were collected from investigation plots established within the wetland communities of the Liaohekou Nature Reserve, including P. australis, S. salsa, and others. These samples were then measured using the loss-on-ignition method, following the standard “Measurement of Organic Matter in Solid Waste: Loss on Ignition Method” (HJ761–2015) [52]. The collected soil samples were oven-dried at 105 °C until a stable weight was achieved; then, approximately 1–2 g of the dried soil was placed in a crucible and heated at 550 °C for four hours inside a muffle furnace until it reached a constant weight. The conversion coefficient of the organic matter and the organic carbon is 1.724 [53], and the soil bulk weight takes the value of 1.45 g/${\mathrm{c}\mathrm{m}}^3$ [54]. The calculation formula is as follows:

$$\begin{array}{c}w={\displaystyle \frac{m_0-m_1}{M}}\times 100\\ C_b=w÷1.724\\ D_a=B_a\times C_c\times \mathrm{h}\end{array}$$

(7)

Here, w represents the proportion of organic material in the dry mass (%); $m_0$ represents the combined weight of the crucible and the sample after drying (g); $m_1$ denotes the weight of the crucible and the dried sample after being combusted (g); M represents the weight of the sample before drying (g); and 100 serves as the conversion factor for units. $C_b$ represents the organic carbon content, with 1.724 acting as the conversion coefficient between organic carbon and soil organic content. $D_a$ denotes the amount of carbon stored in the soil (g/${\mathrm{c}\mathrm{m}}^2$); $B_a$ is the soil bulk weight (g/${\mathrm{c}\mathrm{m}}^3$), and generally takes the value of 1.45 g/${\mathrm{c}\mathrm{m}}^3$; $C_c$ refers to the observed carbon concentration within the soil (%); and h denotes sampling depth (cm).

The biomass carbon pool primarily consists of living vegetation above the ground and root systems, including rhizomes, beneath the surface. Its carbon storage is estimated by determining the dry weight of each component and applying the appropriate carbon conversion factor. The vegetation survey plots in this study were conducted simultaneously with the soil survey, with 10 plots of 1 m × 1 m set up in P. australis and S. salsa communities. Differential GPS was employed to pinpoint the plot locations, and all vegetation within these plots was harvested entirely. The collected vegetation was then transported to the laboratory for the determination of its dry weight and carbon content. The procedures followed the “Technical Specification for Coastal Wetland Ecological Monitoring (HY/T080–2005)” [55]. The calculation can be expressed using the following formula:

$$C_{abo}={\displaystyle \sum _{n=1}^i\omega c_{orge,i}\times M_{spe,i}}\times S_i/\left(S_{sp,i}\times 10\right)$$

(8)

where $C_{abo}$ represents the carbon storage within the biomass carbon pool; ${\omega c}_{orge,i}$ represents the proportion of organic carbon contained in the biomass sample within the i sampling unit; $M_{spe,i}$ denotes the dry biomass weight of the sample within the i sampling unit; $S_i$ represents the sampling unit’s surface area; and $S_{sp,i}$ denotes the coverage area of plant samples within the i sampling unit.

This study overlays the soil thickness values simulated by SoLIM with land use type data for the study area. By integrating the carbon density data obtained from measurements, the refined estimation of wetland BCS is conducted within the InVEST modeling framework. The land use classification and carbon density information for the study region in 2023 are presented in Figure 6 and Table 1.

4. Results and Analysis

4.1. Prediction and Accuracy Verification of Soil Thickness in Liaohekou

This study utilizes the covariate functionality module of the SoLIM software to import an environmental factor dataset related to soil conditions, comprising terrain factors (DEM, slope, aspect, plan curvature, profile curvature, and TWI), remote sensing factors (NDVI), and hydrological factors (IIC and PCI) as covariates. These factors can effectively reflect the spatial variability characteristics of the soil growth environment. Representative sample points are then selected, and their location and attribute information are extracted as soil sample data to be imported into the model. The covariates are quantified through fuzzy membership and record the spatial variation in soil environmental conditions, while the sample point data precisely record specific soil properties such as soil type and thickness. After multiple parameter adjustments and model optimizations, the predicted soil thickness map obtained is shown in Figure 7.

The soil thickness ranges from 0 to 80 cm and is evenly divided into four levels, each with a 20 cm interval. Soil thickness correlates with color intensity, where darker shades indicate greater depth, while lighter tones signify shallower layers. As shown in Figure 6, The pattern of soil thickness distribution is well-defined, with the majority of the area exhibiting a soil depth within the range of 40–60 cm, constituting 52% of the entire study region, approximately 66,916.2 ha. Conversely, areas where soil thickness ranges between 60–80 cm and 20–40 cm occupy smaller proportions of the total area, 34%, and 13%, corresponding to approximately 43,752.9 ha and 16,729.1 ha, respectively. The 0–20 cm soil layer is the least common, primarily in densely populated urban residential areas, covering only 1% of the study area, about 1286.9 ha. Areas covered by P. australis tend to have thicker soil, with thickness values generally falling within the 40–80 cm range. Coastal tidal flat areas, on the other hand, exhibit relatively thinner soils, typically within the 20–60 cm range. This could be due to the intricate root systems of P. australis, which exhibit superior soil stabilization capabilities. The soil thickness along the eastern coastline is somewhat less than that found along the western coastline, which may be related to the direction of river flow and associated geological processes. The eastern shore faces the main flow direction of the river, where erosion is more pronounced. The river’s downcutting and lateral erosion processes lead to soil layer removal, resulting in thinner soil. On the western shore, the river flow is relatively slower, and sedimentation is more significant, gradually increasing soil layer thickness. The variation in soil thickness across the area is largely shaped by the interplay of the natural environmental factors and human-induced activities. The Liaohekou region, an estuarine wetland with significant sediment deposition, experiences changes in soil thickness as a result of alterations in the river’s course and vegetation development.

In order to assess the reliability of the soil thickness estimation model, this research identified 40 representative sampling locations within the Liaohekou region for validation. To ensure the representativeness of the sample data, the selected points cover typical areas with various plant characteristics, water features, and topographical positions in the Liaohekou. These sampling locations provide an accurate representation of how different plant species affect soil depth under diverse growth environments, and the impact of varying distances from water sources. Specifically, this includes the P. australis, S. salsa, and tidal flat sample zones in typical wetland types, such as along the coastline, around ponds, areas distant from water sources, wetland mudflats, and near river mouths. This study collected data from 40 sampling points for model validation. The results showed that for 35 sampling points (87.5%), the measured soil thickness corresponded to the same soil level as predicted by the SoLIM model, indicating a high prediction accuracy. Error analysis revealed five anomalous predictions: for three sampling points, the predicted values deviated from the measured values by 20 cm (one soil level), while for two sampling points, the deviation reached 40 cm (two soil levels). In general, the verification outcomes demonstrate that the SoLIM demonstrates strong capability in precisely predicting soil thickness, making it a suitable tool for estimating soil thickness.

4.2. Estimation Results and Accuracy Analysis of BCS in the Research Area

This study uses the spatial analysis function of ArcGIS to link the land use type map with the soil thickness layer, differentiating land cover types with varying soil thicknesses by using different fields. Combining carbon density values at different soil depths, the BCS for 2023 is estimated under the InVEST model, as shown in Figure 8.

In 2023, the total BCS in the Liaohe Estuary was 389.85 × 10⁶ t. The areas with the highest carbon stocks were primarily concentrated in the northern region of the Liaohe Estuary, including P. australis, a tidal flat, S. salsa, and cropland, which exhibited higher carbon capture efficiency. In contrast, regions with lower carbon stocks were predominantly situated along the southern coastline, encompassing rivers, aquaculture ponds, reservoirs, bare land, ponds, and built-up land. Low-value areas are associated with multiple wetland types, but their distribution is limited in area. In contrast, high-value areas are primarily concentrated in wetlands with richer vegetation cover, demonstrating a positive correlation between vegetation types and BCS. In particular, the northern vegetated wetlands played a key role in blue carbon storage. In contrast, areas with intense human activity (such as built-up land) exhibited lower BCS.

Soil thickness is one of the key factors influencing BCS. According to the soil thickness prediction map, areas with thicker soil generally have higher BCS. This phenomenon is strongly linked to the spatial arrangement of soil carbon stocks, as organic carbon in the soil serves as the primary form of carbon storage. Thicker soil layers typically have higher carbon densities and more extensive organic matter storage capacity, facilitating long-term carbon accumulation and sequestration, particularly in wetland ecosystems where increased soil thickness is often associated with more carbon deposition. Additionally, the contribution of wetland vegetation types to BCS should not be overlooked. As a typical blue carbon plant, P. australis has a significantly higher carbon storage capacity than S. salsa. During the growing season, P. australis can rapidly accumulate large amounts of organic matter. Its well-developed root systems anchor the soil, preventing carbon loss and promoting long-term carbon accumulation. In contrast, S. salsa has a lower biomass, weaker soil-binding capacity, and limited carbon capture ability, making its contribution to BCS less significant than P. australis. Moreover, human activities significantly influence the spatial distribution of BCS, mainly through changes in land use. The soil in the southern coastal areas of the study region has undergone degradation due to human disturbances, leading to the destruction of wetland ecosystems and a significant reduction in carbon accumulation capacity. Development activities in coastal areas significantly impact BCS, highlighting the urgent need for enhanced monitoring and research.

5. Discussion

5.1. Prediction and Accuracy Verification of Soil Thickness in Liaohekou

In recent years, the SoLIM model has demonstrated strong applicability in digital soil mapping [56], and its coupling with other models has also shown significant scientific value. Lei et al. combined SoLIM with the SWAT, providing an effective solution for hydrological modeling in areas with limited high-precision soil data [57]. They further revealed how soil data uncertainty influences the SWAT hydrological simulation results. Although previous studies have applied SoLIM to wetland soil analysis [58], this study builds upon that foundation by enhancing the accuracy of the environmental factor dataset to improve model prediction precision. It further investigates the spatial distribution characteristics of soil thickness and its influencing factors while exploring the applicability of soil thickness data in the refined estimation of BCS in coastal wetlands. A new soil thickness estimation method was developed by integrating soil–landscape modeling theory and field survey data and introducing remote sensing and hydrological factors as key variables of the wetland soil environment. The validation results show that the soil thickness prediction accuracy reaches 87.5%, demonstrating the applicability of the SoLIM model in wetland environment research. Compared to traditional spatial interpolation methods (such as kriging and inverse distance weighting), this model can still rapidly generate high-precision soil thickness information under limited sample points. While maintaining prediction accuracy, it significantly reduces both time and economic costs. The predicted results closely match the actual soil thickness distribution. This discovery broadens the scope of the SoLIM model’s application and provides an important methodological reference for soil thickness research. Moreover, it offers an efficient tool for soil studies in wetland ecosystems.

Some discrepancies between the predicted results and the actual observations were found at specific sample points. These discrepancies may be attributed to several factors: inevitable random errors during field sampling, which could arise from variations in the field environment (such as microtopography or vegetation cover differences) and human subjective factors (such as sampling point selection bias), and second, the complexity of the soil formation process. This study only considered terrain, hydrological, and remote sensing factors as the primary environmental drivers of soil formation in wetlands and did not fully account for other potential influencing factors, which may limit the prediction accuracy.

5.2. Potential Ecological Impacts of Wetland Soil Thickness

Influenced by river erosion and sedimentation, the spatial variation in soil thickness exhibits specific patterns. Soil thickness is generally concentrated between 40 and 60 cm, with relatively fewer areas falling within the 20–40 cm and 60–80 cm ranges, and the smallest frequency of occurrence in the 0–20 cm range. Additionally, the soil thickness in the western part exceeds that in the eastern part. The variation in soil thickness results from the combined influence of multiple environmental factors, with hydrological conditions being a key driver in the development of wetland soils. The role of plant roots in soil development varies significantly depending on the vegetation type [59,60]. The roots of P. australis are intricate and have strong soil-binding abilities, predominantly found in areas where the soil thickness ranges from 40 to 80 cm. In contrast, the roots of S. salsa are shallower, distributed in regions with soil thickness between 20 and 60 cm. The influence of soil thickness on blue carbon sequestration efficiency and its ecological functions varies significantly. Thicknesses ranging from 40 to 80 cm provide more stable environmental conditions, exhibit lower microbial activity, and enhance carbon sequestration efficiency, playing a crucial role in supporting the stability of wetland ecosystems. Soils with 0–20 cm thickness are influenced by environmental factors and anthropogenic actions, exhibiting shorter carbon retention times, lower carbon sequestration efficiency, and a more sensitive response to ecosystem changes. The main sources of carbon input are root decomposition and vertical migration. In BCS estimation, soil thickness is a key parameter for representing organic carbon’s vertical distribution and storage potential. Its spatial heterogeneity directly influences the spatial pattern of carbon stocks. An in-depth exploration of the driving mechanisms behind soil thickness spatial variation, including hydrological dynamics, vegetation characteristics, and topographical conditions, is essential for enhancing the precision of blue carbon assessments and optimizing wetland carbon sink management strategies.

The human settlements in the study area are located in regions with lower blue carbon storage, and the soil thickness is within the range of 0–20 cm, thus significantly affecting the wetland’s carbon sequestration function. Human activities in the Liaohe Estuary region, such as reservoir construction, coastal aquaculture, and oil and gas development, have significantly altered the hydrological conditions and vegetation cover of the wetland, thereby having a significant impact on the soil carbon storage of the wetland [61,62]. Specifically, these activities have caused river disconnection and groundwater level decline, promoting soil desiccation in the wetland, thereby accelerating organic carbon decomposition and reducing the input of organic carbon. In addition, the reduction in vegetation biomass and surface exposure further weakened the carbon storage capacity of the wetland. At the same time, the accumulation of inorganic carbon in deep soils has intensified, leading to a shift in the wetland’s carbon storage structure from organic carbon dominance to inorganic carbon dominance. This change has significantly impacted the wetland’s carbon sequestration function, leading to its degradation. Therefore, reducing the negative impact of human activities on wetlands and accurately estimating blue carbon storage is of great significance for the ecological restoration and carbon management of coastal wetlands.

5.3. Estimation of BCS in Coastal Wetlands

In recent years, coupling the InVEST model with other models to improve the accuracy of carbon stock estimation has become a trend. Several studies have confirmed that model coupling, compared to a single model, can significantly improve the accuracy and reliability of the estimation results [63,64,65]. By integrating the strengths of different models, the coupling method effectively overcomes the limitations of single models in addressing complex environmental factors and spatial heterogeneity, thus optimizing the spatial prediction of carbon stocks [66]. The spatial variation in soil thickness and its significant differences across different wetland types are key factors influencing the distribution of organic carbon density, which significantly impacts the accuracy of BCS estimation [67]. This study addresses the limitation of the InVEST blue carbon module, which does not fully account for the spatial heterogeneity of soil thickness in BCS estimation for coastal wetlands. We innovatively combine soil thickness data with the InVEST model, establishing a refined BCS estimation method based on the SoLIM-InVEST coupled model for coastal wetlands. This method provides a reliable tool for soil thickness prediction and BCS estimation. However, it can also be applied to other coastal wetland ecosystems, offering methodological insights and practical references for regional-scale wetland soil carbon stock studies.

A comparison was made between the BCS estimates obtained in this study and those from previous research [68,69,70] and the estimates in this study are generally higher. This discrepancy may be due to the following factors. First, previous studies typically relied on carbon density reference data from the same region or regions with similar latitudes. These values are lower than the carbon density measurements obtained in this study, leading to higher carbon stock estimates in this work. Second, the InVEST model used in this study can estimate both vegetation and soil carbon stocks, while some studies focus only on the carbon stocks of a single carbon pool. The limitations of this approach may lead to lower estimation results. Furthermore, this study takes into account the spatial variability of soil in estimating BCS. At the same time, traditional methods typically calculate carbon stocks only for surface soils (<20 cm), neglecting the carbon storage potential of deeper soils. This further explains the higher estimates in this study. Mason et al. [15] noted significant differences in carbon stock and accumulation rates among salt marshes in different regions worldwide. The east coasts of North America and Australia have high carbon stock values due to the combined effects of relative sea level rise gradients, abundant sediment input, and the widespread distribution of high-productivity vegetation such as Spartina alterniflora. The Liaohe Estuary wetland, located adjacent to the Bohai Sea, is significantly influenced by tidal dynamics, and high-productivity plants such as P. australis and S. salsa are widely distributed in the region. These factors may collectively promote organic carbon accumulation, contributing to this study area’s higher BCS. These differences indirectly confirm that the BCS estimates in this study are more in line with the actual conditions of Liaohekou, thereby enhancing the scientific rigor and reliability of the methodology.

5.4. Limitations and Future Outlook

This research acknowledges several limitations and uncertainties that require further refinement and resolution in subsequent studies:

(1) Given the flat terrain and slight elevation differences in coastal wetlands, the accuracy of DEM needs to be further improved by using higher-resolution remote sensing imagery. Airborne LiDAR (Light Detection and Ranging) technology could be utilized to obtain terrain data with sub-meter precision, or multispectral and hyperspectral remote sensing data could be combined to identify surface features more accurately. The integrated use of these technologies will help capture the spatial distribution patterns of wetland soil thickness more accurately, thus significantly enhancing the reliability and generalizability of model predictions.

(2) In the process of blue carbon estimation, regional variability is a crucial factor that cannot be overlooked. Significant differences exist in the carbon storage mechanisms and environmental driving factors among different wetland types; thus, a high-precision blue carbon calculation model must be developed for multiple wetland types. Furthermore, human activities (such as wetland reclamation, landfilling, and water pollution) are increasingly disturbing wetland ecosystems, particularly in coastal wetlands, which are highly impacted by human influences. Future research should quantitatively assess the negative impacts of human activities on BCS and explore synergistic pathways for wetland ecosystem restoration and BCS enhancement, providing a scientific basis for wetland protection and sustainable management.

(3) This study’s BCS estimation largely relies on static or short-term data, making it challenging to fully assess the long-term cumulative effects of human activities on land use change. Future research should incorporate historical remote sensing data and long-term monitoring records and use coupled prediction models to reveal the dynamic evolution of soil thickness and BCS, further analyzing the spatiotemporal distribution characteristics of BCS and their driving mechanisms, which will provide a more reliable theoretical foundation for refined estimation and prediction of BCS. Furthermore, the sampling in this study was conducted during the plant growing season, when aboveground biomass and carbon stock are relatively high. However, during the non-growing season (e.g., winter), aboveground vegetation may die back or decrease, which could lead to seasonal variations in BCS in the study area. This dynamic change was not considered in the current study, which may affect the comprehensive assessment of the annual BCS. Therefore, future studies should conduct sampling across different seasons to reveal the intra-annual variation characteristics of wetland BCS, thereby improving the accurate understanding of carbon sequestration functions.

The results of this research provide a robust scientific basis for effective coastal wetlands management strategies and the promotion of steady BCS growth, offering theoretical support for governmental agencies in formulating carbon emission policies and implementing wetland restoration projects. Specifically, soil thickness, as a key factor affecting BCS, should be prioritized in coastal wetland management, mainly to prevent soil erosion and carbon loss caused by human activities. Furthermore, differentiated protection strategies should be implemented based on varying soil thicknesses across regions. In areas with thicker soils and higher carbon stocks, priority should be given to designating ecological protection zones, restricting development activities, and enhancing soil carbon sequestration through vegetation restoration and hydrological regulation. Human interference should be strictly controlled in areas with thinner soils and more fragile ecosystems to prevent further degradation. Lastly, the scientific management of BCS should be strengthened, and wetland protection and restoration plans based on soil carbon sequestration functions should be developed to support achieving regional carbon neutrality goals. In conclusion, the close relationship between soil thickness and wetland blue carbon management implies that the protection and management of coastal wetlands require multidisciplinary collaboration. By integrating remote sensing technology and model predictions, more scientifically sound management strategies can be developed for wetland ecosystems to tackle the combined challenges of climate change and ecological degradation.

6. Conclusions

This study optimizes traditional blue carbon spatial quantification methods in wetland BCS calculations by creating a link between soil thickness and carbon density. The calculations fully account for the spatial variability in soil thickness in wetlands, effectively coupling SoLIM-InVEST models, which improves the precision and accuracy of BCS estimations results. The main conclusions are as follows:

(1) The distribution of soil thickness in the Liaohekou study area exhibits clear spatial patterns. Most of the region has soil thicknesses between 40–60 cm and 60–80 cm, which account for 86% of the total study area, approximately 110,669.1 ha. Soil layers of 20–40 cm are relatively sparse, covering 13% of the study area, approximately 16,729.1 ha. Soil layers of 0–20 cm are the least common, occupying only 1%. Additionally, the soil thickness in the P. australis wetlands is generally more significant than in the S. salsa wetlands, and the soil thickness on the eastern shore is generally shallower than on the western shore. The soil thickness in urban residential areas is the shallowest, closely related to human activities and natural environmental factors.

(2) The coupled SoLIM-InVEST model compensates for the rough expression of the influence of soil thickness on soil carbon density. It enables a more refined estimate of BCS. The calculated BCS for the Liaohekou study area 2023 is 389.85 × 10⁶ t. Regarding spatial distribution, areas with higher BCS per unit are mainly situated in the northern region, with wetland types including P. australis, tidal flats, S. salsa, and cropland. Areas with lower BCS are concentrated along the southern coastal areas, with wetland types mainly consisting of construction land, river water areas, bare soil, aquaculture ponds, pits, shallow marine waters, and reservoirs, and predominantly characterized by construction land. This research suggests that regions with increased soil depth generally exhibit higher BCS.