A method and system for estimating carbon storage of a sonneratia caseolaris ecosystem

Abstract

The application provides a method and system for estimating carbon storage of a Sonneratia apetala ecosystem, comprising: obtaining a Sonneratia apetala single optimal community and constructing a global five-dimensional remote sensing feature set; obtaining a global tree height spatial distribution based on the global five-dimensional remote sensing feature set and a tree height inversion model; calculating a global aboveground biomass spatial distribution based on the global tree height spatial distribution, and converting the global aboveground biomass spatial distribution into a global aboveground carbon density spatial distribution by using a carbon content coefficient; respectively calculating a global underground carbon density spatial distribution and a global litter carbon density spatial distribution based on the global aboveground carbon density spatial distribution and a root crown ratio coefficient and a litter distribution coefficient; obtaining a global soil organic carbon density spatial distribution based on the five-dimensional remote sensing feature set and global mangrove soil organic carbon prior data; and finally obtaining a global carbon storage spatial distribution based on the aboveground carbon density, the underground carbon density, the litter carbon density and the soil organic carbon density. The application realizes high-precision estimation of the global carbon pool of the Sonneratia apetala single optimal community.

Description

Technical Field

[0001] This application relates to the field of carbon sink monitoring, and more specifically, to a method and system for estimating carbon storage in the *Sonneratia apetala* ecosystem. Background Technology

[0002] Mangroves are important blue carbon ecosystems, and assessing their carbon sequestration capacity is a crucial step in achieving the "dual carbon" goals. *Avicennia marina*, due to its rapid growth and strong adaptability, has been widely introduced for mangrove ecological restoration and has become an important component of mangroves in southern my country. However, its rapid expansion as an introduced species has also brought ecological management challenges. Accurately quantifying the carbon sequestration contribution of *Avicennia marina* is essential for scientifically assessing its ecological benefits and developing differentiated management strategies. Traditional plot survey methods are costly and inefficient, making them unsuitable for large-scale, dynamic monitoring. Remote sensing technology, with its macroscopic and rapid advantages, has become the mainstream method for estimating mangrove carbon storage.

[0003] However, existing remote sensing methods do not take into account the significant differences between high-productivity tree species such as Avicennia marina and native tree species, which leads to an underestimation of carbon storage in the spatial distribution area of ​​Avicennia marina and an overestimation in the native tree species area. The overall accuracy is not high, and it is impossible to provide accurate data for carbon sink trading or ecological benefit assessment of Avicennia marina as a specific tree species. Summary of the Invention

[0004] In view of this, this application provides a method and system for estimating the carbon storage of the Avicennia marina ecosystem, in order to solve the technical problem that existing technologies cannot achieve accurate and complete carbon sink accounting for this key tree species.

[0005] The first aspect of this application provides a method for estimating the carbon storage of the *Sonneratia apetala* ecosystem, the method comprising:

[0006] The spatial distribution of Avicennia marina in the target area was obtained, and patches of Avicennia marina with vegetation cover that met the preset vegetation cover conditions were selected from the spatial distribution of Avicennia marina and used as Avicennia marina single-superior communities. Construct a five-dimensional remote sensing feature set of lobeless Sassafras monomorphic communities across the target region; Based on the global five-dimensional remote sensing feature set and the pre-trained tree height inversion model, the global tree height spatial distribution is obtained. Based on the spatial distribution of tree height across the entire area, the spatial distribution of aboveground biomass across the entire area was calculated using a pre-established allometric growth model of tree height and diameter at breast height and an allometric growth equation of aboveground biomass. The carbon content coefficient was then used to convert the spatial distribution of aboveground carbon density across the entire area. Based on the spatial distribution of aboveground carbon density across the entire area and the root-to-shoot ratio and litter distribution coefficient determined through literature review, the spatial distribution of underground carbon density and the spatial distribution of litter carbon density across the entire area were calculated respectively. Using a pre-constructed soil organic carbon prediction model, based on the global five-dimensional remote sensing feature set and prior data on global mangrove soil organic carbon, the spatial distribution of soil organic carbon density in the entire region is obtained. The spatial distribution of carbon density above ground, below ground, litter, and soil organic carbon in the entire region is superimposed and fused at the pixel level to obtain the spatial distribution of total carbon storage in the Avicennia marina ecosystem.

[0007] Optionally, the method further includes: Ground quadrats were set up in the study area. The true value of tree height in the quadrats was obtained by UAV lidar, and the true values ​​of diameter at breast height, aboveground biomass, and soil organic carbon density were measured. Construct a five-dimensional remote sensing feature set specifically for lobeless Sedum aizoon mono-superior communities in the study area; A tree height inversion model was trained based on a dedicated five-dimensional remote sensing feature set of the study area and the true value of tree height. Based on the true values ​​of tree height and diameter at breast height in the study area, an allometric growth model for tree height and diameter at breast height specifically for Avicennia marina was constructed. Based on the true values ​​of tree height, diameter at breast height and aboveground biomass in the study area, an allometric growth equation for aboveground biomass specific to Avicennia marina was constructed. A soil organic carbon prediction model was trained based on a dedicated five-dimensional remote sensing feature set for the study area, prior data on global mangrove soil organic carbon, true values ​​of aboveground biomass, true values ​​of soil organic carbon density, and environmental factor data.

[0008] Optionally, obtaining the true value of the sample plot tree height using UAV lidar includes: For point cloud data from UAV lidar aerial surveys, a progressive triangulation filtering algorithm is used to separate ground points from vegetation points, generating digital elevation models and digital surface models respectively. The canopy height model is obtained by difference calculation, and the local maximum method is used to extract the true values ​​of individual tree positions and tree heights.

[0009] Optionally, the allometric growth model for tree height and diameter at breast height is as follows: ; in, , It is a constant. For the height of the tree, It refers to the diameter at breast height.

[0010] Optionally, the feature is that the allometric growth equation for aboveground biomass is: ; in, , These are the coefficients of the allometric growth equation. This refers to aboveground biomass.

[0011] Optionally, obtaining the lobeless Sorbus hymenoplastic spatial distribution of the target region includes: Acquire time-series satellite remote sensing images covering the target area; The spatial distribution of lobeless Sorbus hirsuta was obtained by using a pre-defined lobeless Sorbus hirsuta remote sensing identification algorithm to identify satellite time-series remote sensing images.

[0012] Optionally, the decision rule of the lobeless Sentinel-1VH remote sensing identification algorithm is set as follows: the polarization backscattering coefficient of Sentinel-1VH is ≥-14.390 and the shortwave infrared reflectance of Sentinel-2B11 is <0.084.

[0013] Optionally, the five-dimensional remote sensing feature set of the lobeless Sorbus hainanensis monoculture community in the target region includes: Acquire multi-source satellite remote sensing data and spaceborne lidar data for the target area; Based on the acquired multi-source satellite remote sensing data and spaceborne lidar data, a full-domain five-dimensional remote sensing feature set is constructed. The global five-dimensional remote sensing feature set includes: spectral dimension features, spatial dimension features, temporal dimension features, backscattering dimension features, and point cloud dimension features.

[0014] Optionally, the calculation of the spatial distribution of aboveground biomass based on the spatial distribution of tree height across the entire domain, using a pre-established tree height-diameter-thickness allometric growth model and aboveground biomass allometric growth equation, includes: Substituting the spatial distribution of tree height across the entire domain into the allometric growth model of tree height and diameter at breast height, we obtain the spatial distribution of diameter at breast height across the entire domain. Based on the spatial distribution of diameter at breast height (DBH) and the allometric growth equation of aboveground biomass, the spatial distribution of aboveground biomass across the entire area was calculated.

[0015] A second aspect of this application provides a carbon storage estimation system for the Avicennia marina ecosystem, the system comprising: The single-superior community delineation unit is used to obtain the spatial distribution of Avicennia marina in the target area, and to select patches of Avicennia marina whose vegetation cover meets the preset vegetation cover conditions from the spatial distribution of Avicennia marina, as Avicennia marina single-superior communities. Remote sensing feature construction unit, used to construct a five-dimensional remote sensing feature set of lobeless Sorbus hainanensis mono-superior communities across the target area; The global tree height inversion unit is used to obtain the global tree height spatial distribution based on the global five-dimensional remote sensing feature set and the pre-trained tree height inversion model. The aboveground biomass computing unit is used to calculate the spatial distribution of aboveground biomass based on the spatial distribution of tree height across the entire domain, using a pre-established tree height-diameter-thickness allometric growth model and aboveground biomass allometric growth equation, and converts it into the spatial distribution of aboveground carbon density across the entire domain using the carbon content coefficient. The underground biological computing unit is used to calculate the spatial distribution of underground carbon density and the spatial distribution of litter carbon density in the whole area based on the spatial distribution of aboveground carbon density and the root-to-shoot ratio and litter distribution coefficient determined by literature survey. The soil density calculation unit is used to obtain the spatial distribution of soil organic carbon density in the whole region by using a pre-built soil organic carbon prediction model based on the whole-domain five-dimensional remote sensing feature set and the prior data of global mangrove soil organic carbon. The total carbon storage output unit is used to perform pixel-level superposition and fusion of the spatial distribution of aboveground carbon density, underground carbon density, litter carbon density, and soil organic carbon density across the entire region to obtain the spatial distribution of total carbon storage in the Avicennia marina ecosystem.

[0016] This application selects patches of *Avicennia marina* with vegetation cover meeting preset conditions from the spatial distribution of *Avicennia marina* in the target area, designating them as single-superior *Avicennia marina* communities. By limiting these to single-superior communities, the estimation bias caused by mixed forests is avoided, improving the targeting and accuracy of carbon storage accounting. A comprehensive five-dimensional remote sensing feature set is constructed to fully characterize vegetation structure and growth status, providing robust input for subsequent inversion. Based on the comprehensive five-dimensional remote sensing feature set and a pre-trained tree height inversion model, the spatial distribution of tree height across the entire area is obtained, realizing the spatial inversion of tree height from sample point measurements to the entire area, meeting the needs of large-scale rapid monitoring. The spatial distribution of aboveground biomass across the entire area is calculated from the spatial distribution of tree height, and then converted into the spatial distribution of aboveground carbon density across the entire area using the carbon content coefficient. Distribution: Based on the spatial distribution of aboveground carbon density across the entire region and the root-to-shoot ratio and litter distribution coefficient determined through literature review, the spatial distribution of underground carbon density and litter carbon density across the entire region are calculated respectively, ensuring the integrity of carbon pool accounting. Using a pre-constructed soil organic carbon prediction model, based on the aforementioned five-dimensional remote sensing feature set of the entire region and prior data of global mangrove soil organic carbon, the spatial distribution of soil organic carbon density across the entire region is obtained, solving the problem of the difficulty in observing the underground carbon pool over a large area. Finally, through pixel-level fusion, a unified accounting of the spatial distribution of aboveground carbon density, underground carbon density, litter carbon density, and soil organic carbon density across the entire region is achieved, outputting a high-precision, spatially continuous total carbon storage distribution. This application breaks through the limitation of existing technologies that only focus on aboveground biomass. By simultaneously calculating aboveground, belowground, litter, and soil organic carbon, it fully covers the carbon pool of the mangrove ecosystem, allowing the carbon storage estimation results of *Sonneratia apetala* to truly reflect its carbon sequestration capacity. While ensuring the accuracy of the calculation, it significantly improves the efficiency of implementation and effectively solves the core problems of mixed tree species and incomplete carbon pool calculation in traditional mangrove carbon storage estimation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the training process of the carbon storage estimation model for the petalless Sorrelia ecosystem provided in the embodiments of this application; Figure 2 This is a flowchart illustrating the carbon storage estimation method for the petalless Kapok ecosystem provided in this application embodiment; Figure 3 This is a schematic diagram of the module of the carbon storage estimation system for the petalless Sorrelia ecosystem provided in this application embodiment. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0020] Sonneratia apettata, belonging to the genus Sonneratia in the family Sonneraceae, is a fast-growing, tall mangrove tree with an extremely strong carbon sequestration capacity. Compared to native mangroves such as Avicennia marina, Kandelia candel, and Avicennia marina, Sonneratia apettata exhibits significant advantages in growth rate, carbon sequestration capacity, and habitat adaptability. However, its significant differences in biomass and carbon allocation compared to other tree species make it impossible to co-calculate with other tree species in mangrove carbon storage estimation, which has become one of the core challenges of traditional estimation methods.

[0021] To address the core issues in existing mangrove carbon storage estimation, such as mixed tree species, incomplete carbon pools, and difficulty in direct remote sensing inversion of underground carbon pools, this application proposes a carbon storage estimation method for the Avicennia marina ecosystem, aiming to build a high-precision, replicable, and scalable carbon storage estimation technology system.

[0022] This application uses the high-precision distribution of Avicennia marina monotypic communities as the basis for calculation, and solves the problem of insufficient estimation accuracy caused by mixed tree species in traditional methods from the source. By constructing a global five-dimensional remote sensing feature set, integrating multi-source remote sensing data from air, space, and ground, and combining machine learning algorithms, while introducing global soil carbon prior knowledge, it achieves simultaneous estimation of the entire carbon pool of aboveground biomass, underground biomass, and soil organic carbon in Avicennia marina monotypic communities.

[0023] like Figure 1 The diagram shown illustrates the training process of the carbon storage estimation model for the lobeless Sorbus aegyptiacus ecosystem provided in this application embodiment, including the following steps: 101. Ground quadrats were set up in the study area. The true value of tree height in the quadrats was obtained by UAV lidar, and the true values ​​of diameter at breast height, aboveground biomass, and soil organic carbon density were measured.

[0024] Typical *Sonneratia apetala* forest areas, including Qi'ao Island in Guangdong, Maowei Sea in Guangxi, and Zhangjiang Estuary in Fujian, were selected as study areas. Typical ground plots were established within these areas, and UAV-based lidar aerial surveys were conducted at a flight altitude of 100m and a point cloud density of no less than 20 points / m² to acquire sub-meter-level canopy 3D structure data. For the same set of ground plots, the true values ​​of diameter at breast height (DBH), aboveground biomass, and soil organic carbon density were simultaneously measured to eliminate statistical errors caused by spatial differences between plots and ensure strict spatial matching of aboveground and belowground carbon pool data, providing a precise sample basis for subsequent model construction.

[0025] In one optional implementation, soil samples are collected at preset depths (0–20cm, 20–40cm, 40–60cm, 60–100cm), the organic carbon content and bulk density of each layer are measured, the soil organic carbon density is calculated layer by layer and accumulated to obtain the true value of soil organic carbon density.

[0026] In an optional implementation, the step of obtaining the true value of tree height in a sample plot using UAV lidar includes: using a progressive triangulation filtering algorithm to separate ground points and vegetation points from aerial survey point cloud data, generating a digital elevation model (DEM) and a digital surface model (DSM) respectively, calculating the difference between the two to obtain a canopy height model (CHM), and using the local maximum method to extract the location and true value of individual trees.

[0027] The raw point cloud data acquired by UAV lidar aerial surveying was preprocessed to remove noise, redundant, and outlier points, ensuring the purity and reliability of the data. Subsequently, a progressive triangulation filtering algorithm was used to separate ground points from vegetation points in the preprocessed point cloud data: by iteratively constructing a triangulation network, low-lying ground points and high-lying canopy points were gradually distinguished, ultimately resulting in two types of data: ground point cloud and vegetation point cloud. Based on the separated ground point cloud, a digital elevation model (DEM) was constructed using an interpolation algorithm to represent the true surface elevation of the study area. Based on the fused data of vegetation and ground point clouds, a digital surface model (DSM) was constructed to represent the surface elevation including the vegetation canopy height. The pixel-by-pixel difference between the DSM and DEM was calculated to obtain the canopy height model (CHM), where the value of each pixel in the CHM represents the relative height of the vegetation at that location. Finally, the CHM data is traversed through a preset window to identify the elevation maximum point in each window. This maximum point is determined as the top position of a single tree, and the corresponding CHM value is extracted as the true value of the tree height. This completes the extraction of the true tree height of all single trees in the quadrat.

[0028] 102. Construct a five-dimensional remote sensing feature set specifically for the lobeless Sorbus hainanensis monoculture community in the study area.

[0029] Multi-source satellite remote sensing data and spaceborne lidar data covering the study area were acquired. Based on the acquired multi-source satellite remote sensing data and spaceborne lidar data, a dedicated five-dimensional remote sensing feature set was constructed. This dedicated five-dimensional remote sensing feature set includes: spectral dimension features, spatial dimension features, temporal dimension features, backscattering dimension features, and point cloud dimension features. Spectral dimension features include multispectral band reflectance and vegetation indices; spatial dimension features include vegetation texture features; temporal dimension features include time-series phenological features; backscattering dimension features include radar polarimetric scattering features; and point cloud dimension features include spaceborne lidar canopy height features, providing high-dimensional and highly interpretable input variables for subsequent model training.

[0030] 103. Based on the dedicated five-dimensional remote sensing feature set of the study area and the true value of tree height, a tree height inversion model was trained.

[0031] Using a dedicated five-dimensional remote sensing feature set of the sample plot locations in the study area as the independent variable and the true tree height measured by UAV lidar at the same location as the dependent variable, sample pairs were constructed. Machine learning algorithms such as random forest and gradient boosting machine were used to train the regression model. The model parameters were optimized and overfitting was controlled through five-fold cross-validation to obtain a high-precision tree height inversion model, which was used for subsequent large-scale tree height spatial distribution inversion in the entire region.

[0032] 104. Based on the true values ​​of tree height and diameter at breast height in the study area, an allometric growth model for tree height and diameter at breast height specifically for *Sorbus apetala* was constructed.

[0033] The tree height ground value extracted by UAV lidar is spatially matched with the ground measured diameter at breast height (DBH) ground value to construct a tree height-DBH paired dataset. Based on the growth characteristics of Avicennia marina, a power function model is used for fitting to obtain an allometric growth model of tree height and DBH specific to Avicennia marina, enabling rapid calculation of DBH spatial distribution from remote sensing tree height.

[0034] In one optional implementation, the allometric growth model for tree height and diameter at breast height is: ;in, , It is a constant. For the height of the tree, It refers to the diameter at breast height.

[0035] 105. Based on the true values ​​of tree height, diameter at breast height, and aboveground biomass in the study area, an allometric growth equation for aboveground biomass specific to Avicennia marina was constructed.

[0036] Using the true values ​​of tree height and diameter at breast height as independent variables and the true value of measured aboveground biomass as the dependent variable, a unique allometric growth relationship for Avicennia marina was constructed to ensure that the biomass estimation is consistent with the rapid and tall growth characteristics of this tree species and to improve the accuracy of aboveground carbon pool accounting.

[0037] In one optional implementation, the allometric growth equation for aboveground biomass is: ;in, , These are the coefficients of the allometric growth equation. This refers to aboveground biomass.

[0038] 106. Based on the dedicated five-dimensional remote sensing feature set of the study area, prior data of global mangrove soil organic carbon, true values ​​of aboveground biomass, true values ​​of soil organic carbon density, and environmental factor data, a soil organic carbon prediction model was trained.

[0039] Prior data on global mangrove soil organic carbon were resampled to ensure spatial resolution was consistent with a dedicated five-dimensional remote sensing feature set. The resampled prior data, the dedicated five-dimensional remote sensing feature set, ground biomass data, and environmental factor data such as topography, salinity, and tidal level were spatially registered at the pixel level to form a joint input feature set for model training. Ground-measured soil organic carbon density from the same ground sample plot was used as the dependent variable and fitting target for model training. Random forest and gradient boosting tree machine learning algorithms were employed for model training. Cross-validation was used to optimize model parameters and reduce the risk of overfitting. Ultimately, a soil organic carbon prediction model suitable for the *Avicennia marina* ecosystem was obtained, enabling simultaneous estimation of soil organic carbon density from remote sensing features and prior data, thus solving the technical challenge of large-scale direct observation of underground carbon pools.

[0040] It should be noted that step 101 is executed first to obtain all true value samples; steps 102 to 106 have no strict execution order and can be executed in parallel, partially in parallel, sequentially, or individually in any order, making the model training process flexible and adaptable.

[0041] like Figure 2 The diagram shown is a flowchart illustrating a method for estimating carbon storage in a lobeless mangrove ecosystem according to an embodiment of this application, including the following steps: 201. Obtain the spatial distribution of Avicennia marina in the target area, and select patches of Avicennia marina whose vegetation cover meets the preset vegetation cover conditions from the spatial distribution of Avicennia marina as Avicennia marina single-superior communities.

[0042] Step 201 is the step of defining the accounting unit. Its purpose is to spatially locate the distribution area of ​​pure Avicennia marina, avoid the estimation bias caused by mixed forests, and ensure the specificity and accuracy of subsequent carbon storage calculations.

[0043] First, the spatial distribution of *Avicennia marina* within the target area is obtained using remote sensing technology. Then, patches are selected based on preset vegetation cover conditions. Patches meeting the coverage criteria are identified as superior *Avicennia marina* communities and used as the dedicated accounting units for carbon storage estimation in this application. The target area refers to the area where carbon storage estimation is actually applied and carried out.

[0044] In an optional implementation, obtaining the lobeless Sorbus hainanensis spatial distribution of the target region includes: Acquire time-series satellite remote sensing images covering the target area; The spatial distribution of lobeless Sorbus hirsuta was obtained by using a pre-defined lobeless Sorbus hirsuta remote sensing identification algorithm to identify satellite time-series remote sensing images.

[0045] Satellite time-series remote sensing images can reflect the spectral, structural, and phenological characteristics of Avicennia marina in different seasons and growth stages, providing multi-dimensional and robust data support for accurate tree species identification.

[0046] A pre-defined remote sensing identification algorithm for Avicennia marina was adopted. This algorithm is specifically adapted to the identification method of Avicennia marina species characteristics. It can accurately extract relevant feature information of Avicennia marina from satellite time-series remote sensing images, effectively distinguish Avicennia marina from other mangrove species and surrounding vegetation types. Through professional processing and analysis of satellite time-series remote sensing images, the spatial distribution of Avicennia marina is finally generated, which can clearly reflect the specific spatial distribution location, range and patch characteristics of Avicennia marina in the target area. This lays a precise spatial data foundation for subsequent screening of superior Avicennia marina communities.

[0047] In some embodiments, the satellite time-series remote sensing imagery comes from one or more of Sentinel-1 and Sentinel-2. Both Sentinel-1 and Sentinel-2 are high-quality remote sensing data sources, suitable for the technical requirements of remote sensing identification of *Avicennia marina*. Using one or a combination of them can provide multi-dimensional, high-quality time-series remote sensing data support for accurate spatial identification of *Avicennia marina*. Sentinel-1 is a synthetic aperture radar satellite image, possessing all-weather, all-day imaging capabilities, unaffected by common weather conditions such as rain and high clouds in mangrove distribution areas, or by day and night cycles. It can stably acquire time-series imagery data and capture the geometric and structural features of *Avicennia marina*, such as canopy structure, texture, and topography, compensating for the imaging limitations of optical imagery. Sentinel-2 is a multispectral optical satellite image, possessing high spatial resolution and rich spectral bands. It can accurately capture the spectral differences in leaf reflectance and photosynthetic characteristics between *Avicennia marina* and other mangrove species such as *Kandelia candel* and *Avicennia marina*, which are the core characteristic basis for achieving spectral identification of tree species.

[0048] In a preferred embodiment, the decision rule of the lobeless Sentinel-1VH remote sensing identification algorithm is set as follows: the polarization backscattering coefficient of Sentinel-1VH is ≥-14.390 and the shortwave infrared reflectance of Sentinel-2B11 is <0.084.

[0049] In some embodiments, patches with Avicennia marina vegetation cover ≥85% are selected and defined as Avicennia marina monotypic communities, which are used as the accounting unit in this application. The clearly defined monotypic community accounting unit makes the subsequent modeling and inversion processes of aboveground and belowground carbon pools more targeted, allowing the Avicennia marina-specific model to play its full role and ensuring the scientific validity of the carbon pool estimation data. The carbon storage data output using monotypic communities as the accounting unit provides refined spatial unit data support for the management of Avicennia marina-specific blue carbon projects and carbon sink value assessment, avoiding the problem that averaged data from mixed patches cannot meet the needs of differentiated ecological management.

[0050] As a tall, fast-growing tree species, *Avicennia marina* (also known as *Avicennia apetalum*) exhibits significantly different biomass and carbon allocation ratios per unit area compared to native mangrove species such as *Kandelia candel* and *Avicennia marina*. If the calculated carbon storage patch contains a large number of other tree species, it will directly cause the carbon storage estimate to deviate from the true carbon sequestration level of *Avicennia apetalum*. A vegetation cover of ≥85% is a commonly used scientific threshold in ecology for defining a single dominant community. This threshold ensures that *Avicennia apetalum* is the absolutely dominant species in the target patch, with other tree species only appearing sporadically and having a negligible impact on the overall carbon storage.

[0051] 202. Construct a five-dimensional remote sensing feature set of lobeless Sedum monocultures in the target region.

[0052] Step 202 involves constructing a global five-dimensional remote sensing feature set, which provides high-dimensional, high-precision input variables for subsequent tree height inversion and underground carbon inversion.

[0053] First, multi-source satellite remote sensing data and spaceborne lidar data for the target area are acquired. Multi-source satellite remote sensing data can include Sentinel-2 multispectral imagery and Sentinel-1 SAR imagery, while spaceborne lidar data can include GEDI spaceborne lidar data. Next, based on the acquired multi-source satellite remote sensing data and spaceborne lidar data, a global five-dimensional remote sensing feature set is constructed. This global five-dimensional remote sensing feature set refers to a collection of features across five dimensions, specifically: spectral dimension features (extracting reflectance and vegetation indices such as NDVI, EVI, and red-edge index from Sentinel-2 bands), spatial dimension features (extracting texture features based on the gray-level co-occurrence matrix), temporal dimension features (constructing NDVI / EVI time series using growing season time-series imagery and extracting phenological parameters), backscattering dimension features (extracting VV and VH polarization backscattering coefficients from Sentinel-1 and calculating the VH / VV ratio), and point cloud dimension features (canopy height sampling data acquired by the GEDI spaceborne lidar).

[0054] The aforementioned multi-source data characterize the structure and spectral properties of the lobeless Sorbus canopy from different dimensions, providing a rich data foundation for tree height inversion modeling. Features from each dimension are optimized, and features with excessively high correlation or poor robustness are removed, ultimately generating a global five-dimensional remote sensing feature set for subsequent modeling.

[0055] In some embodiments, the GEDI spaceborne lidar is used to extract large-scale canopy height sampling points, providing prior information for the tree height inversion model. The prior data on global mangrove soil organic carbon is the 30-meter resolution global mangrove soil organic carbon density data published by Sanderman et al., used to provide background constraints for soil carbon estimation.

[0056] In practical applications, satellite data sources can be flexibly selected according to the actual conditions of the target area, such as climate and terrain. Combining Sentinel-1 and Sentinel-2 can also achieve complementarity between the structural features of radar and the spectral features of optics, enriching the remote sensing identification feature dimensions of lobeless Sangha and improving the computational accuracy of the lobeless Sangha remote sensing identification algorithm.

[0057] 203. Based on the global five-dimensional remote sensing feature set and the pre-trained tree height inversion model, the global tree height spatial distribution is obtained.

[0058] Step 203 is the inversion of the spatial distribution of tree height across the entire region, realizing the expansion of tree height from "sample point measurement" to "continuous mapping of the entire region".

[0059] The five-dimensional remote sensing feature set obtained in step 202 is input pixel by pixel into the pre-trained tree height inversion model. The model outputs the predicted tree height value pixel by pixel based on the nonlinear mapping relationship between remote sensing features and tree height, and finally forms the spatial distribution of tree height covering the entire lobeless Sorbus hainanensis monoculture community.

[0060] The above steps extend the true values ​​of finite quadrats to the entire region, solving the problem of difficulty in measuring the height of each tree in the field over a large area of ​​Avicennia marina.

[0061] 204. Based on the spatial distribution of tree height across the entire domain, the spatial distribution of aboveground biomass across the entire domain was calculated using a pre-established allometric growth model of tree height and diameter at breast height and an allometric growth equation of aboveground biomass. The carbon content coefficient was then used to convert the spatial distribution of aboveground carbon density across the entire domain.

[0062] Step 204 is to estimate the spatial distribution of aboveground biomass across the entire area, and to realize the spatial representation of the aboveground carbon pool through the tree height inversion results.

[0063] In an optional implementation, the calculation of the spatial distribution of aboveground biomass based on the spatial distribution of tree height across the entire domain, using a pre-established tree height-diameter-at-breast-width allometric growth model and aboveground biomass allometric growth equation, includes: Substituting the spatial distribution of tree height across the entire domain into the allometric growth model of tree height and diameter at breast height, we obtain the spatial distribution of diameter at breast height across the entire domain. Based on the spatial distribution of diameter at breast height (DBH) and the allometric growth equation of aboveground biomass, the spatial distribution of aboveground biomass across the entire area was calculated.

[0064] Based on the spatial distribution of tree height obtained from step 203, the tree height-to-diameter-at-breast-length (DBH) allometric growth model is substituted into the spatial distribution of DBH, calculated pixel by pixel. Then, based on the calculated DBH spatial distribution, the spatial distribution of DBH is substituted into the aboveground biomass allometric growth equation specific to *Sonneratia apetala* to calculate the spatial distribution of aboveground biomass. Finally, the product of the aboveground biomass spatial distribution and the carbon content coefficient (e.g., 0.5) is used to convert it into the spatial distribution of aboveground carbon density.

[0065] The above steps enable continuous spatial extrapolation from remotely sensed tree height to diameter at breast height (DBH) to aboveground biomass, and finally calculate the spatial distribution of aboveground carbon density across the entire region.

[0066] 205. Based on the spatial distribution of aboveground carbon density across the entire region and the root-to-shoot ratio and litter distribution coefficient determined through literature review, the spatial distribution of underground carbon density and litter carbon density across the entire region were calculated respectively.

[0067] Step 205 involves estimating the spatial distribution of underground carbon density and the spatial distribution of litter carbon density across the entire region. The spatial distribution of underground carbon density is calculated based on the product of the spatial distribution of aboveground carbon density and the root-to-shoot ratio determined through literature review; the spatial distribution of litter carbon density is calculated based on the product of the spatial distribution of aboveground carbon density and the litter distribution coefficient determined through literature review.

[0068] The root-to-shoot ratio coefficient is the aboveground-to-underground biomass allocation coefficient, and the litter allocation coefficient is the aboveground-to-litter allocation coefficient. Both the root-to-shoot ratio coefficient and the litter allocation coefficient are determined through literature review, for example, based on IPCC guidelines, mangrove blue carbon literature, or field measurement data; they are empirical coefficients or measured calibration coefficients. Using the allocation coefficient method, the spatialization of underground biomass and litter carbon pools can be quickly achieved, ensuring the completeness of the total carbon pool accounting.

[0069] The root-to-shoot ratio and litter distribution coefficient can be the same, for example, both 0.5, or they can be different.

[0070] 206. Using a pre-constructed soil organic carbon prediction model, based on the global five-dimensional remote sensing feature set and prior data of global mangrove soil organic carbon, the spatial distribution of soil organic carbon density in the entire region is obtained.

[0071] Step 206 is the soil organic carbon pool inversion step, used to achieve spatial mapping of the largest carbon pool in the mangrove ecosystem.

[0072] The trained soil organic carbon prediction model is applied to the entire region. Specifically, the five-dimensional remote sensing feature set of the entire region and the prior data of soil organic carbon in global mangrove forests are input pixel by pixel into the dedicated soil organic carbon prediction model. The model estimates the soil organic carbon density of each pixel based on the input features corresponding to each pixel, thereby obtaining the spatial distribution of soil organic carbon density in the entire region.

[0073] Prior data on global mangrove soil organic carbon can be obtained by using the 30-meter resolution global mangrove soil organic carbon dataset published by Sanderman et al. After resampling, it can be matched with remote sensing features to provide background constraints for soil carbon inversion and improve inversion accuracy and spatial consistency.

[0074] 207. The spatial distribution of aboveground carbon density, underground carbon density, litter carbon density, and soil organic carbon density of the entire region were superimposed and fused at the pixel level to obtain the spatial distribution of total carbon storage of the Avicennia marina ecosystem.

[0075] Step 207 outputs the carbon storage of the entire ecosystem. The spatial distributions of aboveground carbon density, belowground carbon density, litter carbon density, and soil organic carbon density across the entire region are overlaid and fused at the pixel level. The total carbon storage is obtained by summing the data pixel by pixel. This generates and outputs the spatial distribution of the total carbon storage of the *Sonneratia apetala* ecosystem, which can intuitively and accurately reflect the spatial distribution of carbon storage in the entire ecosystem of the single dominant *Sonneratia apetala* community. This completes the entire carbon storage estimation process.

[0076] Total carbon storage = Total surface carbon density + Total underground carbon density + Total litter carbon density + Total soil organic carbon density.

[0077] In some embodiments, the formula for calculating total carbon storage is as follows:

[0078] in, For total carbon reserves, The aboveground carbon density of the entire area is calculated as (aboveground biomass × 0.5). For the overall underground carbon density, For the total soil organic carbon density, The carbon density of litter across the entire area.

[0079] The above steps enable unified accounting of the entire carbon pool, including above-ground, underground, vegetation litter, and soil carbon. The output results can be directly used for blue carbon monitoring, carbon sink assessment, ecological management, and carbon trading accounting.

[0080] Compared with the prior art, this application has the following significant advantages: More precise accounting targets: This application focuses on single-superior communities of Avicennia marinae, and by pre-screening pure forest patches that meet preset vegetation cover conditions, it completely avoids the problem of mixed accounting with other mangrove species, ensuring the accuracy and relevance of carbon storage accounting results from the source. In contrast, although existing technologies can identify Avicennia marinae, their application scenarios may still cover mixed forest areas, which are prone to accounting deviations due to mixed tree species, making it difficult to guarantee accuracy. More complete accounting content: This application systematically integrates the full carbon pool accounting of aboveground biomass, belowground biomass, litter, and soil organic carbon, fully complying with the relevant national blue carbon accounting standards and achieving comprehensive coverage of the carbon storage of the Avicennia marina ecosystem. In contrast, most existing technologies are limited to the estimation of aboveground biomass and do not involve the belowground carbon pool and soil organic carbon, resulting in significant gaps in the accounting results and failing to fully reflect the carbon sequestration capacity of Avicennia marina. The estimation method is more advanced: This application innovatively constructs a full-domain five-dimensional remote sensing feature set covering spectrum, spatial texture, temporal phenology, radar backscattering, and spaceborne lidar point cloud. Combined with machine learning algorithms, a nonlinear regression model is constructed, which greatly improves the accuracy of the estimation. At the same time, in the estimation of soil organic carbon, the prior data of global mangrove soil organic carbon are innovatively integrated, which effectively solves the technical pain point that it is difficult to directly remotely invert the underground carbon pool, and further improves the intelligence level and reliability of the entire estimation system. Greater application value: The spatial distribution data of total carbon storage of the Avicennia marina ecosystem produced by this application can be directly applied to the monitoring, reporting and verification (MRV), comprehensive ecological benefit assessment and differentiated operation and management of Avicennia marina special carbon sink projects. It is in line with the current development needs of the blue carbon industry, has stronger policy adaptability and market orientation, and has a wider range of application scenarios and more prominent practical value.

[0081] Figure 3 This is a functional block diagram of the carbon storage estimation system for the petalless Sorrelia ecosystem provided in the embodiments of this application.

[0082] The carbon storage estimation system 30 for the *Sonneratia apetala* ecosystem runs on a computer device. The system 30 may include multiple functional modules composed of program code segments. The program code of each segment in the system 30 may be stored in the memory of the computer device and executed by at least one processor to perform (see details). Figure 2 (Description) Function for estimating carbon storage in the genus *Sonneratia* ecosystem.

[0083] In this embodiment, the carbon storage estimation system 30 for the *Sonneratia apetala* ecosystem can be divided into multiple functional modules based on its functions. These functional modules may include: a single-superior community delineation unit 301, a remote sensing feature construction unit 302, a global tree height inversion unit 303, a surface biological calculation unit 304, a subsurface biological calculation unit 305, a soil density calculation unit 306, and a total carbon storage output unit 307. The term "module" in this application refers to a series of computer-readable instruction segments that can be executed by at least one processor and perform a fixed function, stored in memory. In this embodiment, the functions of each module will be detailed in subsequent embodiments.

[0084] The single-superior community delineation unit 301 is used to obtain the spatial distribution of Avicennia marina in the target area, and to screen out patches of Avicennia marina whose vegetation cover meets the preset vegetation cover conditions from the spatial distribution of Avicennia marina, as Avicennia marina single-superior communities. Remote sensing feature construction unit 302 is used to construct a five-dimensional remote sensing feature set of lobeless Sangika lichen monogenean communities across the target area. The global tree height inversion unit 303 is used to obtain the global tree height spatial distribution based on the global five-dimensional remote sensing feature set and the pre-trained tree height inversion model. The aboveground biomass computing unit 304 is used to calculate the spatial distribution of aboveground biomass based on the spatial distribution of tree height, using a pre-established tree height-diameter-thickness allometric growth model and aboveground biomass allometric growth equation, and convert it into the spatial distribution of aboveground carbon density based on the carbon content coefficient. The underground biological computing unit 305 is used to calculate the spatial distribution of underground carbon density and the spatial distribution of litter carbon density in the whole area based on the spatial distribution of aboveground carbon density in the whole area and the root-shoot ratio coefficient and litter distribution coefficient determined by literature survey. Soil density calculation unit 306 is used to obtain the spatial distribution of soil organic carbon density in the whole region by using a pre-constructed soil organic carbon prediction model based on the whole-domain five-dimensional remote sensing feature set and the prior data of global mangrove soil organic carbon. The total carbon storage output unit 307 is used to perform pixel-level superposition and fusion of the spatial distribution of aboveground carbon density, the spatial distribution of underground carbon density, the spatial distribution of litter carbon density, and the spatial distribution of soil organic carbon density in the entire region to obtain the spatial distribution of total carbon storage of the Avicennia marina ecosystem.

[0085] It should be understood that the various variations and specific embodiments of the carbon storage estimation method for the Avicennia marina ecosystem provided in the above embodiments are also applicable to the carbon storage estimation device for the Avicennia marina ecosystem in this embodiment. Through the detailed description of the carbon storage estimation method for the Avicennia marina ecosystem described above, those skilled in the art can clearly understand the implementation process of the carbon storage estimation device for the Avicennia marina ecosystem in this embodiment. For the sake of brevity, it will not be described in detail here.

[0086] This application provides a computer program product comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform a method for estimating the carbon storage of a *Sonneratia apetala* ecosystem. Applying a carbon storage estimation method for a *Sonneratia apetala* ecosystem to a computer program product facilitates execution.

[0087] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the carbon storage estimation method for a petioleless Sangha ecosystem as described above.

[0088] The computer storage medium of this application can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. This application applies a carbon storage estimation method for a petioleless Sangha ecosystem to a computer-readable storage medium storing a computer program. When executed by a processor, this program implements the steps of the clothing simulation method provided in this application, which is simple, fast, easy to store, and not easily lost.

[0089] Those skilled in the art will understand that the modules described above can be implemented using general-purpose computing systems. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Optionally, they can be implemented using computer-executable program code, allowing them to be stored in a storage system for execution by the computing system. Alternatively, they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0090] Note that the above description is merely a preferred embodiment and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments. Many other equivalent embodiments may be included without departing from the concept of this application, and the scope of this application is determined by the scope of the appended claims.

[0091] The above disclosures are only a few specific implementation scenarios of this application. However, this application is not limited to these. Any variations that can be conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. A method for estimating carbon storage in a genus *Avicennia marina* ecosystem, characterized in that, The method includes: The spatial distribution of Avicennia marina in the target area was obtained, and patches of Avicennia marina with vegetation cover that met the preset vegetation cover conditions were selected from the spatial distribution of Avicennia marina and used as Avicennia marina single-superior communities. Construct a five-dimensional remote sensing feature set of lobeless Sassafras monomorphic communities across the target region; Based on the global five-dimensional remote sensing feature set and the pre-trained tree height inversion model, the global tree height spatial distribution is obtained. Based on the spatial distribution of tree height across the entire domain, the spatial distribution of aboveground biomass across the entire domain was calculated using a pre-established allometric growth model of tree height and diameter at breast height and an allometric growth equation of aboveground biomass. The carbon content coefficient was then used to convert the spatial distribution of aboveground carbon density across the entire domain. Based on the spatial distribution of aboveground carbon density across the entire area and the root-to-shoot ratio and litter distribution coefficient determined through literature review, the spatial distribution of underground carbon density and the spatial distribution of litter carbon density across the entire area were calculated respectively. Using a pre-constructed soil organic carbon prediction model, based on the global five-dimensional remote sensing feature set and prior data on global mangrove soil organic carbon, the spatial distribution of soil organic carbon density in the entire region is obtained. The spatial distribution of carbon density above ground, below ground, litter, and soil organic carbon in the entire region is superimposed and fused at the pixel level to obtain the spatial distribution of total carbon storage in the Sangharama apetala ecosystem.

2. The method for estimating carbon storage in the *Avicennia marina* ecosystem according to claim 1, characterized in that, The method further includes: Ground quadrats were set up in the study area. The true value of tree height in the quadrats was obtained by UAV lidar, and the true values ​​of diameter at breast height, aboveground biomass, and soil organic carbon density were measured. Construct a five-dimensional remote sensing feature set specifically for lobeless Sassafras monotypic communities in the study area; A tree height inversion model was trained based on a dedicated five-dimensional remote sensing feature set of the study area and the true value of tree height. Based on the true values ​​of tree height and diameter at breast height in the study area, an allometric growth model for tree height and diameter at breast height specifically for Avicennia marina was constructed. Based on the true values ​​of tree height, diameter at breast height and aboveground biomass in the study area, an allometric growth equation for aboveground biomass specific to Avicennia marina was constructed. A soil organic carbon prediction model was trained based on a dedicated five-dimensional remote sensing feature set for the study area, prior data on global mangrove soil organic carbon, true values ​​of aboveground biomass, true values ​​of soil organic carbon density, and environmental factor data.

3. The method for estimating carbon storage in the *Avicennia marina* ecosystem according to claim 2, characterized in that, The acquisition of the true value of the sample plot tree height via UAV lidar includes: For point cloud data from UAV lidar aerial surveys, a progressive triangulation filtering algorithm is used to separate ground points from vegetation points, generating digital elevation models and digital surface models respectively. The canopy height model is obtained by difference calculation, and the local maximum method is used to extract the true values ​​of individual tree positions and tree heights.

4. The method for estimating carbon storage in the *Avicennia marina* ecosystem according to claim 2, characterized in that, The allometric growth model for tree height and diameter at breast height is as follows: ; in, , It is a constant. For the height of the tree, It refers to the diameter at breast height.

5. The carbon storage estimation method for the Aeonium apetalis ecosystem according to claim 4, characterized in that, The equation for allometric growth of aboveground biomass is: ; in, , These are the coefficients of the allometric growth equation. This refers to aboveground biomass.

6. The method for estimating carbon storage in the *Avicennia marina* ecosystem according to claim 1, characterized in that, The acquisition of the lobeless Sanghae spatial distribution in the target region includes: Acquire time-series satellite remote sensing images covering the target area; The spatial distribution of lobeless Sorbus hirsutus was obtained by using a pre-defined lobeless Sorbus hirsutus remote sensing identification algorithm to identify satellite time-series remote sensing images.

7. The method for estimating carbon storage in the *Avicennia marina* ecosystem according to claim 1, characterized in that, The decision rules for the lobeless Sentinel-1VH remote sensing identification algorithm are set as follows: the polarization backscattering coefficient of Sentinel-1VH is ≥-14.390 and the shortwave infrared reflectance of Sentinel-2B11 is <0.

084.

8. The method for estimating carbon storage in the *Sonneratia apetala* ecosystem according to claim 1, characterized in that, The five-dimensional remote sensing feature set of the lobeless Sorbus hainanensis mono-superior community in the target region includes: Acquire multi-source satellite remote sensing data and spaceborne lidar data for the target area; Based on the acquired multi-source satellite remote sensing data and spaceborne lidar data, a full-domain five-dimensional remote sensing feature set is constructed. The global five-dimensional remote sensing feature set includes: spectral dimension features, spatial dimension features, temporal dimension features, backscattering dimension features, and point cloud dimension features.

9. The method for estimating carbon storage in the *Avicennia marina* ecosystem according to claim 1, characterized in that, The spatial distribution of aboveground biomass, calculated based on the spatial distribution of tree height across the entire area, using a pre-established allometric growth model of tree height and diameter at breast height (DBH) and an allometric growth equation of aboveground biomass, includes: Substituting the spatial distribution of tree height across the entire domain into the allometric growth model of tree height and diameter at breast height, we obtain the spatial distribution of diameter at breast height across the entire domain. Based on the spatial distribution of diameter at breast height (DBH) and the allometric growth equation of aboveground biomass, the spatial distribution of aboveground biomass across the entire area was calculated.

10. A carbon storage estimation system for the Avicennia marina ecosystem, characterized in that, The system includes: The single-superior community delineation unit is used to obtain the spatial distribution of Avicennia marina in the target area, and to select patches of Avicennia marina whose vegetation cover meets the preset vegetation cover conditions from the spatial distribution of Avicennia marina, as Avicennia marina single-superior communities. Remote sensing feature construction unit, used to construct a five-dimensional remote sensing feature set of lobeless Sorbus hainanensis mono-superior communities across the target area; The global tree height inversion unit is used to obtain the global tree height spatial distribution based on the global five-dimensional remote sensing feature set and the pre-trained tree height inversion model. The aboveground biomass computing unit is used to calculate the spatial distribution of aboveground biomass based on the spatial distribution of tree height across the entire domain, using a pre-established tree height-diameter-thickness allometric growth model and aboveground biomass allometric growth equation, and converts it into the spatial distribution of aboveground carbon density across the entire domain using the carbon content coefficient. The underground biological computing unit is used to calculate the spatial distribution of underground carbon density and the spatial distribution of litter carbon density in the whole area based on the spatial distribution of aboveground carbon density and the root-to-shoot ratio and litter distribution coefficient determined by literature survey. The soil density calculation unit is used to obtain the spatial distribution of soil organic carbon density in the whole region by using a pre-built soil organic carbon prediction model based on the whole-domain five-dimensional remote sensing feature set and the prior data of global mangrove soil organic carbon. The total carbon storage output unit is used to perform pixel-level superposition and fusion of the spatial distribution of aboveground carbon density, underground carbon density, litter carbon density, and soil organic carbon density across the entire region to obtain the spatial distribution of total carbon storage in the Avicennia marina ecosystem.

Images