Method for estimating soil carbon storage in vegetation-covered areas of tidal wetlands
By constructing a vegetation functional zoning system and a biomass inversion model, and combining remote sensing data and soil samples, the problems of efficiency and reliability in estimating soil carbon storage in tidal flat wetlands were solved, enabling accurate assessment of deep carbon storage and supporting blue carbon measurement and wetland management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- STATE OCEANIC ADMINISTRATION NANTONG MARINE ENVIRONMENT MONITORING CENT STATION
- Filing Date
- 2026-02-27
- Publication Date
- 2026-06-02
AI Technical Summary
Existing technologies for estimating carbon storage in tidal flat wetland soils suffer from high costs, insufficient representativeness, and difficulty in achieving dynamic monitoring. Traditional field sampling methods are inefficient and prone to large errors, while remote sensing methods lack reliable physical models for estimating deep carbon storage.
By combining temporal multispectral remote sensing image data with soil stratification sample data from the root zone of vegetation, a vegetation functional zoning system was constructed. Through biomass inversion model and machine learning, a quantitative relationship between aboveground biomass and soil carbon storage was established, and the soil carbon storage of the whole profile was estimated. By utilizing localized parameters based on ecological knowledge of salt marsh wetlands, a high-quality simulation dataset was generated for refined estimation.
It enables efficient, accurate, and large-scale assessment of soil carbon storage in tidal flat wetland vegetation areas, providing operable technical tools to support blue carbon metering and wetland management.
Smart Images
Figure CN122132747A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of soil carbon storage estimation technology, specifically a method for estimating soil carbon storage in tidal flat wetland vegetation-covered areas. Background Technology
[0002] Currently, the mainstream method for assessing soil carbon storage mainly relies on traditional field sampling. This method involves setting up sampling points in the study area, obtaining soil samples, and measuring parameters such as organic carbon content and bulk density in the laboratory. Then, the regional carbon storage is estimated through spatial interpolation. This method is widely used in terrestrial ecosystems such as forests and grasslands, but when it is directly applied to tidal flat wetlands, it reveals inherent defects that are difficult to overcome: (1) poor feasibility and high cost: Tidal flat wetlands have complex terrain, are difficult to access, and are affected by tidal cycles. Field sampling requires a lot of manpower, material resources, and time, resulting in extremely low sampling efficiency; (2) insufficient representativeness and significant errors: Due to the difficulty of sampling, only sparse and discontinuous sampling point data can often be obtained. Tidal flat wetland soils are highly spatially heterogeneous due to the coupling influence of multiple factors such as hydrology, vegetation, and sedimentation. Using sparse point data to "represent the whole area" makes it difficult to truly reflect the spatial distribution pattern of carbon storage, resulting in significant uncertainty in the assessment results; (3) It is difficult to achieve dynamic monitoring: traditional methods cannot meet the management needs of rapid and periodic monitoring of carbon storage in large-scale tidal flat wetlands to assess its dynamic changes.
[0003] To overcome the limitations of traditional methods, remote sensing technology has been introduced due to its advantages of large-scale and periodic observation. However, existing pure remote sensing inversion methods also face severe challenges when applied to the estimation of soil carbon storage in tidal flat wetlands: (1) lack of direct spectral response mechanism: soil organic carbon lacks direct and sensitive diagnostic spectral features such as vegetation chlorophyll in remote sensing spectra. Especially under vegetation cover, soil signals are severely interfered with, making it difficult to establish a robust direct spectral inversion model; (2) helpless in dealing with deep carbon storage: remote sensing signals mainly reflect surface or shallow information. For soil carbon pools buried at depths (such as 1 meter), there is a lack of effective physical models for direct detection and quantitative inversion; (3) contradiction between model universality and accuracy: empirical models based on statistical relationships are effective in specific areas, but are limited by training samples and have poor spatiotemporal universality; while models based on physical mechanisms often have poor accuracy in the special habitat of tidal flat wetlands due to complex parameters and difficulties in localization.
[0004] In summary, existing technologies present a dual dilemma: "traditional methods cannot achieve efficient large-area assessments" and "remote sensing methods cannot guarantee the reliability of deep estimation mechanisms." Therefore, there is an urgent need in this field for an innovative technical solution that can organically integrate the efficiency of remote sensing technology with the reliability of traditional methods, enabling accurate, efficient, and large-scale estimation of soil carbon storage (including deep carbon pools) in tidal flat wetlands, particularly in their vegetated areas, through an innovative technical approach. Summary of the Invention
[0005] To address the dual challenges of high cost and insufficient representativeness of traditional field sampling methods in large-scale tidal flat wetland applications, and the lack of reliable physical mechanisms for estimating deep soil carbon storage using existing pure remote sensing inversion methods, this invention proposes a method for estimating soil carbon storage in tidal flat wetland vegetation-covered areas. This method enables mechanistic, refined, and large-area remote sensing estimation of soil carbon storage in tidal flat wetland vegetation zones.
[0006] To achieve the above objectives, the present invention is implemented through the following technical solution:
[0007] This invention relates to a method for estimating soil carbon storage in tidal flat wetland vegetation cover areas, comprising:
[0008] Acquire temporal multispectral remote sensing image data of the target area and corresponding soil stratification sample data of the vegetation root zone;
[0009] A vegetation functional zoning system was constructed based on temporal multispectral remote sensing image data.
[0010] A biomass inversion model was constructed and inverted to obtain the spatial distribution of aboveground biomass in the target area.
[0011] Spatiotemporal matching of aboveground biomass and measured soil carbon storage data in the target area was performed to analyze and establish a quantitative relationship model between aboveground biomass and soil carbon storage at a specific depth.
[0012] Based on the spatial distribution and quantitative relationship model of aboveground biomass in the target area, the carbon storage of shallow soil is estimated.
[0013] Based on the statistical relationship between shallow and deep soil carbon storage, the total soil carbon storage of the entire profile is estimated.
[0014] A further improvement of this invention lies in: constructing and performing a biomass inversion model to obtain the spatial distribution of aboveground biomass in the target area, including:
[0015] A simulated dataset was generated using a localized vegetation radiative transfer model.
[0016] Based on a simulated dataset, a machine learning model is trained to obtain a biomass inversion model.
[0017] The spatial distribution of aboveground biomass in the target area is obtained by using a biomass inversion model.
[0018] A further improvement of the present invention is that the localized vegetation radiation transfer model is obtained by: applying prior ecological knowledge of the stress of tidal salt water on vegetation in salt marsh wetlands to the biophysical parameters of the vegetation radiation transfer model, thereby forming a localized parameter space that conforms to the habitat conditions of the target area.
[0019] A further improvement of the present invention is that a non-uniform sampling strategy based on parameter sensitivity is used when generating the simulated dataset, including:
[0020] Sensitivity analysis was performed on the constrained biophysical parameters to identify a subset of core parameters;
[0021] The core parameter subset is sampled within the constrained value range using the first sampling density, while the parameters of the non-core parameter subset are sampled using a second sampling density lower than the first sampling density or assigned a fixed typical value.
[0022] A further improvement of the present invention is that: in the simulated dataset, the aboveground biomass label corresponding to each simulated canopy reflectance spectral data is calculated synchronously based on the leaf area index parameter value and the leaf dry matter content parameter value through a preset biomass estimation function.
[0023] A further improvement of this invention lies in: constructing a vegetation functional zoning system based on temporal multispectral remote sensing image data, including:
[0024] Classify temporal multispectral remote sensing image data to divide vegetation-covered areas into non-vegetated areas;
[0025] The normalized vegetation index time series of each pixel in the vegetation cover area is calculated. Based on the phenological characteristics of the normalized vegetation index time series, the vegetation cover area is divided into primary partitions according to vegetation type using a classification algorithm.
[0026] Based on aboveground biomass thresholds, carbon sink potential sub-regions are divided on the basis of primary zoning.
[0027] A further improvement of this invention is that the quantitative relationship model is a linear empirical model of aboveground biomass and soil carbon storage at a depth of 10-50 cm.
[0028] A further improvement of the present invention is that, based on the statistical relationship between shallow soil carbon storage and deep soil carbon storage, the total profile soil carbon storage is estimated, including: based on each zone of the vegetation functional zoning system, a conversion model between shallow soil carbon storage and total profile carbon storage is established using the corresponding conversion coefficient, and the total profile soil carbon storage of the corresponding zone is estimated using the conversion model.
[0029] The beneficial effects of this invention are as follows: By transforming the ecological stress knowledge unique to salt marsh wetlands, such as tides and salinity, into prior constraints on the biophysical parameters of the PROSAIL vegetation radiative transfer model, this invention solves the fundamental problems of parameter blindness and simulation distortion in the application of the PROSAIL model in tidal flat wetlands, providing high-quality training data that conforms to local ecological mechanisms for machine learning models. This invention collaboratively constructs a simulated dataset by using parameter-sensitive non-uniform sampling and synchronously calculating biomass labels using model input parameters. This ensures the representativeness of the simulated dataset on key biomass gradients and the accuracy of the spectral-label correspondence from the source, significantly improving the reliability and generalization ability of the biomass inversion model. This invention finely characterizes the impact of different growth patterns of the same plant species on carbon input by dividing carbon sink potential sub-regions. By establishing a linear relationship between aboveground biomass and soil carbon storage, it achieves indirect, mechanistic inversion from remotely sensed vegetation information to soil carbon storage. This invention provides a complete technical process from multi-source data collaborative processing, mechanism-data fusion modeling, quantitative relationship establishment to final carbon storage mapping and scale extrapolation. This method mainly relies on publicly available remote sensing data and limited validation samples. The process is clear and highly repeatable. It has successfully achieved efficient, accurate and large-scale assessment of soil carbon storage in large, hard-to-access tidal wetland vegetation areas, providing a powerful and operable technical tool for blue carbon measurement, wetland protection and management decision-making. Attached Figure Description
[0030] Figure 1 This is a flowchart of the method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the fitting results between vegetation biomass and soil carbon storage in an embodiment of the present invention. Detailed Implementation
[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.
[0032] like Figure 1 As shown, in conjunction with a specific salt marsh wetland, the method for estimating soil carbon storage in the vegetation-covered area of a tidal flat wetland in this embodiment includes the following steps:
[0033] S1: Acquire temporal multispectral remote sensing image data of the target area and corresponding soil stratification sample data of the vegetation root zone, and perform collaborative preprocessing.
[0034] This embodiment uses Landsat series satellites to acquire temporal multispectral remote sensing image data of the target area during the growing season (August-November) from 2013 to 2023. The temporal multispectral remote sensing image data is preprocessed, including radiometric calibration, atmospheric correction (using the FLAASH model), and geometric correction, to ensure the accuracy and spatial consistency of the surface reflectance data, with the geometric registration error controlled within 0.5 pixels.
[0035] Sampling points were set up along eight typical cross-sections in the target area, with no fewer than five points per cross-section. A 1-meter deep columnar soil sample was collected at each point, layered at 10-centimeter intervals. The precise location and elevation of the sampling points were recorded using RTK-GPS. Laboratory analysis of the soil samples was used to obtain data on the organic carbon content, bulk density, and sediment composition of each soil layer.
[0036] S2, Constructing and retrieving a biomass inversion model. Specifically, this includes:
[0037] The parameter space of the PROSAIL model for salt marshes was optimized, and a simulation dataset was generated using the optimized PROSAIL model. Based on the ecological knowledge of salt marsh vegetation under tidal saline stress, prior constraints were imposed on the biophysical parameters of the PROSAIL model to compress its theoretical value range and form a parameter space that conforms to local habitat conditions. The parameters of the PROSAIL model in this embodiment are shown in Table 1.
[0038] Table 1: Parameters of the PROSAIL model
[0039]
[0040] Within the constrained parameter space, highly sensitive parameters were identified through sensitivity analysis. This included local sensitivity analysis using the OTA algorithm and global sensitivity analysis using the EFAST method. The results showed that the highly sensitive parameters for *Spartina alterniflora* included leaf area index (LAI), leaf dry matter content (Cm), leaf inclination distribution parameter (LIDFA), leaf structure parameter (N), leaf equivalent water thickness (Cw), and chlorophyll content (Cab), while carotenoid content (Car) and soil background parameter (Psoil) were considered low-sensitivity parameters. The highly sensitive parameters for *Suaeda salsa* mainly included Cm, LAI, N, LIDFA, Cab, and Psoil, with Car and Cw showing lower sensitivity. The highly sensitive parameters for *Phragmites australis* were consistent with those for *Spartina alterniflora*. Leaf area index and leaf dry matter content were proven to be the two most crucial parameters for biomass estimation. LAI and Cm were sampled intensively at a first sampling density with a small step size, while parameters in the non-core parameter subset were sampled at a second sampling density lower than the first sampling density or assigned fixed typical values. The specific values of each parameter in this embodiment are shown in Table 2.
[0041] Table 2: Parameter values for the PROSAIL model
[0042]
[0043] The PROSAIL model is run to generate a large number of simulated canopy reflectance spectra. The aboveground biomass label corresponding to each simulated canopy reflectance spectrum is calculated synchronously by the input parameters driving this simulation and the preset biomass estimation function. The calculation expression is: AGB = LAI × Cm × 10.0, thereby generating a high-quality "canopy reflectance-belowground biomass" simulation dataset.
[0044] A backpropagation neural network (BP neural network) was trained using a simulated dataset to learn the complex mapping relationship between multi-band reflectance and aboveground biomass, resulting in a biomass inversion model. Using this biomass inversion model and preprocessed real Landsat reflectance images, a spatial distribution map of aboveground biomass across the entire vegetation cover area was retrieved.
[0045] S3, Constructing a vegetation functional zoning system. Specifically, this includes:
[0046] S3.1 Extracting vegetation cover areas: The maximum likelihood method is used to classify the preprocessed time-series multispectral remote sensing image data, and the target area is initially divided into vegetation cover areas and non-vegetation areas.
[0047] S3.2, Identify vegetation types and perform first-level zoning: Calculate the normalized difference vegetation index (NDVI) time series for each pixel within the vegetation cover area, expressed as:
[0048]
[0049] in: The spectral reflectance is in the near-infrared band. The spectral reflectance is in the red light band.
[0050] The random forest classification algorithm is used to identify vegetation types within the vegetation-covered area and to perform primary partitioning based on vegetation type. In this embodiment, the vegetation-covered area is divided into three partitions according to three dominant vegetation types: Spartina alterniflora, Phragmites australis, and Suaeda salsa.
[0051] S3.3, Based on the aboveground biomass threshold, carbon sink potential sub-regions are divided on the basis of the primary zoning. In this embodiment, to further characterize the differences in carbon input potential within the same vegetation type, the most widely distributed *Spartina alterniflora* area is further subdivided on the basis of the primary zoning. The aboveground biomass threshold, which maximizes the distinction between high and low carbon storage levels, is used to divide the carbon sink potential sub-regions. This embodiment utilizes the spatial distribution map of aboveground biomass obtained from S2 to analyze frequency distribution characteristics and determines the aboveground biomass threshold based on the analysis results. In this embodiment, the aboveground biomass threshold is 5.5 kg / m². *Spartina alterniflora* areas with aboveground biomass AGB ≥ 5.5 kg / m² are classified as *Spartina alterniflora* high carbon sink potential areas, and *Spartina alterniflora* areas with aboveground biomass AGB < 5.5 kg / m² are classified as *Spartina alterniflora* low carbon sink potential areas. Finally, a zoning system comprising four types of vegetation functional zones is formed.
[0052] S4. Establish a quantitative relationship model between aboveground biomass and soil carbon storage at a specific depth. Specifically, this includes:
[0053] S4.1, the historical aboveground biomass data retrieved over the years is spatiotemporally matched with the carbon storage data of soil sampling points with precise spatial locations to construct the "point AGB-point soil carbon storage" analysis dataset.
[0054] S4.2, based on the analyzed dataset, the relationship between aboveground biomass and soil carbon storage at different depths was analyzed. The results showed that the linear relationship between aboveground biomass and soil carbon storage at a depth of 10-50 cm was the most robust, with a coefficient of determination (R²) of 0.75. Furthermore, the regression slope in the 10-50 cm depth range showed a statistically significant difference compared to adjacent depth ranges. Based on this, a quantitative relationship model, namely a linear empirical model, was established: Where k is the regression slope and b is the intercept, the linear empirical model obtained in this embodiment is as follows: Figure 2 As shown.
[0055] S5, Target Area Soil Carbon Storage Estimation and Scale Extrapolation. Specifically includes:
[0056] S5.1 substitutes the latest spatial distribution map of aboveground biomass into the linear empirical model to generate a spatial distribution map of soil carbon storage at a depth of 10-50cm in the vegetated area, i.e., a shallow carbon storage distribution map.
[0057] S5.2, perform heterogeneous-scale extrapolation of carbon storage across the entire 1 m profile. A stable proportional relationship exists between soil carbon storage at a depth of 10-50 cm and the carbon storage at the 1 m full profile, but this relationship varies across different vegetation functional zones. In this embodiment, independent conversion coefficients are fitted for the reed zone, the Suaeda salsa zone, the Spartina alterniflora high carbon sequestration potential zone, and the Spartina alterniflora low carbon sequestration potential zone. Using the corresponding conversion coefficients, the shallow carbon storage distribution map is converted into the corresponding spatial distribution map of total organic carbon storage in the 1 m full profile, and the total amount is estimated.
[0058] This invention combines the efficiency of remote sensing with the reliability of mechanisms, overcoming the technical bottlenecks of traditional methods such as low efficiency and insufficient representativeness, as well as the weak mechanisms and difficulty in estimating deep carbon reserves of pure remote sensing methods. It provides an efficient, accurate and operational technical means for the assessment of blue carbon in tidal flat wetlands.
[0059] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.
[0060] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for estimating soil carbon storage in tidal flat wetland vegetation cover areas, characterized in that, include: Acquire temporal multispectral remote sensing image data of the target area and corresponding soil stratification sample data of the vegetation root zone; A biomass inversion model was constructed and inverted to obtain the spatial distribution of aboveground biomass in the target area. Based on time-series multispectral remote sensing image data, a vegetation functional zoning system is constructed; Spatiotemporal matching of aboveground biomass and measured soil carbon storage data in the target area was performed to analyze and establish a quantitative relationship model between aboveground biomass and soil carbon storage at a specific depth. Based on the spatial distribution and quantitative relationship model of aboveground biomass in the target area, the carbon storage of shallow soil is estimated. Based on the statistical relationship between shallow and deep soil carbon storage, the total soil carbon storage of the entire profile is estimated.
2. The method for estimating soil carbon storage in tidal flat wetland vegetation cover areas according to claim 1, characterized in that, A biomass inversion model was constructed and inverted to obtain the spatial distribution of aboveground biomass in the target area, including: A simulated dataset was generated using a localized vegetation radiative transfer model. Based on a simulated dataset, a machine learning model is trained to obtain a biomass inversion model. The spatial distribution of aboveground biomass in the target area is obtained by using a biomass inversion model.
3. The method for estimating soil carbon storage in tidal flat wetland vegetation cover areas according to claim 2, characterized in that, The localized vegetation radiation transfer model was obtained in the following way: based on the ecological prior knowledge of the vegetation in the salt marsh wetland being subjected to tidal salt water stress, prior constraints were imposed on the biophysical parameters of the vegetation radiation transfer model to form a localized parameter space that conforms to the habitat conditions of the target area.
4. The method for estimating soil carbon storage in tidal flat wetland vegetation cover areas according to claim 3, characterized in that, When generating the simulated dataset, a parameter-sensitive non-uniform sampling strategy is employed, including: Sensitivity analysis was performed on the constrained biophysical parameters to identify a subset of core parameters; The core parameter subset is sampled within the constrained value range using the first sampling density, while the parameters of the non-core parameter subset are sampled using a second sampling density lower than the first sampling density or assigned a fixed typical value.
5. The method for estimating soil carbon storage in tidal flat wetland vegetation cover areas according to claim 3, characterized in that, In the simulated dataset, the aboveground biomass label corresponding to each simulated canopy reflectance spectral data is calculated synchronously based on the leaf area index parameter value and the leaf dry matter content parameter value through a preset biomass estimation function.
6. The method for estimating soil carbon storage in tidal flat wetland vegetation cover areas according to claim 1, characterized in that, Based on temporal multispectral remote sensing image data, a vegetation functional zoning system was constructed, including: Classify temporal multispectral remote sensing image data to divide vegetation-covered areas into non-vegetated areas; The normalized vegetation index time series of each pixel in the vegetation cover area is calculated. Based on the phenological characteristics of the normalized vegetation index time series, the vegetation cover area is divided into primary partitions according to vegetation type using a classification algorithm. Based on aboveground biomass thresholds, carbon sink potential sub-regions are divided on the basis of primary zoning.
7. The method for estimating soil carbon storage in tidal flat wetland vegetation cover areas according to claim 1, characterized in that, The quantitative relationship model is a linear empirical model of aboveground biomass and soil carbon storage at a depth of 10-50 cm.
8. The method for estimating soil carbon storage in tidal flat wetland vegetation cover areas according to claim 1, characterized in that, Based on the statistical relationship between shallow and deep soil carbon storage, the total soil carbon storage of the whole profile is estimated. This includes: establishing a conversion model between shallow and total soil carbon storage for each zone based on the vegetation functional zoning system, using the corresponding conversion coefficients, and using the conversion model to estimate the total soil carbon storage of the corresponding zone.