A method and system for evaluating carbon sink benefits of wetlands based on remote sensing data

Abstract

The application discloses a kind of wetland carbon sink benefit evaluation method and system based on remote sensing data, it is related to carbon emission evaluation technical field.A kind of wetland carbon sink benefit evaluation system based on remote sensing data, including have: carbon sink evaluation zoning module and carbon sink benefit evaluation module.The application is realized to the high-resolution resolution of wetland spatial heterogeneity by multimodal remote sensing data fusion and fine grid division, different types of carbon sink area and its distribution boundary can be accurately identified, improve the scientificity and particle of carbon sink zoning;Dynamic index feature layer is constructed by remote sensing time series, combined with hydrological data to construct hydrological impulse response factor layer, and it is clustered and divided, realizes the dynamic identification and pixel level benefit calculation of dynamic carbon sink area.

Description

Technical Field

[0001] This invention relates to the field of carbon emission assessment technology, and in particular to a method and system for assessing the carbon sequestration benefits of wetlands based on remote sensing data. Background Technology

[0002] In existing technologies, wetland carbon sequestration benefit assessment methods mostly rely on single-modal remote sensing data or traditional statistical models, which are insufficient to comprehensively reflect the diversity characteristics of wetlands and the dynamic process of carbon sequestration. Furthermore, the accuracy of carbon sequestration benefit assessment is insufficient for special areas such as seasonal wetlands and areas with significant hydrological changes. At the same time, the lack of a unified classification and integrated assessment mechanism leads to fragmented and poorly adaptable regional carbon sequestration benefit results.

[0003] To address the aforementioned issues, a wetland carbon sequestration benefit assessment method based on the fusion of multimodal remote sensing data and hydrological information is proposed. This method conducts targeted carbon sequestration benefit assessments for wetland areas with ambiguous boundaries, achieving accurate assessments of carbon sequestration benefits across multiple regions, factors, and scales, thereby enhancing the comprehensiveness and dynamic adaptability of carbon sequestration benefit assessments. Summary of the Invention

[0004] The present invention aims to provide a method and system for assessing the carbon sequestration benefits of wetlands based on remote sensing data, so as to achieve accurate assessment of carbon sequestration benefits across multiple regions, factors, and scales.

[0005] A method for assessing the carbon sequestration benefits of wetlands based on remote sensing data includes the following steps: Acquire remote sensing images of wetlands; delineate target wetland areas based on the remote sensing images to obtain standard carbon sink areas (C1). n And special carbon sink areas, n=1, 2, 3; For standard carbon sink area C n The corresponding regional remote sensing image data and field sampling data were reacquired; based on the regional remote sensing image data and field sampling data, a regional carbon sink benefit assessment was conducted to obtain the regional carbon sink benefit assessment result G. n ; For special carbon sink areas, regional hydrological data and special regional remote sensing image data are acquired; based on the regional hydrological data and special regional remote sensing image data, the regional carbon sink benefits are assessed to obtain the special carbon sink benefits assessment results; The regional carbon sequestration benefit assessment results G n The carbon sequestration benefit assessment results are combined with the special carbon sequestration benefit assessment results to obtain the wetland carbon sequestration benefit assessment results.

[0006] As a preferred embodiment of the present invention, the specific steps for delineating target wetland areas based on wetland remote sensing images include: Wetland remote sensing images contain multimodal remote sensing data Y i, i = 1, 2, ..., I, where I is the total number of modal categories in the multimodal remote sensing data; The target wetland area is divided into grids to obtain the wetland carbon sequestration assessment grid G. m m=1,2,…,M; each wetland carbon sequestration assessment grid G m The unit area is the same; For extracting wetland carbon sequestration assessment grid G ​​from wetland remote sensing images m Spectral features, texture features, and topographic features; Carbon sink areas are classified based on spectral, textural, and topographic features, dividing the target wetland area into standard carbon sink areas C. n n=1, 2, 3; Exclude classified standard carbon sink areas C from all wetland carbon sink assessment grids within the target wetland area. n The corresponding wetland carbon sequestration assessment grid yields the wetland carbon sequestration assessment grid G ​​to be classified. m '; For the assessment grid G ​​of wetland carbon sinks to be classified m ', Based on a pre-constructed carbon sink potential thermofield, each wetland carbon sink assessment grid G ​​to be classified m 'Assign carbon sink potential score F' m Based on carbon sequestration potential score F m Further subdivision using fuzzy clustering algorithm will classify the wetland carbon sequestration assessment grid G ​​that meets the standard carbon sequestration area characteristics. m 'Divided into corresponding standard carbon sink areas C n Otherwise, it will be classified as a special carbon sink area.

[0007] As a preferred technical solution of the present invention, for standard carbon sink area C n The specific steps for assessing regional carbon sequestration benefits based on regional remote sensing image data and field sampling data include: Preprocessing of regional remote sensing image data yields preprocessed regional remote sensing image data; Feature extraction is performed on the field sampling data to obtain the sampling data features; feature analysis is performed based on the sampling data features to obtain the intensity of anthropogenic carbon interference; Using a pre-trained regional carbon sink assessment model, standard carbon sink predictions are performed on preprocessed regional remote sensing image data, sampling data features, and the intensity of anthropogenic carbon disturbance, yielding the regional carbon sink benefit assessment result G. n .

[0008] As a preferred embodiment of the present invention, the specific steps for assessing the regional carbon sink benefits based on regional hydrological data and remote sensing image data of specific regions in special carbon sink areas include: Preprocessing is performed on remote sensing image data of a specific region to obtain preprocessed remote sensing image data of the specific region; a time series set of remote sensing images is then established based on the temporal characteristics of the preprocessed remote sensing image data of the specific region. The regional hydrological data is projected onto the time series set of remote sensing images using spatial interpolation methods to obtain a hydrological data raster layer; The vegetation index time series and water body index time series are calculated for each unit time of the remote sensing image time series set; a dynamic index feature layer is constructed based on the vegetation index time series and water body index time series. A hydrological impulse response factor layer is constructed by combining a hydrological data raster layer and a dynamic index feature layer; Clustering of the hydrological impulse response factor layer yields multiple dynamic carbon sink sub-regions; Pixel-level carbon sequestration benefits are calculated for each dynamic carbon sequestration sub-region to obtain the sub-region carbon sequestration benefits. By integrating the carbon sink benefits of all sub-regions, a special carbon sink benefit assessment result is obtained; Among these methods, the boundary hydrological data of special carbon sink areas are obtained to update these areas.

[0009] As a preferred embodiment of the present invention, the specific steps for updating the special carbon sink area by acquiring boundary hydrological data of the special carbon sink area include: Obtain boundary hydrological data corresponding to the wetland carbon sink assessment grid in special carbon sink areas; Based on the boundary hydrological data, the characteristic change rate is identified, and the fluctuation characteristic change rate is obtained; Based on the preset mapping rules, the rate of change of fluctuation characteristics is judged, and the target wetland area corresponding to the wetland carbon sink assessment grid of the boundary hydrological data is reallocated for the next wetland carbon sink benefit assessment.

[0010] As a preferred technical solution of the present invention, the regional carbon sequestration benefit assessment result G n The specific steps for integrating carbon sink benefit assessment results with special carbon sink benefit assessment results include: Based on standard carbon sink area C n And special carbon sink areas, extract the regional boundary carbon sink assessment results; The assessment results of the regional boundary carbon sink are weighted based on classification confidence levels to obtain the assessment results of the boundary carbon sink benefits. The regional carbon sequestration benefit assessment results G n The wetland carbon sink benefit assessment results are obtained by weighting the special carbon sink benefit assessment results and the boundary carbon sink benefit assessment results.

[0011] A wetland carbon sequestration benefit assessment system based on remote sensing data includes: The carbon sink assessment zoning module includes wetland assessment zoning units for acquiring wetland remote sensing images; based on the wetland remote sensing images, target wetland areas are delineated to obtain the standard carbon sink zone C. n And special carbon sink areas, n=1, 2, 3; The carbon sink benefit assessment module includes standard carbon sink assessment units, special carbon sink assessment units, and wetland carbon sink assessment units; Standard carbon sink assessment units are used for standard carbon sink areas C n The corresponding regional remote sensing image data and field sampling data were reacquired; based on the regional remote sensing image data and field sampling data, a regional carbon sink benefit assessment was conducted to obtain the regional carbon sink benefit assessment result G. n ; The special carbon sink assessment unit is used to acquire regional hydrological data and special regional remote sensing image data for special carbon sink areas; based on the regional hydrological data and special regional remote sensing image data, the regional carbon sink benefit assessment is carried out to obtain the special carbon sink benefit assessment results; Wetland carbon sequestration assessment units are used to assess the regional carbon sequestration benefits (G) n The carbon sequestration benefit assessment results are combined with the special carbon sequestration benefit assessment results to obtain the wetland carbon sequestration benefit assessment results.

[0012] The present invention has the following advantages: 1. This invention achieves high-resolution analysis of the spatial heterogeneity of wetland areas through multimodal remote sensing data fusion and fine grid division, which can accurately identify different types of carbon sink areas and their distribution boundaries, thus improving the scientific nature and granularity of carbon sink zoning. By constructing a dynamic index feature layer through remote sensing time series and constructing a hydrological impulse response factor layer in combination with hydrological data, and then clustering and dividing it, dynamic identification of dynamic carbon sink areas and pixel-level benefit calculation are realized.

[0013] 2. This invention constructs a hydrological boundary update mechanism, enabling the boundaries of special carbon sink areas to be dynamically adjusted according to the temporal changes in hydrological conditions. This avoids estimation bias caused by static boundaries and enhances the model's adaptability and accuracy in response to climate change, extreme events, or human interference. At the same time, the weighted mechanism that integrates classification confidence and ecological sensitivity factor weights achieves continuity and credibility control in boundary area carbon sink estimation, greatly improving the robustness and ecological rationality of the fusion results. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of a wetland carbon sequestration benefit assessment system based on remote sensing data, used in an embodiment of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this invention.

[0016] Example 1: A method for assessing the carbon sequestration benefits of wetlands based on remote sensing data, comprising the following steps: Acquire remote sensing images of wetlands; delineate target wetland areas based on the remote sensing images to obtain standard carbon sink areas (C1). n And special carbon sink areas, n=1, 2, 3; The standard carbon sink area includes a vegetation carbon sink area, a water body carbon sink area, and a soil carbon sink area; n=1 corresponds to the vegetation carbon sink area, n=2 corresponds to the water body carbon sink area, and n=3 corresponds to the soil carbon sink area. In assessing the carbon sequestration benefits of wetlands, classifying target areas based on remote sensing images is fundamental to identifying the spatial structure of carbon sequestration. Vegetation carbon sequestration zones refer to areas where atmospheric carbon dioxide is absorbed and fixed in biomass and litter primarily through wetland vegetation (such as emergent plants like reeds, cattails, and water onions, as well as floating-leaved or submerged plants). Their carbon sequestration function manifests as the formation of organic carbon through photosynthesis, with some of it subsequently transferred to the soil through litter to form stable carbon storage. Water body carbon sequestration zones refer to the carbon sequestration function achieved by water bodies themselves (such as lakes, floodplains, and river floodplains) through the processes of carbon dissolution, suspension, and sedimentation. This includes the long-term storage of dissolved organic carbon (DOC) and particulate organic carbon (POC). In some areas, such as still water areas, methane absorption or carbonate deposition reactions may also occur. Soil carbon sequestration zones refer to areas where wetland subsoil has long-term carbon sequestration through organic matter accumulation and microbial activity, especially in peat layers and anaerobic soils formed under reducing conditions. These soils exhibit high carbon storage stability and are the core part of long-term carbon sequestration in wetland systems.

[0017] Special carbon sink areas differ from the above standard classifications. They refer to areas with atypical carbon sink processes or special carbon behaviors, and these areas exhibit carbon sink behaviors that vary significantly over time.

[0018] The specific steps for delineating target wetland areas based on wetland remote sensing images include: Wetland remote sensing images contain multimodal remote sensing data Y i , i=1,2,…,I, where I is the total number of modal categories in the multimodal remote sensing data; Multimodal remote sensing data typically includes four main data sources: hyperspectral data, synthetic aperture radar (SAR) data, lidar data, and thermal infrared (TIR) ​​remote sensing data. These data are responsible for acquiring multidimensional information such as spectrum, structure, elevation, and temperature. Hyperspectral data has hundreds of bands and can identify vegetation types (C3 / C4 plants), dissolved substances in water (CDOM, DOC), and subtle spectral features such as organic matter and carbonates in the soil. It is mainly used to extract spectral features. Secondly, SAR data can penetrate clouds and some vegetation cover and is used to extract surface roughness, water boundaries, and micro-topographic relief. Further extraction of texture features using methods such as arrays can also reflect the wetland surface moisture state; secondly, lidar data (such as GEDI or ICESat-2) can provide high-precision three-dimensional topographic point cloud data for calculating the topographic features of wetland areas, such as surface curvature, concavity index, slope, and terrain distribution. This information is particularly crucial for identifying active soil carbon sink areas such as stagnant water areas, accumulation areas, and micro-depressions; finally, thermal infrared data is used to provide surface temperature information, which helps to identify phenomena such as methane release hotspots, water body temperature differentiation, and wetland-dryland boundaries, indirectly reflecting carbon flux fluctuations and microbial activity potential, and is an important clue for identifying special carbon sink areas.

[0019] The target wetland area is divided into grids to obtain the wetland carbon sequestration assessment grid G. m m=1,2,…,M; each wetland carbon sequestration assessment grid G m The unit area is the same; The specific steps for grid division are as follows: Using vector data (such as administrative boundaries, wetland reserve red lines) or masks in remote sensing images, the spatial extent of the target wetland is clearly defined, i.e., the target wetland area, with the unit being square meters or square kilometers; the size of each grid cell is manually set by professional technicians or the grid side length is determined based on the actual pixel size of the remote sensing image, i.e., the size of the unit area; using the boundary of the target area as the clipping range, it is divided into equal intervals according to the set unit area to form a regular grid system. Each grid has a fixed number and spatial location and can be spatially aligned with remote sensing data and topographic data; For extracting wetland carbon sequestration assessment grid G ​​from wetland remote sensing images mThe data includes spectral features, textural features, and topographic features. Spectral features refer to indicators extracted from multispectral or hyperspectral remote sensing images that reflect differences in the reflectivity of land cover surfaces. These mainly include the reflectance characteristics of different land surface types such as vegetation, water bodies, and soil under different electromagnetic wave bands. Textural features describe the spatial relationships and grayscale variation patterns between pixels in remote sensing images, reflecting the spatial organization of land cover, such as vegetation density, water fragmentation, and mudflat roughness. Topographic features are mainly derived from digital elevation models (DEMs) or lidar (LiDAR) data, describing micro-geomorphic information such as surface undulations, slope, morphology, and water flow paths. Carbon sink areas are classified based on spectral, textural, and topographic features, dividing the target wetland area into standard carbon sink areas C. n n=1, 2, 3; Exclude classified standard carbon sink areas C from all wetland carbon sink assessment grids within the target wetland area. n The corresponding wetland carbon sequestration assessment grid yields the wetland carbon sequestration assessment grid G ​​to be classified. m '; In the process of classifying wetland carbon sink areas, the first step is to assess each wetland carbon sink grid G. m The corresponding spectral, texture, and topographic features are extracted. These features form a high-dimensional feature vector, used to represent the ecological state and physical properties of the grid. A trained classification model is constructed or loaded, such as one based on random forest, support vector machine, or an improved U-Net deep learning network. These features are used as input to predict the carbon sink type of the grid. The model outputs multi-class classification results and their corresponding confidence values, representing the probability that the grid belongs to each type of carbon sink region (vegetation carbon sink region, water carbon sink region, soil carbon sink region). To ensure the reliability of the classification results, a confidence threshold is set. When the model predicts a probability of a certain category higher than this threshold, the grid is directly assigned to the corresponding standard carbon sink region C. n The entire target wetland is divided into zones. All grids with high confidence and clear classification are directly marked as classified areas, corresponding to their respective standard carbon sink zones. However, for grids whose probability values ​​for each category output by the model are close and cannot form a significant preference during the classification process, their maximum confidence does not reach the set threshold due to reasons such as ambiguous ecological performance, unclear boundaries, or being located in the classification boundary zone. Therefore, these grids that are not classified into the standard carbon sink zone are extracted from the target wetland area to obtain the wetland carbon sink assessment grid to be classified for subsequent operations.

[0020] For the assessment grid G ​​of wetland carbon sinks to be classified m ', Based on a pre-constructed carbon sink potential thermofield, each wetland carbon sink assessment grid G ​​to be classified m 'Assign carbon sink potential score F' mBased on carbon sequestration potential score F m Further subdivision using fuzzy clustering algorithm will classify the wetland carbon sequestration assessment grid G ​​that meets the standard carbon sequestration area characteristics. m 'Divided into corresponding standard carbon sink areas C n Otherwise, it will be designated as a special carbon sink area; In the delineation of wetland carbon sink areas, a carbon sink probability thermodynamic field is introduced for the carbon sink assessment grids of wetlands whose confidence in the initial classification is insufficient. This is to explore their potential carbon sink attributes and further assist in determining whether they should be classified into standard carbon sink areas or separately marked as special carbon sink areas. The carbon sink probability thermodynamic field is essentially a spatialized and continuous probability auxiliary layer. It represents the spatial potential gradient of carbon sink capacity at a certain point in a specific area. This thermodynamic field is not directly derived from a single image, but is modeled based on the fusion of multi-source information such as historical ecological monitoring data, remote sensing time series indices, topographic water accumulation models, hydrological inundation frequency, and microbial index change layers. The process of constructing this thermal field includes: First, collecting historical remote sensing data such as hyperspectral, SAR, and thermal infrared data from multiple periods, and overlaying them with spatial monitoring layers of ecological evolution records, such as vegetation expansion, peat accumulation points, and methane release zones. Statistical learning or spatial weighted regression (such as GWR) is used to estimate the carbon sink contribution potential of each region. Second, these potentials are spatially smoothed to construct a continuous raster layer, called the carbon sink potential thermal field. On this map, each pixel location has a value, namely the carbon sink potential score, which represents the probability or comprehensive score of the location being judged as having substantial carbon sink capacity. The higher the score, the closer the ecological conditions and historical response characteristics of the pixel are to known carbon sink areas, even though it was not clearly classified in the previous step. The carbon sink potential scores of all pixels in the wetland carbon sink assessment grid to be classified are combined to obtain the carbon sink potential score of the corresponding grid.

[0021] The wetland carbon sink assessment grid to be classified is further divided by a fuzzy clustering algorithm. Fuzzy clustering (such as the fuzzy C-means algorithm) is different from hard classification methods. It allows each grid to have membership in multiple categories at the same time, and is suitable for spatial phenomena with fuzzy boundaries and overlapping types in wetlands.

[0022] In practice, the spectral, textural, and topographical features of the wetland carbon sink assessment grid to be classified, along with their corresponding carbon sink potential scores, are first used as feature vectors and input into a fuzzy clustering algorithm. The algorithm automatically fits the optimal cluster centers (i.e., carbon sink types). Then, the membership degree of each grid to each type is calculated. If the membership degree of a grid to a certain standard carbon sink type (such as vegetation or water) is greater than a set threshold and the carbon sink potential score is higher than an empirical critical value, then it is classified into the corresponding standard carbon sink area C. nIf the membership of the grid to all standard types is low, or its carbon sink potential score differs significantly from the known standard samples, it is classified as a special carbon sink area and marked as requiring further dynamic monitoring or separate modeling.

[0023] For standard carbon sink area C n The corresponding regional remote sensing image data and field sampling data were reacquired; based on the regional remote sensing image data and field sampling data, a regional carbon sink benefit assessment was conducted to obtain the regional carbon sink benefit assessment result G. n ; For standard carbon sink area C n The specific steps for assessing regional carbon sequestration benefits based on regional remote sensing image data and field sampling data include: Preprocessing of regional remote sensing image data yields preprocessed regional remote sensing image data. The preprocessing steps involve quality enhancement and spatial registration of the original remote sensing data to make it suitable for analysis. Preprocessing typically includes the following steps: 1) Radiometric calibration and atmospheric correction to remove atmospheric scattering and inconsistencies in radiation; 2) Geometric correction and orthorectification to ensure the image accurately matches the actual geographical location of the wetland; 3) Cloud masking and noise removal using methods such as Fmask and SCL classification to remove clouds, cloud shadows, fog, and edge noise regions; 4) Spectral index construction to calculate remote sensing indices reflecting carbon absorption and hydrological characteristics, forming a unified multi-channel input image; finally, image cropping and segmentation are performed to confine the image within the standard carbon sink area boundary and convert it into an evaluation grid structure input model. Feature extraction is performed on field sampling data to obtain sampling data features; feature analysis is conducted based on the sampling data features to obtain the intensity of anthropogenic carbon interference; field sampling data is an important source for model training and correction; field sampling within the standard carbon sink area should include carbon sink element sampling and environmental auxiliary parameters at representative points, mainly including: aboveground biomass, growth rate, litter amount, leaf area index (LAI) of vegetation quadrats; dissolved organic carbon (DOC), particulate organic carbon (POC), pH value, carbonate concentration, transparency, etc. of water samples; soil organic carbon (SOC), bulk density, moisture, redox potential, iron oxide content, etc. of soil samples; sampling data features refer to the input feature variable vector formed after normalizing, interpolating, statistically transforming or combining these raw indicators to reflect the carbon sink capacity and related ecological mechanisms of the sampling point; Anthropogenic carbon disturbance intensity refers to the degree of decline in carbon sink function within a standard carbon sink area caused by human activities (such as drainage, farming, dam construction, canal digging, and engineering construction). The process of deriving anthropogenic carbon disturbance intensity through feature analysis based on sampled data is essentially a comprehensive interpretation mechanism that couples ecological observation data with remote sensing change characteristics. Its purpose is to identify the degree of intervention caused by human activities on the carbon sink system within the region. The sampling points of the field data are mapped to remote sensing images, and the changing trends of indicators such as vegetation index, water index, surface temperature, and synthetic aperture radar texture in the time series are extracted to construct a time curve matrix containing multiple rate-of-change indicators. A change point detection algorithm is used to determine the time periods and significance of abrupt changes in the remote sensing indicators and match them with the abrupt change times of ecological factors at the sampling points. If the two are highly synchronized in time and the direction of the abrupt change is consistent with typical anthropogenic disturbance responses, then the point can be determined to have a high-confidence anthropogenic carbon disturbance intensity.

[0024] Using a pre-trained regional carbon sink assessment model, standard carbon sink predictions are performed on preprocessed regional remote sensing image data, sampling data features, and the intensity of anthropogenic carbon disturbance, yielding the regional carbon sink benefit assessment result G. n ; The core of the regional carbon sink benefit assessment model is a pre-trained prediction model, the goal of which is to learn the nonlinear mapping relationship between remote sensing features and carbon sink observations. This model can be constructed using methods such as random forest (RF), XGBoost, lightweight neural networks (such as 1D-CNN or shallow MLP), or hybrid ensemble models.

[0025] The input to the regional carbon sink benefit assessment model consists of three parts: first, the grid features of preprocessed remote sensing images, such as NDVI, NDWI, and SAR texture; second, the features of sampled data, such as derived indices like LAI, biomass estimation, and SOC; and third, the anthropogenic carbon disturbance intensity index. The model output is the predicted carbon sink benefit value for each grid cell. The training dataset mainly comes from field monitoring points and high-confidence remote sensing samples, and sample labels can be constructed by combining historical ecological evolution sample areas. The training objective is to minimize the difference between the model's predicted value and the measured carbon storage data (e.g., RMSE). The training termination condition can be set to the error converging to a set threshold (e.g., when MAE < 0.15, the model terminates training) or reaching the maximum number of iterations. The final model can be used to map carbon sink benefits to any standard carbon sink area and has a certain generalization ability, supporting dynamic predictions for different years and regions.

[0026] For special carbon sink areas, regional hydrological data and special regional remote sensing image data are acquired; based on the regional hydrological data and special regional remote sensing image data, the regional carbon sink benefits are assessed to obtain the special carbon sink benefits assessment results; For special carbon sink areas, the specific steps for assessing regional carbon sink benefits based on regional hydrological data and remote sensing image data of specific areas include: Preprocessing is performed on remote sensing image data of a specific region to obtain preprocessed remote sensing image data of the specific region; a time series set of remote sensing images is established based on the temporal characteristics of the preprocessed remote sensing image data of the specific region; the remote sensing images are organized into a time series set of remote sensing images in chronological order, which covers multiple temporal observation nodes within at least one year and maintains image alignment in space, thereby enabling the construction of dynamically changing surfaces.

[0027] Regional hydrological data are projected onto a time series dataset of remote sensing images using spatial interpolation methods to obtain a hydrological data raster layer. Regional hydrological observation data (including water level, flow velocity, rainfall, evapotranspiration, soil moisture, etc.) are spatiotemporally normalized and standardized, and discrete hydrological point data are mapped onto a raster structure that matches the time series of remote sensing images using spatial interpolation methods (such as Kriging interpolation, spline interpolation, or inverse distance weighting) to form a hydrological data raster layer corresponding to each time period.

[0028] The vegetation index time series and water body index time series are calculated for each unit time of the remote sensing image time series set; a dynamic index feature layer is constructed based on the vegetation index time series and water body index time series. Vegetation and water indices are calculated for remote sensing images at each time point to generate corresponding vegetation and water index time series. Then, statistical parameters such as dynamic change rate, intra-annual variation, and frequency of crossing upper and lower critical values ​​are calculated through these series to construct dynamic index feature layers. These layers reflect the sensitivity and response elasticity of carbon sink function with seasonal changes.

[0029] A hydrological impulse response factor layer is constructed by combining a hydrological data raster layer and a dynamic index feature layer. The hydrological data raster layer and the dynamic index feature layer are then fused at the pixel level to construct the hydrological impulse response factor layer. This factor is an indicator reflecting the ability of the wetland system to undergo structural transformation of the carbon sink mechanism under the stimulation of hydrological events (such as flooding during the rainy season and retreat during the dry season). The construction methods include multi-factor weighting, principal component analysis, or extraction of abrupt and synchronization signals using a time-series aggregation model, ultimately generating a spatially expressible hydrological response potential energy map.

[0030] Clustering of the hydrological impulse response factor layer yields multiple dynamic carbon sink sub-regions. Subsequently, unsupervised clustering algorithms (such as K-means, ISODATA, or density clustering) are used to further divide the hydrological impulse response factor layer into multiple dynamic carbon sink sub-regions. Each sub-region represents a type of hydrological-ecological response pattern with different dynamic carbon sink regulation mechanisms.

[0031] Pixel-level carbon sink benefits are calculated for each dynamic carbon sink sub-region to obtain the sub-region carbon sink benefits. After the dynamic carbon sink sub-regions are divided, the pixel-level carbon sink benefit calculation stage begins. This process requires fitting the relationship between vegetation index, water index, hydrological factors and historical carbon flux measurements for each pixel in the time period, or introducing regional empirical models, such as semi-empirical models based on the relationship between vegetation index and carbon flux, biomass inversion models, or spatial-temporal regression models, to calculate the net carbon absorption of each pixel in the carbon sink cycle of the sub-region.

[0032] By integrating the carbon sequestration benefits of all sub-regions, a special carbon sequestration benefit assessment result is obtained. Finally, the carbon sequestration values ​​of pixels in each sub-region are weighted or area-normalized to generate the carbon sequestration benefit of each sub-region. The carbon sequestration benefits of all sub-regions are spatially integrated (using methods such as area-weighted average or regional weighted overlay) to obtain a complete special carbon sequestration benefit assessment result.

[0033] Among these measures, the boundary hydrological data of special carbon sink areas are obtained to update the special carbon sink areas; The specific steps for updating special carbon sink areas by obtaining boundary hydrological data include: Obtain boundary hydrological data corresponding to the wetland carbon sink assessment grid in special carbon sink areas; It is necessary to extract the boundary hydrological data of the corresponding wetland carbon sink assessment grid from the edge of the special carbon sink area. The specific steps are to spatially overlay the carbon sink assessment grid with the time series of remote sensing images, and extract the grid cells located on the boundary of the current special carbon sink area at each time node. The spatial location of these grids is close to the boundary between the carbon sink area and other areas (such as vegetation area and water body area). Subsequently, the hydrological observation values ​​corresponding to these boundary grids are extracted from the spatial interpolation hydrological raster at each time node. The indicators may include flooding frequency, surface water level, evapotranspiration, flow velocity, soil moisture, albedo change, surface temperature change, etc.

[0034] Based on the characteristic change rate identified from boundary hydrological data, the fluctuation characteristic change rate is obtained. In the time series, time-varying statistical parameters such as monthly variability, annual amplitude, and frequency of crossing upper and lower critical values ​​are calculated for the aforementioned boundary hydrological data. Combined with the amplitude and frequency of changes in dynamic indices derived from remote sensing, such as vegetation indices and water body indices, the fluctuation characteristic change rate characterizing the transition of spatial ecological states is further obtained. For example, if a boundary grid maintains a high water body index, a decrease in surface temperature, and a persistently higher soil moisture content than the regional median value for two consecutive months, accompanied by a positive jump in the plant index, then the composite change rate of the hydrological and ecological linkage characteristics at that location can be calculated to measure whether it is migrating from a water state to a vegetation state or a carbon sink enhancement state.

[0035] Based on preset mapping rules, the rate of change of fluctuation characteristics is judged, and the target wetland area corresponding to the wetland carbon sink assessment grid of the boundary hydrological data is reassigned for the next wetland carbon sink benefit assessment. Based on the above rate of change of fluctuation characteristics, a set of preset change identification and mapping rules needs to be established to determine whether the regional affiliation of the boundary grid should be adjusted. The preset mapping rules are usually composed of rule models or threshold logic. For example, certain critical change rates of indicators are set. If the vegetation index rise rate exceeds 0.1% per month or the hydrological impulse response intensity is greater than the set threshold, then the current grid is judged to have changed from a water body area to a vegetation carbon sink area. Or if the soil moisture fluctuation frequency is more than three times per month and the water body index is stable at a high level, then it is judged to be a special carbon sink area boundary with potential water accumulation carbon sequestration capacity, and should be included in the subsequent monitoring model.

[0036] Based on this, the spatial attribution labels of the boundary grid are updated, either by removing them from or incorporating them into the original special carbon sink area boundaries, forming a new regional mask layer or regional zoning vector map. This is used to redefine carbon sink functional areas in the next round of wetland carbon sink assessment, thereby realizing a dynamic update mechanism for the boundaries of special carbon sink areas in response to hydrological and ecological changes. This method can effectively avoid carbon sink estimation bias caused by static boundary assumptions and improve the model's spatiotemporal adaptability and ability to reflect ecological processes.

[0037] The regional carbon sequestration benefit assessment results G n By integrating the results of special carbon sink benefit assessments with the carbon sink benefit assessment results, the carbon sink benefit assessment results of wetlands are obtained. The regional carbon sequestration benefit assessment results G n The specific steps for integrating carbon sink benefit assessment results with special carbon sink benefit assessment results include: Based on standard carbon sink area C n For standard and special carbon sink areas, extract the carbon sink assessment results at the boundary of the region. Based on the already divided standard and special carbon sink areas, extract the carbon sink assessment results corresponding to the boundary of the two types of areas. This operation requires first constructing the boundary vector layer of the standard and special carbon sink areas, then spatially identifying the boundary grid or overlapping buffer zone of the two types of areas (usually set to three to five pixels wide), and extracting the carbon sink benefit values ​​corresponding to these boundary areas from all the assessment results to form the regional boundary carbon sink assessment results.

[0038] The assessment results of the regional boundary carbon sink are weighted based on classification confidence levels to obtain the assessment results of the boundary carbon sink benefits. By retrospectively analyzing the multi-class probability results output for each boundary grid during the previous classification stage, the maximum membership value is extracted as the classification confidence index. Then, a weighting function is constructed, using confidence as the weight, to linearly weight the boundary grid between the two assessment results for standard and special carbon sink areas, generating the boundary carbon sink benefit assessment result. Regions with higher confidence have a greater weight in retaining their original assessment value, while regions with lower confidence rely more on the average value of neighboring regions or fuzzy fusion, thus enhancing the assessment stability of the boundary region.

[0039] The regional carbon sequestration benefit assessment results G n The assessment results of special carbon sink benefits and boundary carbon sink benefits are weighted and calculated to obtain the assessment results of wetland carbon sink benefits. The results of regional carbon sink benefit assessment, special carbon sink benefit assessment, and boundary carbon sink benefit assessment are weighted and fused according to regional area and ecological importance to obtain a complete wetland carbon sink benefit assessment result layer. The weighting method can be a combination function of area proportion weight, ecological sensitivity factor weight, or uncertainty correction weight. The three types of results are spatially aligned before fusion calculation. The ecological sensitivity factor weight refers to the weight coefficient used to express the degree of response of different regions to ecological disturbances or environmental changes when spatially fusing or weighting the estimation of wetland carbon sink benefits. This weight represents the vulnerability or variability of a region's ecosystem to carbon sink function and is used to adjust its influence in the overall carbon sink benefit calculation. Simply put, the higher the ecological sensitivity of a region, the greater its weight in the carbon sink estimation result can be, to reflect its priority in ecological protection and carbon management.

[0040] For example, a confidence-corrected model is used in the boundary transition zone, and independent assessment values ​​are used in the standard zone and the special zone. The final output is a high-resolution wetland carbon sink benefit assessment result with continuous spatial distribution and integration of response characteristics of different ecological mechanisms, which can be used for subsequent carbon trading, restoration priority ranking or policy simulation scenario analysis.

[0041] In this embodiment, a wetland nature reserve is selected as the target research area. A wetland remote sensing image database is constructed by acquiring multimodal remote sensing images, including hyperspectral data, multispectral data, synthetic aperture radar data, lidar data, and thermal infrared remote sensing data. Based on the image processing workflow, atmospheric correction, geometric registration, texture reconstruction, and topographic factor extraction are performed on the data. The entire wetland is divided into a standard 10m × 10m carbon sink assessment grid. The spectral, texture, and curvature topographic features of each grid are extracted. Subsequently, an improved deep learning model is used to classify carbon sink areas. The classification results identify vegetation carbon sink areas (C1), water carbon sink areas (C2), soil carbon sink areas (C3), and areas with blurred boundaries and significant mixing states that are not yet classified. To process these grids to be classified, a carbon sink potential thermodynamic field based on historical carbon flux observations and temporal vegetation and water index evolution is constructed. Each grid is assigned a carbon sink potential score. The system combines this score with a fuzzy clustering algorithm to reclassify some grids into C1, C2, or C3, while the rest are labeled as special carbon sink areas, such as silt accumulation areas, carbonate crystallization zones, and areas with abnormal methane release.

[0042] Within the standard carbon sink area, approximately X sampling points were deployed in the field to acquire ground data such as biomass, soil organic carbon, vegetation diversity, and dissolved organic carbon in water. Combined with carbon stable isotope tracking and microbial activity detection, a carbon cycle factor library was constructed. Through analysis linked to remote sensing features, anthropogenic disturbance intensity indices (e.g., drainage ditch distribution, intensity of agricultural activity disturbance) were estimated. These indices, along with remote sensing imagery, were input into a pre-trained carbon sink assessment model (built based on an ensemble learning regression network) to obtain regional carbon sink benefit assessment results, measured in tons of carbon per hectare per year. For special carbon sink areas, a remote sensing time series set was established using multi-temporal remote sensing images from the previous y years. Vegetation indices, water indices, and humidity variations were calculated over the time series, and a hydrological data raster layer was constructed using measured water level data from lake hydrological stations. By constructing a hydrological impulse response factor layer, carbon dynamic evolution zones sensitive to hydrological fluctuations within the region were extracted. Multiple dynamic carbon sink sub-regions were then divided using a clustering algorithm. Net carbon sink benefits were calculated at the pixel level within each sub-region, and the results were finally aggregated to obtain the special carbon sink benefit assessment results. In the boundary region, targeting the intersection of the standard and special zones, the carbon sink estimate is corrected by extracting the historical classification confidence scores of the boundary grids and weighting them with the carbon potential scores, thus generating a boundary carbon sink benefit assessment result. Finally, the regional carbon sink benefit assessment result G is used to... n The results of special carbon sink benefit assessment and boundary carbon sink benefit assessment are fused according to the weight of ecological sensitivity factors to output a complete carbon sink benefit assessment result for the wetland nature reserve, estimating that the current annual carbon sink of the wetland is approximately Z tons of carbon dioxide equivalent; where X, y, and Z are specific values.

[0043] Example 2: A wetland carbon sequestration benefit assessment system based on remote sensing data, see [link to example]. Figure 1 As shown, it includes: The carbon sink assessment zoning module includes wetland assessment zoning units for acquiring wetland remote sensing images; based on the wetland remote sensing images, target wetland areas are delineated to obtain the standard carbon sink zone C. n And special carbon sink areas, n=1, 2, 3; The carbon sink benefit assessment module includes standard carbon sink assessment units, special carbon sink assessment units, and wetland carbon sink assessment units; Standard carbon sink assessment units are used for standard carbon sink areas C n The corresponding regional remote sensing image data and field sampling data were reacquired; based on the regional remote sensing image data and field sampling data, a regional carbon sink benefit assessment was conducted to obtain the regional carbon sink benefit assessment result G. n ; The special carbon sink assessment unit is used to acquire regional hydrological data and special regional remote sensing image data for special carbon sink areas; based on the regional hydrological data and special regional remote sensing image data, the regional carbon sink benefit assessment is carried out to obtain the special carbon sink benefit assessment results; The wetland carbon sequestration assessment unit is used to assess the regional carbon sequestration benefits (G) n The carbon sequestration benefit assessment results are combined with the special carbon sequestration benefit assessment results to obtain the wetland carbon sequestration benefit assessment results.

[0044] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A method for assessing the carbon sequestration benefits of wetlands based on remote sensing data, characterized in that, Includes the following steps: Acquire remote sensing images of wetlands; delineate target wetland areas based on the remote sensing images to obtain standard carbon sink areas (C1). n And special carbon sink areas, n=1, 2, 3; For standard carbon sink area C n Reacquire the corresponding regional remote sensing image data and field sampling data; Regional carbon sequestration benefits were assessed based on regional remote sensing image data and field sampling data, resulting in the regional carbon sequestration benefit assessment result G. n ; For special carbon sink areas, acquire regional hydrological data and special regional remote sensing image data; Regional carbon sink benefits were assessed based on regional hydrological data and remote sensing image data of specific regions, and the results of the assessment of specific carbon sink benefits were obtained. The regional carbon sequestration benefit assessment results G n By integrating the results of special carbon sink benefit assessments with the carbon sink benefit assessment results, the carbon sink benefit assessment results of wetlands are obtained. For special carbon sink areas, the specific steps for assessing regional carbon sink benefits based on regional hydrological data and remote sensing image data of specific areas include: Preprocessing is performed on remote sensing image data of a specific region to obtain preprocessed remote sensing image data of the specific region; a time series set of remote sensing images is then established based on the temporal characteristics of the preprocessed remote sensing image data of the specific region. The regional hydrological data is projected onto the time series set of remote sensing images using spatial interpolation methods to obtain a hydrological data raster layer; The vegetation index time series and water body index time series are calculated for each unit time of the remote sensing image time series set; a dynamic index feature layer is constructed based on the vegetation index time series and water body index time series. A hydrological impulse response factor layer is constructed by combining a hydrological data raster layer and a dynamic index feature layer; Clustering of the hydrological impulse response factor layer yields multiple dynamic carbon sink sub-regions; Pixel-level carbon sequestration benefits are calculated for each dynamic carbon sequestration sub-region to obtain the sub-region carbon sequestration benefits. By integrating the carbon sink benefits of all sub-regions, a special carbon sink benefit assessment result is obtained; Among these methods, the boundary hydrological data of special carbon sink areas are obtained to update these areas.

2. The wetland carbon sequestration benefit assessment method based on remote sensing data according to claim 1, characterized in that, The specific steps for delineating target wetland areas based on wetland remote sensing images include: Wetland remote sensing images contain multimodal remote sensing data Y i , i = 1, 2, ..., I, where I is the total number of modal categories in the multimodal remote sensing data; The target wetland area is divided into grids to obtain the wetland carbon sequestration assessment grid G. m m=1,2,…,M; each wetland carbon sequestration assessment grid G m The unit area is the same; For extracting wetland carbon sequestration assessment grid G ​​from wetland remote sensing images m Spectral features, texture features, and topographic features; Carbon sink areas are classified based on spectral, textural, and topographic features, dividing the target wetland area into standard carbon sink areas C. n n=1, 2, 3; Exclude classified standard carbon sink areas C from all wetland carbon sink assessment grids within the target wetland area. n The corresponding wetland carbon sequestration assessment grid yields the wetland carbon sequestration assessment grid G ​​to be classified. m '; For the assessment grid G ​​of wetland carbon sinks to be classified m ', Based on a pre-constructed carbon sink potential thermofield, each wetland carbon sink assessment grid G ​​to be classified m 'Assign carbon sink potential score F' m Based on carbon sequestration potential score F m Further subdivision using fuzzy clustering algorithm will classify the wetland carbon sequestration assessment grid G ​​that meets the standard carbon sequestration area characteristics. m 'Divided into corresponding standard carbon sink areas C n Otherwise, it will be classified as a special carbon sink area.

3. The wetland carbon sequestration benefit assessment method based on remote sensing data according to claim 2, characterized in that, For standard carbon sink area C n The specific steps for assessing regional carbon sequestration benefits based on regional remote sensing image data and field sampling data include: Preprocessing of regional remote sensing image data yields preprocessed regional remote sensing image data; Feature extraction is performed on the field sampling data to obtain the sampling data features; feature analysis is performed based on the sampling data features to obtain the intensity of anthropogenic carbon interference; Using a pre-trained regional carbon sink assessment model, standard carbon sink predictions are performed on preprocessed regional remote sensing image data, sampling data features, and the intensity of anthropogenic carbon disturbance, yielding the regional carbon sink benefit assessment result G. n .

4. The wetland carbon sequestration benefit assessment method based on remote sensing data according to claim 3, characterized in that, The specific steps for updating special carbon sink areas by obtaining boundary hydrological data include: Obtain boundary hydrological data corresponding to the wetland carbon sink assessment grid in special carbon sink areas; Based on the boundary hydrological data, the characteristic change rate is identified, and the fluctuation characteristic change rate is obtained; Based on the preset mapping rules, the rate of change of fluctuation characteristics is judged, and the target wetland area corresponding to the wetland carbon sink assessment grid of the boundary hydrological data is reallocated for the next wetland carbon sink benefit assessment.

5. The wetland carbon sequestration benefit assessment method based on remote sensing data according to claim 4, characterized in that, The regional carbon sequestration benefit assessment results G n The specific steps for integrating carbon sink benefit assessment results with special carbon sink benefit assessment results include: Based on standard carbon sink area C n And special carbon sink areas, extract the regional boundary carbon sink assessment results; The assessment results of the regional boundary carbon sink are weighted based on classification confidence levels to obtain the assessment results of the boundary carbon sink benefits. The regional carbon sequestration benefit assessment results G n The wetland carbon sink benefit assessment results are obtained by weighting the special carbon sink benefit assessment results and the boundary carbon sink benefit assessment results.

6. A wetland carbon sequestration benefit assessment system based on remote sensing data, characterized in that, The system, as described in any one of claims 1-5, is a method for assessing the benefits of wetland carbon sequestration based on remote sensing data, comprising: The carbon sink assessment zoning module includes wetland assessment zoning units for acquiring wetland remote sensing images; based on the wetland remote sensing images, target wetland areas are delineated to obtain the standard carbon sink zone C. n And special carbon sink areas, n=1, 2, 3; The carbon sink benefit assessment module includes standard carbon sink assessment units, special carbon sink assessment units, and wetland carbon sink assessment units; Standard carbon sink assessment units are used for standard carbon sink areas C n The corresponding regional remote sensing image data and field sampling data were reacquired; based on the regional remote sensing image data and field sampling data, a regional carbon sink benefit assessment was conducted to obtain the regional carbon sink benefit assessment result G. n ; The special carbon sink assessment unit is used to acquire regional hydrological data and special regional remote sensing image data for special carbon sink areas; based on the regional hydrological data and special regional remote sensing image data, the regional carbon sink benefit assessment is carried out to obtain the special carbon sink benefit assessment results; Wetland carbon sequestration assessment units are used to assess the regional carbon sequestration benefits (G) n The carbon sequestration benefit assessment results are combined with the special carbon sequestration benefit assessment results to obtain the wetland carbon sequestration benefit assessment results.

Images