A carbon sequestration data monitoring system for mangrove restoration areas

Abstract

This invention provides a carbon sink data monitoring system for mangrove restoration areas, relating to the field of ecological carbon sink measurement technology. The system includes: an acquisition module for collecting raw carbon exchange datasets at the atmosphere-vegetation-soil interface using a multi-type sensor network deployed in intertidal wetlands within the mangrove restoration area; a data generation module for performing spatiotemporal coordinate matching based on the raw dataset to generate a multidimensional carbon flux fusion dataset jointly indexed by geographic location and time; and an identification module for calculating local spatiotemporal dispersion and global correlation thresholds based on the fused dataset, identifying and marking anomalous data points. This invention enables real-time and multidimensional monitoring of carbon sinks in mangrove restoration areas, improving monitoring efficiency and data reliability, and providing support for evaluating restoration and management strategies.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological carbon sink measurement, and in particular to a carbon sink data monitoring system for a mangrove restoration area. Background Art

[0002] Traditional monitoring relies heavily on single, fixed-point sensors to collect data, making it difficult to capture the diverse ecological landscape of mangroves. For example, in the intertidal zone of the restoration area, carbon dioxide concentration sensors are installed in only a few areas. When the tide rises, some sensors are submerged by seawater, interrupting data collection and preventing the acquisition of carbon exchange information during the submerged period. Furthermore, traditional sensors lack the ability to adapt to environmental changes. In high-salt and high-humidity environments, sensor performance degrades, reducing data accuracy and resulting in significant deviations from actual soil carbon flux data.

[0003] Furthermore, traditional data analysis technologies rely on fixed, pre-set standards and processing procedures, making them unable to keep pace with the dynamic changes in mangrove ecosystems. During typhoon season, mangrove vegetation structure is damaged, and carbon sequestration capacity changes. However, traditional monitoring systems process data according to fixed standards and procedures, failing to promptly adjust data processing strategies based on vegetation damage. This results in calculated carbon sequestration amounts that are inconsistent with actual conditions. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a carbon sequestration data monitoring system for mangrove restoration areas, which realizes dynamic monitoring and spatial correction of carbon sequestration in mangrove restoration areas through multi-type sensor fusion and adaptive grid division.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0006] First, a carbon sequestration data monitoring system for a mangrove restoration area includes:

[0007] The acquisition module is used to collect raw data sets of carbon exchange at the atmosphere-vegetation-soil interface based on a multi-type sensor network deployed in the intertidal wetlands of the mangrove restoration area;

[0008] The data generation module is used to match the spatiotemporal coordinates based on the original dataset to generate a multi-dimensional carbon flux fusion dataset with a joint index of geographic location and time;

[0009] The recognition module is used to calculate the local spatiotemporal discreteness and global correlation threshold based on the fused data set, and identify and mark abnormal data points;

[0010] The gridding module is used to determine four boundary detection points within the restoration area based on abnormal data points, forming an irregular polygonal monitoring area that changes with ecological processes, and performing non-uniform grid cutting on the polygonal area to generate a set of adaptive grid cells covering the monitoring area;

[0011] The indicator generation module is used to calculate the terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency of each grid cell set within the tidal cycle, and generate spatial correction quantitative indicators;

[0012] The compensation and correction module is used to extract sensor data within the grid cell, combine it with the spatial correction quantitative index to compensate for the carbon flux value, and obtain the corrected carbon sink data set;

[0013] The comprehensive assessment module is used to calculate the carbon sink stability index, spatial coverage contribution index and ecological process representation index of each node during the assessment period based on the corrected carbon sink dataset, integrate the three indicators into comprehensive assessment parameters, and determine the prominent nodes as the final layout points for carbon sink monitoring based on the distribution characteristics of the comprehensive assessment parameters.

[0014] Furthermore, the spatiotemporal coordinates are matched based on the original dataset to generate a multi-dimensional carbon flux fusion dataset with a joint index of geographic location and time, including:

[0015] Perform timestamp synchronization on the raw data sets collected by multi-type sensor networks to obtain a time-synchronized data set with a unified time base for all sensor data;

[0016] Based on the time-synchronized dataset, each data point is mapped to the preset unified spatial coordinate system of the restoration area according to the geographic coordinates of the deployment location, generating a mapping dataset with a unified spatiotemporal reference;

[0017] Using the mapped dataset, the geographic location coordinates and synchronized timestamps are used as joint primary keys to correlate and integrate carbon exchange-related parameters from different types of sensors at the same geographic location and timestamp to form a preliminary correlated and integrated dataset.

[0018] In order to address the missing spatiotemporal point data in the preliminary correlation and integration data set due to sensor failure, terrain obstruction, and communication interruption, a complete data set with missing values ​​filled in is obtained based on a unified spatial coordinate system and synchronized timestamps and using valid data from adjacent sensor nodes.

[0019] The complete data set is organized into a data matrix with geographic location coordinates and timestamps as joint indexes, including multidimensional carbon fluxes and related environmental parameters, to generate a multidimensional carbon flux fusion data set.

[0020] Furthermore, based on the fused dataset, the local spatiotemporal dispersion and global correlation thresholds are calculated to identify and mark abnormal data points, including:

[0021] Based on the multidimensional carbon flux fusion dataset, a spatiotemporal sliding window is constructed for each sensor node, and the standard deviation of the data points within the window is calculated to obtain the local spatiotemporal dispersion dataset of each node in the corresponding spatiotemporal window.

[0022] Based on the multidimensional carbon flux fusion dataset, the covariance matrix of all sensor node data is calculated, and the global correlation threshold is calculated based on the covariance matrix to obtain the global correlation threshold that represents the overall correlation of the sensor network data in the entire restoration area.

[0023] Using the local spatiotemporal discreteness dataset and the global correlation threshold, the anomaly decision boundary is generated by comparing the deviation degree between the local spatiotemporal discreteness of each sensor node and the global correlation threshold.

[0024] Based on the abnormal judgment boundary, it is judged whether the data points of each sensor node exceed the boundary. For the node data points that exceed the current boundary, the abnormal state is recorded and marked as abnormal data points.

[0025] Furthermore, based on the abnormal data points, four boundary detection points were determined within the restoration area to form an irregular polygonal monitoring area that changes with the ecological process. The polygonal area was then non-uniformly meshed to generate an adaptive grid cell set covering the monitoring area, including:

[0026] Based on the abnormal data points, the kernel density estimation value of the abnormal data points in the geographical space of the mangrove restoration area is calculated to generate the abnormal point spatial kernel density distribution field covering the entire restoration area;

[0027] Based on the spatial kernel density distribution field of the outlier point, the gradient vector field of the density field in space is calculated, and four spatial position points in the gradient vector field are identified. The four points are determined as boundary detection points. Based on the spatial coordinates of the four boundary detection points, the four points are connected to form a convex hull polygon, that is, the dynamic irregular polygon monitoring area;

[0028] Based on the outlier spatial kernel density distribution field, the outlier density values ​​of each sub-area within the convex hull polygon monitoring area were calculated. At the same time, based on the water level gauge and salinity meter data in the multidimensional carbon flux fusion dataset, data reflecting the hydrological connectivity characteristics within the monitoring area were extracted.

[0029] Based on the outlier density values ​​and hydrological connectivity characteristic data of each sub-region within the polygon, a fine grid size is set in sub-regions with relatively concentrated outliers and key channel areas with more significant hydrological connectivity; a sparse grid size is set in sub-regions with sparse outlier distribution and areas with flat hydrological connectivity, so as to construct a spatial parameter mapping for the non-uniform grid division scheme that adapts to spatial heterogeneity.

[0030] Based on the spatial parameter mapping of the convex hull polygon monitoring area boundary and the non-uniform grid partitioning scheme, the polygon monitoring area is subjected to non-uniform grid cutting to generate a set of adaptive grid cells covering the entire dynamic polygon monitoring area.

[0031] Furthermore, for the grid cell set, the topographic deformation rate, environmental factor gradient change rate, and hydrological inundation frequency of each cell within the tidal cycle are calculated to generate spatially corrected quantitative indicators, including:

[0032] Obtain data recorded by elevation sensors, salinity sensors, temperature sensors, and water level sensors deployed on grid cells in the mangrove restoration area during the tidal cycle;

[0033] Based on elevation sensor data, the surface subsidence of each grid cell within a single tidal cycle is calculated, and based on the surface subsidence and the corresponding tidal cycle time, the terrain deformation rate of each grid cell is calculated. Based on the gradient monitoring values ​​recorded by the salinity sensor and the temperature sensor, the gradient change rate of the environmental factor of each grid cell is calculated. Based on the data recorded by the water level sensor, the average daily inundation frequency of each grid cell is counted to calculate the hydrological inundation frequency value of each grid cell.

[0034] The terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency values ​​are normalized to obtain the corresponding normalized terrain deformation rate parameters, environmental factor change rate parameters and hydrological inundation frequency parameters;

[0035] A weighted fusion calculation is performed on the terrain deformation rate parameters, environmental factor change rate parameters and hydrological inundation frequency parameters to obtain the fusion parameter value of each grid unit. The fusion parameter values ​​are then formed into a correction coefficient matrix that characterizes the spatial heterogeneity of the entire mangrove restoration area, that is, a quantitative indicator of spatial correction.

[0036] Furthermore, the sensor data within the grid cells are extracted and combined with the spatial correction quantitative indicators to compensate for the carbon flux value, thus obtaining the corrected carbon sink dataset, including:

[0037] For each adaptive grid cell, extract the raw carbon flux data recorded by the carbon flux sensor nodes deployed in the cell during the monitoring period;

[0038] Read the quantitative indicators of spatial correction, that is, extract the correction coefficient corresponding to the current processing grid cell from the correction coefficient matrix, and calculate the carbon flux compensation value of the grid cell based on the correction coefficient using the terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency parameters;

[0039] The original carbon flux data were combined with the carbon flux compensation value to generate the grid cell corrected carbon sink data. At the same time, all the corrected carbon sink data were fused to form a corrected carbon sink dataset covering the irregular polygon monitoring area of ​​the entire mangrove restoration area.

[0040] Furthermore, based on the corrected carbon sink dataset, the carbon sink stability index, spatial coverage contribution index, and ecological process representation index of each node during the assessment period were calculated. The three indicators were integrated into a comprehensive assessment parameter. Based on the distribution characteristics of the comprehensive assessment parameters, the prominent nodes were determined as the final deployment points for carbon sink monitoring, including:

[0041] Based on the corrected carbon sink data set, the ratio of the standard deviation to the mean of the carbon sink data of each sensor node within the preset evaluation period is calculated to generate a carbon sink stability index that characterizes the degree of fluctuation of the node data.

[0042] Based on the spatial distribution of the adaptive grid cell set and the node position, the ratio of the sum of the areas of the grid cells associated with each node to the total area of ​​the entire irregular polygon monitoring area is calculated to generate a spatial coverage contribution index that represents the spatial representativeness of the node.

[0043] Based on the hydrological inundation frequency, salinity gradient, and temperature data in the multidimensional carbon flux fusion dataset, the absolute value between each node monitoring data and key ecological process parameters is calculated to generate an ecological process representation index that characterizes the node's ability to reflect ecological processes.

[0044] The carbon sequestration stability index, spatial coverage contribution index and ecological process representation index of each node are normalized and assigned preset weights for weighted summation to generate comprehensive evaluation parameters for each node;

[0045] According to the numerical distribution of comprehensive evaluation parameters of all nodes, the nodes with comprehensive evaluation parameter values ​​greater than the preset threshold are determined as the final layout points for carbon sink monitoring.

[0046] In a second aspect, a computing device includes:

[0047] one or more processors;

[0048] The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the system.

[0049] According to a third aspect, a computer-readable storage medium stores a program, which implements the system when executed by a processor.

[0050] The above solution of the present invention includes at least the following beneficial effects:

[0051] Based on a multi-type sensor network deployed in intertidal wetlands (covering carbon dioxide concentration, temperature and humidity, and soil carbon flux sensors), this system achieves three-dimensional data collection of carbon exchange processes at the atmosphere-vegetation-soil interface, breaking through the coverage gaps of traditional single-point monitoring and fully capturing carbon flux dynamics in complex environments with tidal cycles and topographic changes. Through timestamp synchronization, spatial coordinate system mapping, and missing value filling, multi-source heterogeneous sensor data are integrated into a spatiotemporally aligned, multidimensional carbon flux fusion dataset, addressing the temporal misalignment and spatial discreteness issues inherent in traditional monitoring. Abnormal data identification based on local spatiotemporal discreteness and global correlation thresholds effectively filters out noise data caused by sensor failures and environmental interference. Incorporating spatial correction quantitative indicators such as terrain deformation rate, environmental factor gradient change rate, and hydrological inundation frequency, carbon flux data are adaptively compensated to improve the accuracy of carbon sink calculations and reduce the impact of complex habitats on monitoring results. Dynamic irregular polygonal monitoring areas were constructed based on the distribution of abnormal data points. Using non-uniform grid cutting technology, fine grids were set in areas with dense anomalies and key hydrological channels, while sparse grids were used in sparse areas. This enabled intelligent allocation of monitoring resources, tailored to the spatial heterogeneity of mangrove ecosystems, and improved monitoring efficiency and relevance. The resulting calibrated carbon sequestration dataset and comprehensive assessment parameters can reflect the spatiotemporal evolution of carbon sequestration capacity in mangrove restoration areas in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic diagram of a carbon sequestration data monitoring system for a mangrove restoration area provided by an embodiment of the present invention.

[0053] Figure 2 This is a flow chart of a mangrove restoration area carbon sequestration data monitoring system provided by an embodiment of the present invention, which calculates local spatiotemporal discreteness and global correlation thresholds based on a fused data set, and identifies and marks abnormal data points. DETAILED DESCRIPTION

[0054] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0055] like Figure 1 As shown, an embodiment of the present invention provides a carbon sequestration data monitoring system for a mangrove restoration area, comprising:

[0056] A collection module is used to collect a raw data set of carbon exchange at the atmosphere-vegetation-soil interface based on a multi-type sensor network deployed in the intertidal wetlands of the mangrove restoration area. Specifically, the sensor network includes spatially spaced carbon dioxide concentration sensors, temperature and humidity sensors, and soil carbon flux sensors;

[0057] The data generation module is used to match the spatiotemporal coordinates based on the original dataset to generate a multi-dimensional carbon flux fusion dataset with a joint index of geographic location and time;

[0058] The recognition module is used to calculate the local spatiotemporal discreteness and global correlation threshold based on the fused data set, and identify and mark abnormal data points;

[0059] The gridding module is used to determine four boundary detection points within the restoration area based on abnormal data points, forming an irregular polygonal monitoring area that changes with ecological processes, and performing non-uniform grid cutting on the polygonal area to generate a set of adaptive grid cells covering the monitoring area;

[0060] The indicator generation module is used to calculate the terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency of each grid cell set within the tidal cycle, and generate spatial correction quantitative indicators;

[0061] The compensation and correction module is used to extract sensor data within the grid cell, combine it with the spatial correction quantitative index to compensate for the carbon flux value, and obtain the corrected carbon sink data set;

[0062] The comprehensive assessment module is used to calculate the carbon sink stability index, spatial coverage contribution index and ecological process representation index of each node during the assessment period based on the corrected carbon sink dataset, integrate the three indicators into comprehensive assessment parameters, and determine the prominent nodes as the final layout points for carbon sink monitoring based on the distribution characteristics of the comprehensive assessment parameters.

[0063] In an embodiment of the present invention, a multi-type sensor network deployed across intertidal wetlands in a mangrove restoration area enables comprehensive and dynamic data collection of carbon exchange at the atmosphere-vegetation-soil interface. These sensors (CO2 concentration sensors, temperature and humidity sensors, and soil carbon flux sensors) work together to simultaneously acquire multi-dimensional data on gas concentration, ambient temperature and humidity, and soil carbon flux, comprehensively covering key aspects of the mangrove ecosystem's carbon cycle. Compared to traditional single-point or single-type sensor acquisition methods, this approach avoids the one-sided nature of data collection. For example, monitoring only atmospheric CO2 concentration eliminates the potential for missing critical information on carbon release caused by soil microbial activity. Furthermore, the distributed deployment of the sensor network adapts to the complex and changing intertidal environment, ensuring stable data collection from both inundated areas during high tide and mudflats after low tide, minimizing data loss due to environmental interference.

[0064] The original datasets were matched with spatiotemporal coordinates to generate a multidimensional carbon flux fusion dataset, effectively resolving the spatiotemporal inconsistencies in multi-source data. Timestamp synchronization was used to unify data collected by different sensor types with varying frequencies onto the same time base, avoiding data confusion caused by time misalignment. Geographic coordinate mapping was used to accurately map each sensor's data to a unified spatial coordinate system within the restoration area, ensuring the accuracy of the data's spatial location. Local spatiotemporal dispersion and global correlation thresholds were calculated based on the fused dataset to accurately identify and label anomalous data points.

[0065] Based on the abnormal data points, the monitoring area is dynamically determined and non-uniform grid cutting is performed to improve the pertinence and efficiency of monitoring. By calculating the kernel density estimate of the abnormal data points, it is possible to determine the areas in the mangrove restoration area where carbon flux changes are active or the data fluctuates greatly. Based on this, irregular polygonal monitoring areas are delineated, so that the monitoring range is closely aligned with the actual ecological change hotspots, avoiding the blindness of traditional fixed area monitoring. The non-uniform grid cutting strategy sets fine grids in areas with concentrated abnormal points and complex ecological changes (such as tidal creek intersections and areas with severe vegetation damage) based on the density of abnormal points and hydrological connectivity characteristics in the area to ensure that subtle carbon flux changes can be captured; sparse grids are used in areas with stable data to reduce data redundancy.

[0066] In a preferred embodiment of the present invention, performing spatiotemporal coordinate matching based on the original dataset to generate a multi-dimensional carbon flux fusion dataset with a joint index of geographic location and time may include:

[0067] Perform timestamp synchronization on the raw data sets collected by multi-type sensor networks to obtain a time-synchronized data set with a unified time base for all sensor data;

[0068] Based on the time-synchronized dataset, each data point is mapped to the preset unified spatial coordinate system of the restoration area according to the geographic coordinates of the deployment location, generating a mapping dataset with a unified spatiotemporal reference;

[0069] Using the mapped dataset, the geographic location coordinates and synchronized timestamps are used as joint primary keys to correlate and integrate carbon exchange-related parameters from different types of sensors at the same geographic location and timestamp to form a preliminary correlated and integrated dataset.

[0070] In order to address the missing spatiotemporal point data in the preliminary correlation and integration data set due to sensor failure, terrain obstruction, and communication interruption, a complete data set with missing values ​​filled in is obtained based on a unified spatial coordinate system and synchronized timestamps and using valid data from adjacent sensor nodes.

[0071] The complete data set is organized into a data matrix with geographic location coordinates and timestamps as joint indexes, including multidimensional carbon fluxes and related environmental parameters, to generate a multidimensional carbon flux fusion data set.

[0072] In an embodiment of the present invention, the timestamp information corresponding to each data point is extracted from the raw data set collected by the multi-type sensor network. These timestamps may differ due to the sensor's own clock accuracy and network transmission delay. For example, the timestamps of data recorded by different sensors at the same time may differ by seconds or even minutes. A high-precision time source is selected as a unified time reference, such as the time signal of the Global Positioning System (GPS), which has a time accuracy of nanoseconds. The raw timestamp of each sensor data is compared with the reference time reference to calculate the time deviation. Assume that the raw timestamp of sensor A is , the reference time base is , then the time deviation According to the calculated time deviation, the timestamp of each sensor data is corrected, and the corrected timestamp In this way, the timestamps of all sensor data are unified to a reference time base, resulting in a time-synchronized dataset. The spatial coordinate system used in the restoration area must be clearly defined, such as a common geographic coordinate system (latitude and longitude) or a projected coordinate system (such as the UTM projected coordinate system). The origin, axis directions, and unit of measure parameters of this coordinate system are set as the target coordinate system for data mapping. The actual geographic coordinates of each sensor are obtained from the sensor network deployment records. If the sensor's original coordinates are in longitude and latitude, and the target spatial coordinate system is a projected coordinate system, a coordinate conversion is required. Using the geographic coordinate conversion formula, the original geographic coordinates of the sensor data points are converted to the pre-defined unified spatial coordinate system for the restoration area. For example, the conversion from geographic coordinates (latitude and longitude) to the UTM projected coordinate system involves converting geodetic coordinates to rectangular coordinates and then to UTM projected plane coordinates. This involves the Earth's ellipsoid parameters. Using trigonometric functions and geometric operations, the original longitude and latitude coordinates are converted to coordinates in the target spatial coordinate system, thus generating a mapping dataset with a unified spatiotemporal reference.

[0073] From the mapped dataset, extract the geographic coordinates of each data point (e.g., in a projected coordinate system). 、 The data is then traversed through the entire mapping dataset, associating the data based on the joint primary key. For data points with the same geographic coordinates and timestamp, the sensor type is determined, and all carbon exchange-related parameter data from different types of sensors at the same geographic location and timestamp are found. The carbon exchange-related parameter data from the associated different types of sensors are integrated to form a data record containing multidimensional information. During the integration process, each parameter data is filled into the corresponding position according to the preset data format and field order, thus forming a preliminary associated integrated dataset. The preliminary associated integrated dataset is traversed, and whether the spatiotemporal point data is missing is determined based on whether there are null or invalid values ​​in the data record. Missing data can be identified using the set flag bit and data range. For example, when the carbon dioxide concentration value is -9999, the data is considered missing. Based on the unified spatial coordinate system, the spatial distance between each sensor node with missing data and other sensor nodes is calculated. The Euclidean distance formula can be used. For the coordinates of two sensor nodes in two-dimensional space, several sensor nodes with a closer distance are selected as adjacent nodes. Usually, a distance threshold can be set according to actual conditions, and nodes with a distance less than the threshold are selected as adjacent nodes.

[0074] According to the valid data of the adjacent sensor nodes, a suitable interpolation method is used to fill the missing values, such as the inverse distance weighted interpolation method (IDW), which calculates the missing values ​​by weighted averaging the valid data of the adjacent nodes based on the distance between the adjacent nodes and the node with missing data. The valid data value is , the distance to the missing data node is , then the missing value ,in is the number of adjacent nodes, is the distance weight parameter, usually 2. In this way, the missing values ​​in the preliminary associated integrated data set are filled to obtain a complete data set. Based on the complete data set, the dimension and structure of the data matrix are determined, with the geographic location coordinates and timestamp as the joint index, and the multidimensional carbon flux and related environmental parameters as the elements of the matrix. For example, the rows of the data matrix can represent different geographic location coordinates, and the columns can represent different timestamps. Each element in the matrix corresponds to the multidimensional carbon flux and related environmental parameter data at the geographic location and timestamp. Each data record in the complete data set is accurately filled into the corresponding position of the data matrix according to its geographic location coordinates and timestamp to ensure the integrity and accuracy of the data, thereby generating a multidimensional carbon flux fusion data set with geographic location coordinates and timestamp as the joint index, including multidimensional carbon flux and related environmental parameters.

[0075] Through timestamp synchronization and spatial coordinate mapping, temporal and spatial discrepancies among multi-type sensor data are eliminated, ensuring that all data share a unified spatiotemporal reference. This effectively avoids data errors and confusion caused by spatiotemporal inconsistencies, improving data accuracy and consistency. Data association and integration, using geolocation coordinates and timestamps as joint primary keys, organically combines carbon exchange-related parameters from different sensor types. This integration approach breaks down sensor type boundaries, allowing previously dispersed data to form a cohesive dataset. This allows for a more comprehensive reflection of the relationship between carbon flux and environmental factors, facilitating multidimensional, integrated analysis. For missing spatiotemporal points due to various reasons, valid data from adjacent sensor nodes is used to fill in missing values, ensuring dataset integrity. This complete data set more accurately reflects actual carbon flux changes, avoids analytical bias caused by missing data, and improves data usability and reliability. The complete dataset is organized into a data matrix jointly indexed by geolocation coordinates and timestamps, forming a multidimensional carbon flux fusion dataset. This structured data organization facilitates data storage, management, and querying. The multidimensional carbon flux fusion dataset contains rich information on carbon flux and related environmental parameters, making it possible to conduct more in-depth data analysis and research.

[0076] In a preferred embodiment of the present invention, based on the fused data set, calculating the local spatiotemporal dispersion and the global correlation threshold, and identifying and marking abnormal data points may include:

[0077] Based on the multidimensional carbon flux fusion dataset, a spatiotemporal sliding window is constructed for each sensor node, and the standard deviation of the data points within the window is calculated to obtain the local spatiotemporal dispersion dataset of each node in the corresponding spatiotemporal window.

[0078] Based on the multidimensional carbon flux fusion dataset, the covariance matrix of all sensor node data is calculated, and the global correlation threshold is calculated based on the covariance matrix to obtain the global correlation threshold that represents the overall correlation of the sensor network data in the entire restoration area.

[0079] Using the local spatiotemporal discreteness dataset and the global correlation threshold, the anomaly decision boundary is generated by comparing the deviation degree between the local spatiotemporal discreteness of each sensor node and the global correlation threshold.

[0080] Based on the abnormal judgment boundary, it is judged whether the data points of each sensor node exceed the boundary. For the node data points that exceed the current boundary, the abnormal state is recorded and marked as abnormal data points.

[0081] In the embodiment of the present invention, the size of the spatiotemporal sliding window is set for each sensor node, including the length of the time dimension and the range of the space dimension. For example, the time dimension can be set to be centered at the current time point, and the previous and next time dimensions can be set to be time steps ( Determine based on actual data collection frequency and research needs. For example, if data is collected once per minute, you can set =5, that is, the window covers a 10-minute time period); the spatial dimension is centered on the sensor node, and the radius is set to The circular area ( Determine based on the density of sensor network deployment, such as in denser areas. = 50 meters). Traverse the multidimensional carbon flux fusion data set and, for each sensor node, filter out the data points that fall within the window according to the set spatiotemporal window parameters. These data points contain the multidimensional carbon flux and environmental parameter data related to the sensor node within a specific spatiotemporal range. For each data point within the spatiotemporal sliding window, extract the data features of each dimension, such as carbon flux concentration, temperature, humidity, and other parameters. Assume that there is data points, each containing The features of dimensions can be represented as a data matrix ,in Indicates the The data point dimension feature values. For each dimension , calculate its mean, and calculate each dimension based on the mean The variance of each dimension is squared to obtain the standard deviation. The standard deviations of each dimension are then combined (e.g., averaged or weighted, with weights set based on the importance of each dimension) to obtain the local spatiotemporal dispersion value of the sensor node in the corresponding spatiotemporal window. This process is repeated to calculate the local spatiotemporal dispersion of each sensor node in its corresponding spatiotemporal window, ultimately forming a local spatiotemporal dispersion dataset.

[0082] Extract all data from all sensor nodes from the multidimensional carbon flux fusion dataset and construct a large data matrix containing data from all nodes, all time points and all dimensions. ,in, is the number of sensor nodes, is the number of time points, is the number of data dimensions, Indicates the Space nodes, Indicates the Time points, the data matrix For each dimension, , calculate its mean and standard deviation. Based on the standardized data, calculate the covariance matrix ,in Indicates the Dimensions and The covariance between the dimensions is analyzed. The calculated covariance matrix can be analyzed using methods such as principal component analysis (PCA) to extract key features and identify key information from the covariance matrix. A global correlation threshold calculation method can be set based on the distribution of the covariance matrix's eigenvalues, the actual data characteristics, or research needs. For example, the eigenvalue corresponding to a certain percentile (e.g., the 95th percentile) of the covariance matrix's eigenvalues ​​can be selected as the global correlation threshold. The local spatiotemporal dispersion dataset is traversed. For each sensor node's local spatiotemporal dispersion value, the deviation from the global correlation threshold is calculated. The deviation can be calculated as a ratio or a difference. Based on the distribution of the deviation, combined with the actual data characteristics and research needs, the parameters for the anomaly decision boundary are determined. For example, a data point can be considered anomaly when the deviation exceeds a certain multiple (e.g., 2) of the global correlation threshold, or when the deviation exceeds a preset absolute difference (e.g., 5). These parameter settings generate an anomaly decision boundary for determining whether a data point is anomaly. We then traverse the multidimensional carbon flux fusion dataset again. For each sensor node data point, we determine whether it exceeds the boundary based on the local spatiotemporal discreteness of the node to which it belongs and the anomaly determination boundary. For node data points that exceed the current anomaly determination boundary, we record their abnormal status and mark them as abnormal data points by adding a flag field.

[0083] By calculating local spatiotemporal dispersion and global correlation thresholds, anomalous data points are identified and marked, effectively eliminating erroneous data caused by sensor failures and environmental interference. This ensures that the retained data more accurately reflects actual carbon flux variations and environmental conditions. Anomalous data can seriously impact the accuracy of data analysis results. By marking and removing anomalous data points, analyses of spatiotemporal carbon flux variations and correlations between environmental factors and carbon fluxes are based on more accurate data, resulting in more scientific and reliable conclusions and patterns. This avoids erroneous analysis results and misleading conclusions caused by anomalous data. The process of identifying anomalous data points allows for the timely identification of problematic nodes in the sensor network. For example, sensors with frequent anomalous data may be faulty. This helps operations personnel quickly locate faulty nodes and promptly repair or replace them, optimizing the operational status of the sensor network, improving the efficiency and stability of the entire monitoring system, and ensuring continuous and accurate monitoring of carbon flux and environmental information in the remediation area. Processing anomalous data consumes additional computing resources and storage capacity during data storage and analysis. By marking anomalous data points and performing targeted processing (such as removal or correction), we can reduce the amount of invalid data to be processed, lower data storage costs and computing resource consumption, improve data processing efficiency, and make the data processing process more efficient and economical. Anomalous data point identification technology ensures the quality of multidimensional carbon flux fusion datasets, providing reliable data support for decision-making processes such as carbon emission management and ecological restoration plan development.

[0084] In a preferred embodiment of the present invention, four boundary detection points are determined within the restoration area based on the abnormal data points to form an irregular polygonal monitoring area that changes with the ecological process. The polygonal area is then subjected to non-uniform grid cutting to generate an adaptive grid unit set covering the monitoring area, which may include:

[0085] Based on the abnormal data points, the kernel density estimation value of the abnormal data points in the geographical space of the mangrove restoration area is calculated to generate the abnormal point spatial kernel density distribution field covering the entire restoration area;

[0086] Based on the spatial kernel density distribution field of the outlier point, the gradient vector field of the density field in space is calculated, and four spatial position points in the gradient vector field are identified. The four points are determined as boundary detection points. Based on the spatial coordinates of the four boundary detection points, the four points are connected to form a convex hull polygon, that is, the dynamic irregular polygon monitoring area;

[0087] Based on the outlier spatial kernel density distribution field, the outlier density values ​​of each sub-area within the convex hull polygon monitoring area were calculated. At the same time, based on the water level gauge and salinity meter data in the multidimensional carbon flux fusion dataset, data reflecting the hydrological connectivity characteristics within the monitoring area were extracted.

[0088] Based on the outlier density values ​​and hydrological connectivity characteristic data of each sub-region within the polygon, a fine grid size is set in sub-regions with relatively concentrated outliers and key channel areas with more significant hydrological connectivity; a sparse grid size is set in sub-regions with sparse outlier distribution and areas with flat hydrological connectivity, so as to construct a spatial parameter mapping for the non-uniform grid division scheme that adapts to spatial heterogeneity.

[0089] Based on the spatial parameter mapping of the convex hull polygon monitoring area boundary and the non-uniform grid partitioning scheme, the polygon monitoring area is subjected to non-uniform grid cutting to generate a set of adaptive grid cells covering the entire dynamic polygon monitoring area.

[0090] In an embodiment of the present invention, the geographic space of the mangrove restoration area is discretized and divided into regular small areas (e.g., a grid), each with a well-defined geographic coordinate range. For each outlier data point, a neighborhood with a certain radius is selected around it (this radius can be set based on practical circumstances and experience, for example, the size of the mangrove restoration area, the density of data distribution, and other factors). A kernel function (common kernel functions include Gaussian and uniform kernel functions) is then used to calculate the contribution of the outlier data point to each of the small areas within the neighborhood. For example, the closer the small area is to the outlier data point, the greater the impact and the higher the contribution value; the farther away, the contribution value decreases exponentially. Specifically, the contribution value is calculated by substituting the distance between the small area center and the outlier data point into the Gaussian kernel function formula. This process is repeated for all outlier data points, and the contribution values ​​obtained by each small area from all outlier data points are accumulated to obtain the kernel density estimate for that small area. After traversing all small areas in the entire mangrove restoration area and completing the calculation of the kernel density estimates for all small areas, a spatial kernel density distribution field of outliers covering the entire restoration area is generated. This distribution field intuitively shows the spatial distribution density of outlier data points in the restoration area.

[0091] For each small area in the outlier spatial kernel density distribution field, calculate the rate of change in density, or gradient, along three spatial dimensions (assuming the X, Y, and Z directions; in two-dimensional geographic space, the X and Y directions are primarily considered). Taking the X direction as an example, calculate the difference between the kernel density estimate for the small area and the adjacent X-direction small area, and divide it by the distance between them (the grid side length) to obtain the rate of change in the X direction. Similarly, calculate the rate of change in the Y direction. This generates a two-dimensional gradient vector for each small area, and the gradient vectors of all small areas together constitute the spatial gradient vector field of the density field. Within the gradient vector field, identify locations with significant and representative gradient changes. A threshold can be set to select locations with large gradient moduli (i.e., the length of the gradient vector). From these candidate points, select four points that best cover the boundary of the area where the outlier data points are concentrated as boundary detection points. The principle for selection is that the four points should be as dispersed as possible in space and roughly outline the area where the outlier data points are concentrated. After determining the spatial coordinates of the four boundary detection points, a convex hull algorithm (such as the Graham scan method or the Andrew monotone chain method) is used to connect these four points to form a convex hull polygon. This convex hull polygon is the dynamic irregular polygon monitoring area, which tightly surrounds the area where anomalous data points are concentrated. As ecological processes change and the distribution of anomalous data points shifts, its shape and range will adjust accordingly.

[0092] The convex hull polygon monitoring area is further subdivided into multiple subareas (this division can be performed according to specific rules, such as dividing into smaller grids of equal area). For each subarea, the number of anomalous data points falling within the subarea is counted and then divided by the subarea's area to obtain the anomalous point density value for that subarea. Data and real-time data from water level and salinity meters at each monitoring point within the monitoring area are extracted from the multidimensional carbon flux fusion dataset. Temporal trends in water level data, such as the amplitude and frequency of water level fluctuations, are analyzed; the spatial distribution and temporal variation patterns of salinity data are analyzed. Through this data analysis, the correlation between water level and salinity at different locations within the monitoring area is determined, thereby extracting data reflecting the hydrological connectivity characteristics within the monitoring area. For example, if the water level trends at two locations are highly consistent and the salinity difference is small, the hydrological connectivity between the two locations is good; otherwise, the connectivity is poor.

[0093] Taking into account the outlier density and hydrological connectivity characteristics of each subregion, a higher outlier density indicates that ecological processes in these areas may be more complex and require more detailed monitoring. Therefore, a smaller grid size is set to more accurately capture changes in outlier data points. For key channel areas with significant hydrological connectivity, where frequent water and material exchange have a greater impact on the ecosystem, a smaller grid size is also set to enhance monitoring. Conversely, for subregions with sparse outlier distribution, where ecological processes are relatively stable and do not require overly detailed monitoring, a larger grid size is set to reduce unnecessary monitoring data. For areas with mild hydrological connectivity, where water and material exchange are relatively infrequent and have less impact on the ecosystem, a larger grid size is also set. Based on the spatial parameter mapping of the convex hull polygon monitoring area boundary and the non-uniform grid partitioning scheme, the polygon monitoring area is subjected to non-uniform meshing to generate an adaptive set of grid cells covering the entire dynamic polygon monitoring area. The range of the meshing is determined by the convex hull polygon monitoring area boundary. Based on the spatial parameter mapping of the non-uniform grid partitioning scheme, the subregions are meshed according to the set grid size. Starting from the monitoring area's boundary, the grid is divided inwards according to the set grid size, ensuring that each sub-area is divided into grid cells of the appropriate size. As the division progresses, all generated grid cells are integrated to form an adaptive set of grid cells covering the entire dynamic polygon monitoring area. The size and distribution of these grid cells are optimized based on the distribution of outliers and hydrological connectivity within the monitoring area to better meet monitoring needs.

[0094] Using kernel density estimation and gradient vector field analysis based on outlier data points, the monitoring area boundaries were precisely determined, closely aligning the monitoring area with areas experiencing active ecological changes. Furthermore, non-uniform gridding was implemented based on outlier density and hydrological connectivity. Fine grids were set in key areas, enabling more detailed capture of ecological data changes. This improved monitoring accuracy compared to uniform gridding and facilitated a more accurate understanding of the ecological status of the mangrove restoration area. Sparse grids were set in areas with sparse outlier distribution and flat hydrological connectivity, avoiding unnecessary over-monitoring and reducing the waste of monitoring data and computing resources. This adaptive gridding approach enabled more rational allocation of monitoring resources to key areas, improving resource efficiency and reducing monitoring costs. The generated dynamic irregular polygonal monitoring area and adaptive grid cell set can adapt to changing ecological processes. The shape and extent of the monitoring area will change accordingly as the distribution of outlier data points changes. The gridding also adapts as ecosystem characteristics such as hydrological connectivity change, maintaining effective monitoring of the ecological status of the mangrove restoration area and providing timely and accurate data support for ecological restoration and management. The mangrove restoration area was analyzed and monitored from multiple perspectives, taking into account various ecological factors such as the distribution of outliers and hydrological connectivity.

[0095] In a preferred embodiment of the present invention, for a set of grid cells, the topographic deformation rate, environmental factor gradient change rate, and hydrological inundation frequency of each cell within a tidal cycle are calculated to generate spatially corrected quantitative indicators, which may include:

[0096] Obtain data recorded by elevation sensors, salinity sensors, temperature sensors, and water level sensors deployed on grid cells in the mangrove restoration area during the tidal cycle;

[0097] Based on elevation sensor data, the surface subsidence of each grid cell within a single tidal cycle is calculated, and based on the surface subsidence and the corresponding tidal cycle time, the terrain deformation rate of each grid cell is calculated. Based on the gradient monitoring values ​​recorded by the salinity sensor and the temperature sensor, the gradient change rate of the environmental factor of each grid cell is calculated. Based on the data recorded by the water level sensor, the average daily inundation frequency of each grid cell is counted to calculate the hydrological inundation frequency value of each grid cell.

[0098] The terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency values ​​are normalized to obtain the corresponding normalized terrain deformation rate parameters, environmental factor change rate parameters and hydrological inundation frequency parameters;

[0099] A weighted fusion calculation is performed on the terrain deformation rate parameters, environmental factor change rate parameters and hydrological inundation frequency parameters to obtain the fusion parameter value of each grid unit. The fusion parameter values ​​are then formed into a correction coefficient matrix that characterizes the spatial heterogeneity of the entire mangrove restoration area, that is, a quantitative indicator of spatial correction.

[0100] In an embodiment of the present invention, a set of adaptive grid units that have been divided into mangrove restoration areas is clearly defined, and each grid unit has a unique identifier and spatial location. The installation location and operating status of the elevation sensors, salinity sensors, temperature sensors, and water level sensors deployed on each grid unit are checked to ensure that the sensors are working properly and the data acquisition function is turned on. A complete tidal cycle time range is determined. The tidal cycle can usually be determined based on local tidal patterns (such as obtained through ocean tide tables, data statistics, etc.), generally a semi-diurnal tide (about 12 hours and 25 minutes) or a diurnal tide (about 24 hours and 50 minutes). During the selected tidal cycle, the data generated by each sensor is recorded in real time through the data acquisition system. For each grid unit, the data collected by the corresponding sensor at a certain time interval (such as every minute, every 5 minutes, etc., which can be set according to sensor performance and monitoring requirements) during the entire tidal cycle is stored to form a tidal cycle data sequence for each sensor in each grid unit.

[0101] Calculate terrain deformation rate:

[0102] For each grid cell's elevation sensor data sequence, select the elevation at the beginning of the tidal cycle as the initial elevation and the elevation at the end of the tidal cycle as the final elevation. Subtract the final elevation from the initial elevation to obtain the surface subsidence for that grid cell during a single tidal cycle (a negative result indicates surface uplift). Given the total duration of the tidal cycle (in hours or minutes), divide the surface subsidence by the tidal cycle duration to obtain the terrain deformation per unit time for each grid cell, i.e., the terrain deformation rate (e.g., centimeters per hour).

[0103] Calculate the rate of change of environmental factor gradient:

[0104] Salinity sensors and temperature sensors are usually placed at multiple monitoring points in different locations within a grid unit to obtain gradient monitoring values. For salinity data, the salinity difference between adjacent monitoring points is calculated and then divided by the spatial distance between the two points to obtain the spatial gradient change value of salinity. Similarly, the gradient change value of temperature between adjacent monitoring points is calculated. The gradient change values ​​of salinity and temperature are comprehensively processed. For example, a weighted average method can be used (the weights can be set according to research needs or the importance of salinity and temperature to mangrove ecology) to combine the gradient change values ​​of the two into a single value to obtain the gradient change rate of the environmental factor for each grid unit.

[0105] Calculate the hydrological inundation frequency value:

[0106] Analyze the data recorded by the water level sensor during the tidal cycle and set a water level threshold. When the water level exceeds this threshold, the grid cell is considered to be flooded. Count the total time the grid cell water level is above the threshold during the entire tidal cycle and divide this total time by the tidal cycle length to obtain the flooded percentage of the grid cell during that tidal cycle. Assuming there are multiple tidal cycles in a day, add the flooded percentages for each tidal cycle to obtain the average daily flooding frequency for each grid cell, which is used as the hydrological flooding frequency value.

[0107] Collect data on terrain deformation rate, environmental factor gradient change rate, and hydrological inundation frequency for all grid cells, and identify the maximum and minimum values ​​within these three data types. For terrain deformation rate data, use the normalization formula (current terrain deformation rate - minimum terrain deformation rate) / (maximum terrain deformation rate - minimum terrain deformation rate) for each grid cell's terrain deformation rate to calculate the normalized terrain deformation rate parameter, mapping its numerical range to the [0, 1] interval. Similarly, use the same normalization formula for the environmental factor gradient change rate and hydrological inundation frequency values ​​to calculate the corresponding normalized environmental factor change rate parameter and hydrological inundation frequency parameter, ensuring that all three data types are on the same numerical scale for ease of subsequent processing. Based on the needs of mangrove ecological research and restoration management, assign weights to the terrain deformation rate parameter, environmental factor change rate parameter, and hydrological inundation frequency parameter (for example, if terrain deformation is considered to have a greater impact on mangrove ecology, a higher weight may be assigned). For each grid cell, the normalized terrain deformation rate parameter is multiplied by its corresponding weight, the environmental factor change rate parameter is multiplied by its weight, and the hydrological inundation frequency parameter is multiplied by its weight. This weighted summation yields the fusion parameter value for that grid cell. The fusion parameter values ​​for all grid cells are arranged in order of their spatial position to form a matrix. Each element in this matrix corresponds to a fusion parameter value for a grid cell. This matrix constitutes the correction coefficient matrix that characterizes the spatial heterogeneity of the entire mangrove restoration area, serving as a quantitative indicator of spatial correction.

[0108] By separately calculating the rate of topographic deformation, the rate of change of environmental factor gradients, and the frequency of hydrological inundation, and then performing weighted fusion, we comprehensively and accurately quantify the ecological changes in the mangrove restoration area over the tidal cycle from multiple dimensions. Compared to single-indicator monitoring, this method provides a more detailed picture of ecosystem dynamics, providing rich and accurate data support for in-depth research on mangrove ecology. The resulting spatially corrected quantitative indicators are presented in matrix form, visually demonstrating the spatial heterogeneity between different grid cells across the entire mangrove restoration area. This matrix allows managers and researchers to quickly locate areas of significant ecological change and understand the comprehensive characteristics of ecological factors in each region, assisting in the development of targeted ecological restoration and management strategies. Normalization of each ecological indicator eliminates differences in dimension and numerical range between indicators, making data from different grid cells and different ecological factors comparable. Weighted fusion calculations based on this foundation further enhance the comprehensiveness and scientific nature of the data, improving the quality and application value of the monitoring data. This method, which collects data and calculates indicators based on the tidal cycle, can provide real-time insights into the dynamic characteristics of the mangrove ecosystem as it changes with tides. The generated spatial correction quantitative indicators can be used to correct subsequent monitoring data, improve the accuracy of monitoring data, provide strong support for dynamic evaluation, ecological early warning and scientific decision-making of mangrove restoration projects, and promote the sustainable development of mangrove ecosystems.

[0109] In a preferred embodiment of the present invention, extracting sensor data within a grid cell and performing carbon flux compensation in combination with a spatial correction quantitative index to obtain a corrected carbon sink dataset may include:

[0110] For each adaptive grid cell, extract the raw carbon flux data recorded by the carbon flux sensor nodes deployed in the cell during the monitoring period;

[0111] Read the quantitative indicators of spatial correction, that is, extract the correction coefficient corresponding to the current processing grid cell from the correction coefficient matrix, and calculate the carbon flux compensation value of the grid cell based on the correction coefficient using the terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency parameters;

[0112] The original carbon flux data were combined with the carbon flux compensation value to generate the grid cell corrected carbon sink data. At the same time, all the corrected carbon sink data were fused to form a corrected carbon sink dataset covering the irregular polygon monitoring area of ​​the entire mangrove restoration area.

[0113] In an embodiment of the present invention, a set of adaptive grid cells that have been divided into mangrove restoration areas is identified, each with a unique number and precise spatial location information. For each grid cell, the number and specific locations of carbon flux sensor nodes deployed within the cell are determined. These sensor nodes are typically distributed within the grid cell according to specific layout rules to ensure comprehensive monitoring of the carbon flux in the area. A monitoring period is set, which can be determined based on research objectives and actual needs, such as one day, one week, or one month. During the specified monitoring period, the carbon flux sensor nodes collect carbon flux data at a preset sampling frequency (e.g., once per minute, once per hour, etc.) and store the data locally or transmit it to a data center. From the data storage location or data center, a data query and extraction program is used to extract data collected by all carbon flux sensor nodes within each grid cell during the entire monitoring period. The extracted data is sorted according to the sensor node number and acquisition time sequence to form a raw carbon flux data sequence for each grid cell.

[0114] Find the correction coefficient matrix that stores the spatial correction quantitative index. This matrix is ​​generated based on the terrain deformation rate, environmental factor gradient change rate, and hydrological inundation frequency parameter for each grid cell, as calculated in the previous calculation steps. Each element in the matrix corresponds to a comprehensive correction coefficient for a grid cell. Based on the number or spatial location of the currently processed grid cell, accurately locate the corresponding element in the correction coefficient matrix, which is the correction coefficient for that grid cell. Each grid cell has a corresponding terrain deformation rate parameter, environmental factor change rate parameter, and hydrological inundation frequency parameter. These parameters are normalized data obtained from the previous calculation of the spatial correction quantitative index. The carbon flux compensation value is determined by a calculation formula that comprehensively considers these three parameters and the correction coefficient. For example, a linear weighting method can be used: multiply the terrain deformation rate parameter by a weight related to the terrain's impact on carbon flux, the environmental factor change rate parameter by the corresponding environmental factor weight, and the hydrological inundation frequency parameter by the hydrological weight. These three products are then added together and multiplied by the correction coefficient for that grid cell to obtain the carbon flux compensation value for that grid cell.

[0115] For each grid cell, each data point in its original carbon flux data sequence during the monitoring period is added to the calculated carbon flux compensation value (if the compensation value is negative, a subtraction operation is performed). In this way, the original carbon flux data is corrected to obtain the corrected carbon sequestration data sequence for each grid cell. After processing all grid cells in sequence, the corrected carbon sequestration data of all grid cells are arranged and integrated according to the spatial position order of the grid cells. These data can be stored in a data file or database table so that each data record corresponds to the corrected carbon sequestration data of a grid cell, and the storage order of the data is consistent with the spatial distribution order of the grid cells in the irregular polygon monitoring area of ​​the mangrove restoration area. The final data set is the corrected carbon sequestration data set covering the entire irregular polygon monitoring area of ​​the mangrove restoration area. This data set can more accurately reflect the actual carbon sequestration situation in the mangrove restoration area.

[0116] The original carbon flux data were compensated and corrected using spatially corrected quantitative indicators derived from multiple factors, including topographic deformation rate, environmental gradient change rate, and hydrological inundation frequency. This fully accounts for the impact of multiple environmental factors on carbon flux within mangrove ecosystems. Compared to using only the original carbon flux data, the corrected carbon sequestration dataset more accurately reflects the actual carbon sequestration capacity of mangroves, reduces data bias caused by environmental interference, and provides a reliable data foundation for carbon sequestration research. By closely linking carbon flux data with spatial heterogeneity, the corrected carbon sequestration data contain more ecological information. Researchers can intuitively understand the impact of regional topographic changes, environmental gradients, and hydrological conditions on carbon sequestration, facilitate in-depth exploration of the intrinsic links between carbon cycling and other ecological processes within mangrove ecosystems, and provide strong data support for uncovering the carbon sequestration mechanisms of mangroves. An accurate corrected carbon sequestration dataset provides a more scientific basis for mangrove ecosystem assessments. Based on this dataset, managers can more accurately evaluate the effectiveness of mangrove restoration projects, identify trends in carbon sequestration capacity, and formulate more effective ecological protection and restoration strategies. For example, in areas with low carbon sequestration, we can focus on analyzing the influencing factors and take targeted measures to improve the ecological environment, enhance the carbon sequestration function of mangroves, and promote the sustainable development of the ecosystem. A unified data correction method allows for better comparability of carbon sequestration data across different grid cells and monitoring periods.

[0117] In a preferred embodiment of the present invention, based on the corrected carbon sequestration data set, the carbon sequestration stability index, spatial coverage contribution index, and ecological process representation index of each node during the assessment period are calculated, and the three indicators are integrated into a comprehensive assessment parameter. Based on the distribution characteristics of the comprehensive assessment parameters, the prominent nodes are determined as the final deployment points for carbon sequestration monitoring, which may include:

[0118] Based on the corrected carbon sink data set, the ratio of the standard deviation to the mean of the carbon sink data of each sensor node within the preset evaluation period is calculated to generate a carbon sink stability index that characterizes the degree of fluctuation of the node data.

[0119] Based on the spatial distribution of the adaptive grid cell set and the node position, the ratio of the sum of the areas of the grid cells associated with each node to the total area of ​​the entire irregular polygon monitoring area is calculated to generate a spatial coverage contribution index that represents the spatial representativeness of the node.

[0120] Based on the hydrological inundation frequency, salinity gradient, and temperature data in the multidimensional carbon flux fusion dataset, the absolute value between each node monitoring data and key ecological process parameters is calculated to generate an ecological process representation index that characterizes the node's ability to reflect ecological processes.

[0121] The carbon sequestration stability index, spatial coverage contribution index and ecological process representation index of each node are normalized and assigned preset weights for weighted summation to generate comprehensive evaluation parameters for each node;

[0122] According to the numerical distribution of comprehensive evaluation parameters of all nodes, the nodes with comprehensive evaluation parameter values ​​greater than the preset threshold are determined as the final layout points for carbon sink monitoring.

[0123] In an embodiment of the present invention, a corrected carbon sequestration dataset is determined, comprising corrected carbon sequestration data for all adaptive grid cells in a mangrove restoration area, each grid cell being associated with a sensor node deployed therein. A preset evaluation period is determined, which can be set based on research needs and actual monitoring conditions, such as a month or a quarter. For each sensor node, all carbon sequestration data recorded by the node during the evaluation period is extracted from the corrected carbon sequestration dataset to form a carbon sequestration data sequence for the node. The mean of the data sequence is calculated, all carbon sequestration data in the data sequence are summed and then divided by the number of data to obtain the average carbon sequestration value for the node during the evaluation period. The standard deviation of the data sequence is calculated by first calculating the difference between each data point and the mean, squaring these differences, averaging these squared values, and finally taking the square root of the average to obtain the standard deviation. The standard deviation reflects the degree of data dispersion; a larger standard deviation indicates more severe data fluctuations. The calculated standard deviation is divided by the mean, and the resulting ratio is the carbon sequestration stability indicator for the sensor node. The smaller the value of this indicator, the lower the fluctuation of the carbon sequestration data of the node during the assessment period and the better the data stability; conversely, the data fluctuates greatly and the stability is poor.

[0124] Determine the spatial distribution of the adaptive grid unit set. Each grid unit has precise boundary coordinates. The area of ​​each grid unit can be calculated through geographic information system (GIS) technology. At the same time, determine the grid unit where each sensor node is located. A node may be associated with multiple grid units. For each sensor node, count the areas of the grid units it is associated with, add the areas of these grid units one by one, and get the total area of ​​the grid units associated with the node. Calculate the total area of ​​the entire irregular polygon monitoring area. Use GIS technology to process the polygon boundary and calculate the geographic space area it covers. Divide the total area of ​​the grid units associated with each node by the total area of ​​the entire irregular polygon monitoring area, and then multiply by 100%. The percentage value obtained is the spatial coverage contribution index of the node. The higher the index, the larger the spatial range represented by the node and the stronger the spatial representativeness in the monitoring area.

[0125] Hydrological inundation frequency, salinity gradient, and temperature data were extracted from the multidimensional carbon flux fusion dataset. These data record ecological and environmental information at different locations and times within the monitoring area. Furthermore, the corresponding ecological data recorded by each sensor node during the evaluation period were obtained. For hydrological inundation frequency, the difference between the hydrological inundation frequency data monitored by each node and the average hydrological inundation frequency for the entire monitoring area was calculated. The absolute value of this difference was taken. This absolute value reflects the degree of deviation between the hydrological inundation frequency monitored by the node and the overall average. The smaller the deviation, the closer the node's reflection of the hydrological inundation process is to the overall level. For salinity gradient, the absolute value of the difference between the salinity data monitored by the node and the salinity gradient trend of the surrounding area was calculated. The absolute value of this difference also reflects the ability of the node's salinity data to represent regional salinity changes. The smaller the value, the more consistent the node's monitoring data is with the regional salinity gradient. For temperature data, the absolute value of the difference between the temperature data monitored by the calculation node and the temperature change trend of the entire monitoring area is used. This absolute value reflects the degree to which the node reflects the temperature ecological process. The smaller the value, the more the temperature data monitored by the node can represent the regional temperature changes. The above three absolute values ​​are dimensionlessly processed and then comprehensively calculated. For example, the absolute values ​​of the hydrological inundation frequency, salinity gradient, and temperature data can be divided by the maximum value of each indicator to convert them into dimensionless relative deviation values. These standardized values ​​are then added together, and the resulting sum is the ecological process representation index of the node. The smaller the value of this indicator, the higher the degree of fit between the node monitoring data and the key ecological process parameters, and the stronger the ability to reflect the ecological process.

[0126] Data on the carbon sink stability index, spatial coverage contribution index, and ecological process representation index were collected from all sensor nodes, and the maximum and minimum values ​​for each of these three categories were identified. For the carbon sink stability index, the normalization formula (current index value - minimum index value) / (maximum index value - minimum index value) was used for each node to calculate the normalized carbon sink stability parameter, which was then mapped to the interval [0, 1]. Similarly, the spatial coverage contribution index and ecological process representation index were normalized to obtain the corresponding normalized spatial coverage contribution parameter and ecological process representation parameter. Preset weights were assigned to the normalized carbon sink stability parameter, spatial coverage contribution parameter, and ecological process representation parameter based on the research focus of mangrove carbon sink monitoring and the importance of each indicator to carbon sink monitoring. (For example, if carbon sink stability is considered to have the greatest impact on monitoring results, a higher weight could be assigned.) For each node, the normalized carbon sink stability parameter is multiplied by its corresponding weight, the spatial coverage contribution parameter is multiplied by its weight, and the ecological process representation parameter is multiplied by its weight. This weighted summation yields the node's comprehensive assessment parameter. This comprehensive assessment parameter comprehensively considers the node's performance in multiple aspects, including data stability, spatial representativeness, and ecological process reflection.

[0127] The comprehensive evaluation parameter values ​​of all sensor nodes were analyzed to observe their distribution range and concentration trends. A reasonable preset threshold was set based on research experience and actual monitoring needs. The comprehensive evaluation parameter values ​​of each node were compared with the preset threshold, and nodes with comprehensive evaluation parameter values ​​exceeding the threshold were screened out. These selected nodes demonstrated outstanding performance in carbon sink stability, spatial coverage contribution, and ecological process representation, and were selected as the final carbon sink monitoring locations.

[0128] By calculating multiple metrics, including carbon sink stability, spatial coverage contribution, and ecological process representation, the performance of sensor nodes was comprehensively evaluated, avoiding blind placement of monitoring nodes. The final deployment points determined ensure more scientific and reasonable coverage of the mangrove restoration area, ensuring broad spatial representation and ecological relevance of the monitoring data, and improving the overall rationality and effectiveness of the monitoring node layout. The carbon sink stability metric selects nodes with minimal data fluctuations, ensuring the reliability of the monitoring data; the ecological process representation metric ensures that the nodes accurately reflect key ecological processes, enhancing the data's ability to interpret ecological phenomena. The data collected from the final deployment points determined after comprehensive evaluation is of higher quality and greater reliability, providing more accurate data support for mangrove carbon sequestration research. Determining the final deployment points based on comprehensive evaluation parameters avoids unnecessary deployment of nodes, reducing data redundancy and resource waste. By concentrating limited monitoring resources on key nodes, while ensuring effective monitoring, it reduces costs, improves efficiency, and makes monitoring more targeted and efficient.

[0129] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, executes the system described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0130] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the system described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0131] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A carbon sequestration data monitoring system for mangrove restoration areas, characterized by: include: The acquisition module is used to collect raw data sets of carbon exchange at the atmosphere-vegetation-soil interface based on a multi-type sensor network deployed in the intertidal wetlands of the mangrove restoration area; The data generation module is used to perform spatiotemporal coordinate matching based on the original data set to generate a multidimensional carbon flux fusion data set with a joint index of geographic location and time, specifically including: performing timestamp synchronization processing on the original data set collected by the multi-type sensor network to obtain a time-synchronized data set with a unified time base for all sensor data; based on the time-synchronized data set, mapping each data point to a preset unified spatial coordinate system of the restoration area according to the geographic coordinates of the deployment location to generate a mapping data set with a unified spatiotemporal base; using the mapping data set, with geographic location coordinates and synchronized timestamps as joint primary keys, to associate and integrate carbon exchange related parameters from different types of sensors at the same geographic location and the same timestamp to form a preliminary associated integrated data set; for the missing spatiotemporal point data in the preliminary associated integrated data set caused by sensor failure, terrain obstruction and communication interruption, based on the unified spatial coordinate system and synchronized timestamps, using the valid data of adjacent sensor nodes to obtain a complete data set with the missing values ​​filled; organizing the complete data set into a data matrix with geographic location coordinates and timestamps as joint indexes, including multidimensional carbon flux and related environmental parameters, to generate a multidimensional carbon flux fusion data set; The recognition module is used to calculate the local spatiotemporal discreteness and global correlation threshold based on the fused data set, and identify and mark abnormal data points; The gridding module is used to determine four boundary detection points in the restoration area based on the abnormal data points, form an irregular polygonal monitoring area that changes with the ecological process, and perform non-uniform grid cutting on the polygonal area to generate an adaptive grid unit set covering the monitoring area. Specifically, it includes: based on the abnormal data points, calculating the kernel density estimate of the abnormal data points in the geographical space of the mangrove restoration area, and generating the abnormal point spatial kernel density distribution field covering the entire restoration area; based on the abnormal point spatial kernel density distribution field, calculating the gradient vector field of the density field in space, and identifying four spatial position points in the gradient vector field, determining the four points as boundary detection points, and based on the spatial coordinates of the four boundary detection points, connecting the four points to form a convex hull polygon, that is, a dynamic irregular polygonal monitoring area; based on the abnormal point spatial kernel density distribution field, calculating the convex hull polygon The density of abnormal points in each sub-region within the polygon monitoring area is calculated. At the same time, based on the water level gauge and salinometer data in the multidimensional carbon flux fusion dataset, data reflecting the hydrological connectivity characteristics within the monitoring area are extracted; based on the density of abnormal points and hydrological connectivity characteristic data in each sub-region within the polygon, a fine grid size is set in the sub-regions where abnormal points are relatively concentrated and the key channel areas with more significant hydrological connectivity; in the sub-regions where the abnormal points are relatively sparsely distributed and the areas with flat hydrological connectivity, a sparse grid size is set to construct a spatial parameter mapping of a non-uniform grid division scheme that adapts to spatial heterogeneity; based on the convex hull polygon monitoring area boundary and the spatial parameter mapping of the non-uniform grid division scheme, the polygon monitoring area is non-uniformly gridded to generate an adaptive grid unit set covering the entire dynamic polygon monitoring area; The indicator generation module is used to calculate the terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency of each grid cell set within the tidal cycle, and generate spatial correction quantitative indicators; The compensation and correction module is used to extract sensor data within the grid cell, combine it with the spatial correction quantitative index to compensate for the carbon flux value, and obtain the corrected carbon sink data set; The comprehensive assessment module is used to calculate the carbon sink stability index, spatial coverage contribution index and ecological process representation index of each node during the assessment period based on the corrected carbon sink dataset, integrate the three indicators into comprehensive assessment parameters, and determine the prominent nodes as the final layout points for carbon sink monitoring based on the distribution characteristics of the comprehensive assessment parameters.

2. The mangrove restoration area carbon sequestration data monitoring system according to claim 1 is characterized in that: The sensor network includes spatially spaced carbon dioxide concentration sensors, temperature and humidity sensors, and soil carbon flux sensors.

3. The mangrove restoration area carbon sequestration data monitoring system according to claim 2 is characterized in that: Based on the fused dataset, the local spatiotemporal dispersion and global correlation threshold are calculated to identify and mark abnormal data points, including: Based on the multidimensional carbon flux fusion dataset, a spatiotemporal sliding window is constructed for each sensor node, and the standard deviation of the data points within the window is calculated to obtain the local spatiotemporal dispersion dataset of each node in the corresponding spatiotemporal window. Based on the multidimensional carbon flux fusion dataset, the covariance matrix of all sensor node data is calculated, and the global correlation threshold is calculated based on the covariance matrix to obtain the global correlation threshold that represents the overall correlation of the sensor network data in the entire restoration area. Using the local spatiotemporal dispersion data set and the global correlation threshold, the anomaly decision boundary is generated by comparing the deviation degree between the local spatiotemporal dispersion of each sensor node and the global correlation threshold. Based on the abnormal judgment boundary, it is determined whether the data points of each sensor node exceed the boundary. For the node data points that exceed the current boundary, the abnormal state is recorded and marked as abnormal data points.

4. The mangrove restoration area carbon sequestration data monitoring system according to claim 3 is characterized in that: For a collection of grid cells, the topographic deformation rate, environmental factor gradient change rate, and hydrological inundation frequency of each cell within the tidal cycle are calculated to generate spatially corrected quantitative indicators, including: Obtain data recorded by elevation sensors, salinity sensors, temperature sensors, and water level sensors deployed on grid cells in the mangrove restoration area during the tidal cycle; Based on elevation sensor data, the surface subsidence of each grid cell within a single tidal cycle is calculated, and based on the surface subsidence and the corresponding tidal cycle time, the terrain deformation rate of each grid cell is calculated. Based on the gradient monitoring values ​​recorded by the salinity sensor and the temperature sensor, the gradient change rate of the environmental factor of each grid cell is calculated. Based on the data recorded by the water level sensor, the average daily inundation frequency of each grid cell is counted to calculate the hydrological inundation frequency value of each grid cell. The terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency values ​​are normalized to obtain the corresponding normalized terrain deformation rate parameters, environmental factor change rate parameters and hydrological inundation frequency parameters; A weighted fusion calculation is performed on the terrain deformation rate parameters, environmental factor change rate parameters and hydrological inundation frequency parameters to obtain the fusion parameter value of each grid unit. The fusion parameter values ​​are then used to form a correction coefficient matrix that characterizes the spatial heterogeneity of the entire mangrove restoration area, which is a quantitative indicator of spatial correction.

5. The mangrove restoration area carbon sequestration data monitoring system according to claim 4 is characterized in that: Extract sensor data within the grid cells and perform carbon flux compensation based on spatial correction quantitative indicators to obtain a corrected carbon sink dataset, including: For each adaptive grid cell, extract the raw carbon flux data recorded by the carbon flux sensor nodes deployed in the cell during the monitoring period; Read the quantitative indicators of spatial correction, that is, extract the correction coefficient corresponding to the current processing grid cell from the correction coefficient matrix, and calculate the carbon flux compensation value of the grid cell based on the correction coefficient using the terrain deformation rate, environmental factor gradient change rate and hydrological inundation frequency parameters; The original carbon flux data were combined with the carbon flux compensation value to generate the grid cell corrected carbon sink data. At the same time, all the corrected carbon sink data were fused to form a corrected carbon sink dataset covering the irregular polygon monitoring area of ​​the entire mangrove restoration area.

6. The mangrove restoration area carbon sequestration data monitoring system according to claim 5 is characterized in that: Based on the calibrated carbon sink dataset, the carbon sink stability index, spatial coverage contribution index, and ecological process representation index of each node during the assessment period were calculated. The three indicators were integrated into a comprehensive assessment parameter. Based on the distribution characteristics of the comprehensive assessment parameters, prominent nodes were identified as the final deployment points for carbon sink monitoring, including: Based on the corrected carbon sink data set, the ratio of the standard deviation to the mean of the carbon sink data of each sensor node within the preset evaluation period is calculated to generate a carbon sink stability index that characterizes the degree of fluctuation of the node data. Based on the spatial distribution of the adaptive grid cell set and the node position, the ratio of the sum of the areas of the grid cells associated with each node to the total area of ​​the entire irregular polygon monitoring area is calculated to generate a spatial coverage contribution index that represents the spatial representativeness of the node. Based on the hydrological inundation frequency, salinity gradient, and temperature data in the multidimensional carbon flux fusion dataset, the absolute value between each node monitoring data and key ecological process parameters is calculated to generate an ecological process representation index that characterizes the node's ability to reflect ecological processes. The carbon sequestration stability index, spatial coverage contribution index and ecological process representation index of each node are normalized and assigned preset weights for weighted summation to generate comprehensive evaluation parameters for each node; According to the numerical distribution of comprehensive evaluation parameters of all nodes, the nodes with comprehensive evaluation parameter values ​​greater than the preset threshold are determined and determined as the final layout points for carbon sink monitoring.

7. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the system according to any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which, when executed by a processor, implements the system according to any one of claims 1 to 6.

Images