Forest carbon sink intelligent monitoring system and method thereof

Abstract

The invention discloses an intelligent forest carbon sink monitoring system and a method thereof, relates to the technical field of environmental monitoring, realizes real-time acquisition of forest diameter change and biomass increment through layered monitoring, gridding sensor arrangement and tree species and diameter level screening, and realizes real-time acquisition of forest carbon sink carbon sink by utilizing a multi-head attention mechanism, residual connection and a layer normalization technology. The method comprises the following steps: performing deep fusion and prediction on collected data, accurately calculating the carbon sink quantity of each classification, region and overall forest, generating a change trend and an abnormal region mark, dynamically adjusting the sampling frequency of a sensor based on a prediction result and a system, generating a targeted early warning suggestion in combination with an environment variable, and finally performing early warning. Monitoring results are remotely displayed through a three-dimensional thermodynamic diagram, a tree species distribution diagram and a time sequence animation, and scientific and visual decision support is provided for forest carbon sink management and risk early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental monitoring, and particularly to an intelligent forest carbon sink monitoring system and method thereof. Background Art

[0002] Forest carbon sink monitoring is an important part of ecological protection and carbon neutralization. Traditional methods rely on manual measurement or simplified models, which have problems such as low efficiency, limited coverage, and insufficient data real-time. With the progress of technology, intelligent monitoring systems have gradually been applied in this field, improving the accuracy and operability of monitoring.

[0003] Chinese Patent Application CN104374882A discloses a wireless intelligent carbon sink monitoring system. The system realizes the synchronous monitoring of CO2 concentration and environmental data through multiple modules (such as the MCU minimum system module, sensor module, data storage module, etc.) and a remote control host, with the help of a wireless sensor network. This system effectively improves the real-time of data collection. However, its design mainly relies on a fixed sensor network to collect atmospheric data. Due to the lack of direct monitoring of the growth status of forest trees (such as diameter change, biomass increment), it is difficult to reflect the dynamic changes of forest carbon sinks. In addition, Chinese Patent CN119152438B discloses a remote sensing monitoring method and system for forest carbon sinks. By constructing a carbon storage prediction model through ground survey data and vegetation indices, and using machine learning to generate a high-resolution carbon storage distribution map, it solves the problem of insufficient adaptability of traditional models. However, this system mainly relies on remote sensing images and ground survey data, with problems such as a long data update cycle and insufficient real-time. At the same time, it does not involve the dynamic monitoring of abnormal areas and the generation of management suggestions, restricting its application in real-time forest management.

[0004] The above designs realize the real-time collection of data such as atmospheric environment and vegetation indices and the high-resolution prediction of regional carbon storage through wireless sensor networks, remote sensing technology, and machine learning methods, thereby improving the real-time and large-scale monitoring capabilities. However, there are still certain limitations. For example, there is a lack of direct monitoring of the growth status of individual forest trees (such as diameter change, biomass increment), making it difficult to comprehensively reflect the dynamic changes of carbon sinks inside the forest. At the same time, the data analysis model is relatively single, lacking multi-dimensional data fusion, dynamic anomaly detection, and intelligent warning functions, restricting its application in real-time forest management and precise carbon sink regulation. Summary of the Invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and propose an intelligent forest carbon sink monitoring system and method thereof to solve the above problems.

[0006] The object of the present invention is achieved through the following technical solutions: An intelligent monitoring system for forest carbon sinks, comprising a monitoring level division module, a classified sampling monitoring module, a tree species classification unit, a diameter class division unit, a sensor deployment unit, a data calculation module, a large model analysis module, an intelligent monitoring module, and a data transmission and display module; The monitoring level division module is used to sequentially divide the forest carbon layer, plots or sample points, and sample tree monitoring levels within a preset area range; The diameter class division unit is used to classify the trees of each tree species by breast diameter, and select the trees with normal growth and representativeness from each diameter class as monitoring samples; The sensor deployment unit is used to deploy monitoring points in a grid or typically representative manner within the forest plot, deploy high-precision sensors to collect data on the growth change rate and biomass change of the trees, and support dynamically increasing the monitoring point density according to preliminary data anomalies; The data calculation module includes a sample tree carbon sink calculation unit, which is used to calculate the carbon sink amount of a single sample tree based on the collected sample tree data; The large model analysis module includes: The classification attention unit is used to assign independent attention heads to each tree species and diameter class through a multi-head attention mechanism. Each attention head extracts the carbon sink change law based on the time series characteristics of the input data and calculates the carbon sink amount of each type of tree; The global attention unit is used to receive the output of the classification attention unit, analyze the carbon sink synergy effect between different tree species and diameter classes through a global attention mechanism, calculate the total carbon sink amount of the entire forest, and generate a carbon sink change trend curve; The regional attention unit is used to divide the forest into sub-regions differentiated by site type, dominant tree species, and main stand factors based on geographic information system (GIS) data, and use another set of multi-head attention mechanisms to focus on the carbon sink changes in each region and mark the carbon sink anomaly regions.

[0007] The classified sampling monitoring module is used to collect carbon sink data of trees of different tree species and different diameter classes through sensors deployed in the forest, and classify and organize them into a multi-dimensional data set according to tree species and diameter class; The tree species classification unit is used to classify the trees into coniferous trees, broad-leaved trees, and specific tree species according to the growth characteristics and carbon sink capabilities of the tree species within the forest patch, collect the growth data and environmental variables of each tree species, including temperature, rainfall, and soil humidity; the high-precision sensors in the sensor deployment unit are used to monitor the data of the tree diameter change rate in real time and upload the data to the classified sampling monitoring module through a wireless communication module.

[0008] The comprehensive analysis unit is used to fuse the outputs of the classification attention unit, the global attention unit, and the regional attention unit, achieve interactive analysis of multi-head data through residual connection and layer normalization techniques, output the monitoring calculation results of the classified carbon sink volume, regional carbon sink volume, and total carbon sink volume, and train the model based on historical data and real-time data to support short-term and long-term carbon sink change prediction; the comprehensive analysis unit analyzes the contribution rate of each classified and regional carbon sink volume to the total carbon sink volume through the bidirectional information flow of the classification head, regional head, and global head of the multi-head attention mechanism, identifies key influencing factors, and generates GIS coordinate marks for abnormal regions.

[0009] The intelligent monitoring module is used to dynamically adjust the sensor sampling frequency according to the prediction results of the large model analysis module, identify carbon sink anomalies, and generate targeted early warning suggestions. The intelligent monitoring module includes: The sampling frequency adjustment unit is used to judge carbon sink anomalies according to the prediction results of the comprehensive analysis unit, increase the sampling frequency for abnormal regions, and decrease the sampling frequency for stable regions; the early warning suggestion generation unit is used to analyze the reasons for carbon sink anomalies in combination with environmental variables, including drought, pests and diseases, or insufficient soil nutrients, and generate specific suggestions. The data calculation module also includes: the plot carbon sink estimation unit, which is used to calculate or estimate the carbon sink volume of the plot or sample point based on the carbon sink volume of the sample trees; the forest carbon layer calculation unit, which is used to statistically calculate the carbon sink volumes of multiple plots in combination with sampling theory to obtain the carbon sink volumes of each forest carbon layer; the regional carbon sink statistics unit, which is used to summarize and calculate the carbon sink volumes of each forest carbon layer to calculate the total carbon sink volume of the entire region; the classified sampling monitoring module supports adjusting the standards for tree species classification and diameter class division according to forest types and monitoring requirements, and realizes the dynamic optimal layout of monitoring points through the sensor layout unit.

[0010] The data visualization unit is used to convert carbon sink data into 3D heat maps, tree species distribution maps, and time series animations, and support users to view interactively through remote terminals. The data transmission and display module is used to transmit the classified carbon sink volume, regional carbon sink volume, total carbon sink volume, and early warning suggestions to the remote terminal through a wireless network and present them in a 3D visualization form; the early warning suggestion generation unit of the intelligent monitoring module can generate hierarchical early warning suggestions according to the duration and scope of influence of carbon sink anomalies, and push them to the remote terminal through the data transmission and display module.

[0011] The large model analysis module is used to receive the multi-dimensional data set of the classified sampling monitoring module, analyze the carbon sink changes of different tree species, diameter classes, and regions respectively based on the multi-head attention mechanism, and predict the classified carbon sink volume, regional carbon sink volume, and total carbon sink volume of the entire forest; the data transmission and display module supports users to adjust the 3D visualization perspective through the remote terminal, including zooming in on the abnormal region, switching the tree species distribution view, and playing back the carbon sink change animation.

[0012] An intelligent monitoring method for forest carbon sinks, comprising the following steps: S1: On the basis of a comprehensive survey of the trees in the sample plot, select normal-growing sample trees according to different diameter classes, install high-precision sensors, collect forest carbon sink data of different tree species and diameter classes in the forest through the classification sampling monitoring module, and organize it into a multi-dimensional data set; S2: Input the multi-dimensional data set into the large model analysis module, and use the multi-head attention mechanism to analyze and predict the carbon sink amount of the sample plot, the carbon sink amount of the forest carbon layer, the regional carbon sink amount, and the total carbon sink amount; S3: Adjust the sampling frequency according to the prediction results through the intelligent monitoring module, identify anomalies and generate early warning suggestions; S4: Use the data transmission and display module to transmit the prediction results and suggestions to the remote terminal in a three-dimensional visualization form.

[0013] Step S1 includes the following sub-steps: Classify trees according to tree species growth characteristics and carbon sink capacity through the tree species classification unit, and collect growth data and environmental variables; Classify by diameter class according to breast height diameter through the diameter class division unit, and screen representative trees; Arrange monitoring points in a grid or typical representative manner through the sensor layout unit, and collect diameter change and biomass data.

[0014] Step S2 includes the following sub-steps: Assign multi-head attention to tree species and diameter classes through the classification attention unit, extract time series features, and calculate the classified carbon sink amount; Integrate the classification results through the global attention unit, analyze the synergy effect, calculate the total carbon sink amount and the change curve; Divide sub-regions based on GIS through the regional attention unit, pay attention to regional carbon sink changes, and mark abnormal regions; Output the final prediction result by fusing multi-head data through the comprehensive analysis unit.

[0015] Step S3 includes the following sub-steps: Adjust the sampling frequency according to the prediction results through the sampling frequency adjustment unit; Analyze the reasons for anomalies and generate targeted suggestions through the early warning suggestion generation unit; Convert the carbon sink data into a three-dimensional view through the data visualization unit and push it to the remote terminal.

[0016] The beneficial effects of the present invention are: 1. Through the monitoring level division module, the preset area is successively divided into forest carbon layer, sample plot and sample tree monitoring levels. The tree species classification unit and the diameter class division unit are used to classify and screen the trees in detail, ensuring that the collected data is highly representative and accurate. The high-precision sensors arranged in a grid or typical representative manner can collect the growth change rate and biomass increment data of the trees in real time, and dynamically increase the monitoring point density when the data is abnormal, so as to ensure the comprehensiveness of data coverage and the high precision of collection.

[0017] 2. Utilize the multi-head attention mechanism in the large model analysis module (including classification, global, and regional attention units) to perform in-depth fusion and time series feature extraction on the collected multi-dimensional data. It can not only calculate the carbon sink amounts of different tree species and diameter classes, but also train the model based on historical and real-time data to predict short-term and long-term carbon sink change trends. Residual connection and layer normalization techniques are used to perform interactive analysis on the multi-head data, effectively identify key influencing factors, and output accurate carbon sink amounts and trend curves for classification, regional, and overall forests. At the same time, the GIS coordinates of abnormal areas are calibrated to provide a scientific basis for subsequent risk warnings.

[0018] 3. According to the signal prediction results, the intelligent monitoring module dynamically adjusts the sensor sampling frequency for different regions through the sampling frequency adjustment unit: automatically increases the sampling frequency in abnormal areas to obtain higher-resolution data, while reduces the sampling frequency in stable areas to achieve energy conservation and resource optimization. The early warning suggestion generation unit combines environmental variables (such as temperature, rainfall, soil humidity, etc.) to analyze the reasons for abnormalities and promptly generates specific early warning suggestions such as drought, pests and diseases, or insufficient soil nutrients to help managers quickly take countermeasures.

[0019] 4. In addition to calculating the carbon sink amount of individual sample trees, the system also realizes the accurate calculation of carbon sink amounts from local samples to the overall region through the plot carbon sink calculation unit, forest carbon layer calculation unit, and regional carbon sink statistics unit, providing comprehensive carbon sink assessment data. The data transmission and display module presents these monitoring and prediction results in intuitive forms such as 3D heat maps, tree species distribution maps, and time series animations on remote terminals. Users can zoom in on abnormal areas, switch views, and playback historical data through the interactive interface to achieve real-time monitoring and intuitive judgment.

[0020] 5. The present invention organically integrates data collection, preliminary processing, intelligent analysis, dynamic regulation, and 3D visualization display to form a full-process, automated forest carbon sink monitoring system, greatly improving the monitoring efficiency and system response speed. The accurate prediction results and real-time early warnings not only provide timely risk reminders for forest managers, but also provide solid data support for formulating scientific forest carbon sink regulation strategies, thus contributing to ensuring the healthy and stable development of forest ecosystems and promoting the intelligent process of carbon sink management. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is the system architecture of the present invention Figure 1 ; Figure 2 is the system architecture of the present invention Figure 2 ; Figure 3 is the working flow chart of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0022] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0023] It should be noted that in the following solutions, the orientation concepts of "left", "right", "up", "down", "front", "back", "inside", and "outside" are all relative directions, and will not be listed one by one here.

[0024] Embodiment 1 As Figures 1 to 3 shown, this embodiment mainly realizes the functions of basic data collection, hierarchical management, and preliminary data processing of the forest carbon sink monitoring system, ensuring the representativeness and real-time nature of the collected data, and providing an accurate data basis for subsequent intelligent analysis.

[0025] The monitoring level division module divides the forest into layers based on a preset regional scope: Forest carbon layer: First, divide the basic level of forest carbon sink monitoring in the overall area; Plots or sample points: Divide several plots or monitoring points according to geographical distribution and typical representativeness within the forest carbon layer; Individual tree monitoring level: Further select representative individual trees in the plot as the monitoring objects.

[0026] The sensor deployment unit deploys monitoring points in each forest plot in a grid-like typical representative deployment manner, and installs high-precision sensors at each monitoring point to collect data on the growth change rate and biomass change of forest trees in real time. The system supports dynamically increasing the monitoring point density according to the abnormal situation of the initially collected data, so as to ensure comprehensive and detailed data collection coverage.

[0027] The dynamic layout formula of the sensor grid is as follows: Classification sampling monitoring module: Collect carbon sink data of forest trees with different tree species and different diameter classes through sensors, and organize them into a multi-dimensional data set; Tree species classification unit: Classify the forest trees into coniferous trees, broad-leaved trees, and specific tree species by using pre-set growth characteristics and carbon sink capacity parameters, and collect the growth data of specific tree species and environmental variables (such as temperature, rainfall, soil humidity) under each category.

[0028] The diameter class division unit screens and grades the forest trees of each tree species according to the diameter at breast height, and selects the forest trees with normal growth and representativeness as monitoring samples to ensure that the collected data can truly reflect the growth status and carbon sink changes of forest trees of each diameter class.

[0029] The representative sample scoring formula is as follows: The data calculation module includes a sample tree carbon sink calculation unit, which is used to calculate the carbon sink of each sample tree based on the collected sample tree data as the basic data for subsequent large model analysis.

[0030] The large model analysis module uses the multi-head attention mechanism to achieve in-depth data processing. It has three sub-units inside: Classification attention unit: Assign independent attention heads to each tree species and diameter class, extract the carbon sink change law based on the time series characteristics of the input data, and calculate the carbon sink of each type of forest tree; Global attention unit: Receive the output of the classification attention unit, analyze the synergy effect between different tree species and diameter classes using the global attention mechanism, calculate the total carbon sink of the entire forest, and generate a carbon sink change trend curve; Regional attention unit: Combine Geographic Information System (GIS) data, divide the forest into several sub-regions according to site types, dominant tree species and main stand factors, use another group of multi-head attention mechanisms to focus on the carbon sink changes in each region, and mark the possible abnormal regions.

[0031] The data transmission and display module transmits all calculation and analysis results (including classified carbon sink, total carbon sink and trend curve) to the remote terminal in real time through the wireless network and displays them in intuitive ways such as 3D heat maps, tree species distribution maps and time series animations.

[0032] Working process Monitoring level division After the system is started, the monitoring level division module sequentially divides the forest carbon layer, sample plots (or sample points) and sample tree monitoring levels within the preset area to provide a hierarchical structure for subsequent data collection and management.

[0033] Tree species classification and environmental data collection The tree species classification unit classifies all forest trees, divides them into coniferous trees, broad-leaved trees and specific tree species according to the growth characteristics and carbon sink capabilities of each tree species, and simultaneously collects the growth data and environmental variables (temperature, rainfall, soil humidity) under each category.

[0034] Diameter class screening The diameter class division unit grades according to the diameter at breast height of the forest trees and screens out sample forest trees with normal growth conditions and strong representativeness from each diameter class to provide accurate objects for carbon sink data collection.

[0035] Sensor layout and data collection The sensor layout unit arranges monitoring points in each sample plot in a grid pattern or a typical representative pattern. A high-precision sensor is installed at each point. The sensor collects data on the growth change rate and biomass increment of the sample trees in real time, and uploads the data to the classified sampling monitoring module through a wireless communication module. When the preliminary data shows anomalies, the system can dynamically increase the monitoring point density to improve the data collection accuracy.

[0036] Preliminary data processing The classified sampling monitoring module organizes the collected data of different tree species and diameter classes into a multi-dimensional data set and transmits it to the data calculation module. The carbon sink calculation unit of the sample tree calculates the carbon sink amount of each sample tree based on the collected data to form preliminary data.

[0037] In-depth data analysis The preliminary data enters the large model analysis module. First, the classification attention unit uses the multi-head attention mechanism to extract time series features for each tree species and diameter class, and calculates various carbon sink amounts; Next, the global attention unit integrates various outputs, analyzes the synergy effects between different forest trees, calculates the total carbon sink amount of the entire forest, and generates a carbon sink change trend curve; At the same time, the regional attention unit combines GIS data, divides the forest into different sub-regions, further monitors the carbon sink changes within the regions, and marks abnormal regions.

[0038] Data transmission and visual display Finally, the data transmission and display module transmits the monitoring data such as the classified, global, and regional carbon sink amounts and the trend curve to the remote terminal in real time through the wireless network. Users can view the 3D heat map, tree species distribution map, and time series animation through the interactive interface to intuitively understand the forest carbon sink change situation.

[0039] Through hierarchical monitoring (forest carbon layer, sample plot, and sample tree) and strict screening of tree species and diameter classes, it is ensured that the collected data is highly representative and reflects the true carbon sink situation of each level of the forest.

[0040] Using high-precision sensors and grid-based representative layout to ensure real-time monitoring of the growth dynamics of forest trees; at the same time, the system supports dynamically adjusting the monitoring point density in case of data anomalies to further improve the accuracy and coverage of data collection.

[0041] Using the multi-head attention mechanism for time series feature extraction, synergy effect analysis, and regional division can not only calculate the carbon sink amounts of individual trees and the entire forest, but also generate change trend curves and mark abnormal regions, providing a scientific basis for forest management and carbon sink regulation.

[0042] The three-dimensional heat map, tree species distribution map, and time series animation implemented through the data transmission and display module enable users to intuitively understand the forest carbon sink monitoring data and effectively support remote monitoring and decision-making.

[0043] The seamless connection between each functional module realizes an integrated monitoring system from data collection, preliminary processing to in-depth analysis and visual display, improving the overall monitoring efficiency and management level.

[0044] Through the above implementation methods and working processes, Example 1 can provide an efficient, intelligent and accurate basic platform for forest carbon sink monitoring, realize the whole-process dynamic monitoring of the forest carbon sink status, and provide solid data support for subsequent intelligentization and management decision-making optimization.

[0045] Example 2 As Figures 1 to 3 shown, this example is described based on Example 1. The focus is on realizing the in-depth processing, prediction, anomaly detection and dynamic regulation of the collected data through the comprehensive analysis unit, intelligent monitoring module and data visualization upgrade, thereby greatly improving the intelligent level and monitoring accuracy of the system.

[0046] The comprehensive analysis unit receives the multi-dimensional data output by the large model analysis module in Example 1, including: Output of the classification attention unit (extracting time series features and calculating carbon sinks for each tree species and diameter class) Output of the global attention unit (analyzing the synergistic effect between different tree species and diameter classes, calculating the total carbon sink of the whole forest and generating a carbon sink trend curve) Output of the regional attention unit (dividing sub-regions based on GIS data, monitoring regional carbon sink changes, and marking abnormal regions); The calculation formula for the multi-head attention weights is as follows: To achieve the interactive analysis of multi-head data, the comprehensive analysis unit uses residual connection and layer normalization techniques to fuse the data from each attention module. Through two-way information flow, the classification head, regional head and global head of the comprehensive analysis unit respectively analyze the contribution of each classification and regional carbon sink to the total carbon sink, identify the key influencing factors, and generate GIS coordinate marks for abnormal regions.

[0047] The multi-head output fusion and carbon sink prediction formula are as follows: Model training and prediction Using historical data and real-time data, the comprehensive analysis unit trains a prediction model to support the prediction of short-term and long-term carbon sink changes and outputs: The carbon sink of the sample plot, the carbon sink of the forest carbon layer; the regional carbon sink; the total carbon sink; at the same time, it provides data support for subsequent monitoring and regulation.

[0048] Construction of the intelligent monitoring module Sampling frequency adjustment. The intelligent monitoring module is set according to the prediction result after information fusion as follows: Sampling frequency adjustment unit: According to the prediction result of the comprehensive analysis unit, it judges whether there is a carbon sink anomaly in the monitoring area. For the detected abnormal area, it automatically increases the sensor sampling frequency to obtain higher-resolution data; for the stable area, it reduces the sampling frequency to save system resources.

[0049] The dynamic adjustment formula of the sampling frequency is as follows: The early warning and suggestion generation unit combines environmental variables (temperature, rainfall, soil humidity, etc.) to analyze the causes of carbon sink anomalies (such as drought, pests and diseases, insufficient soil nutrients), and generates specific and targeted early warning suggestions.

[0050] The expansion of the data calculation module adds, in addition to the original sample tree carbon sink calculation unit: Sample plot carbon sink estimation unit: Estimates the carbon sink of the entire sample plot or sample point based on the carbon sink of a single sample tree.

[0051] Forest carbon layer calculation unit: Combines sampling theory to statistically calculate the carbon sink of multiple sample plots to obtain the carbon sink data of each forest carbon layer.

[0052] Regional carbon sink statistics unit: Summarizes and statistics the carbon sink of each forest carbon layer to calculate the total carbon sink of the entire region.

[0053] In addition, the classified sampling monitoring module supports dynamically adjusting the standards of tree species classification and diameter class division according to forest types and actual monitoring needs, and realizes the dynamic optimization of the monitoring point layout through the sensor layout unit.

[0054] Early warning risk index calculation formula: Data visualization and remote interaction The data visualization unit converts the calculated classified, regional and total carbon sinks to generate: Three-dimensional heat map; Tree species distribution map; Time series animation; Upgrade of the data transmission and display module Use the wireless network to transmit the above data and early warning suggestions to the remote terminal. Users can adjust the three-dimensional visualization perspective through the interactive interface, including: Zoom in on the abnormal area; Switch the tree species distribution view; Play back the carbon sink change animation; Working Process Data Fusion and Prediction In Embodiment 1, the system has completed forest stratification, tree species and diameter class classification, sensor layout, and data collection. After being sorted by the classification sampling monitoring module, the collected multi-dimensional data is transmitted to the large model analysis module.

[0055] The comprehensive analysis unit receives the outputs from the classification attention, global attention, and regional attention units, realizes data fusion through residual connection and layer normalization techniques, outputs the prediction results of classification, region, and total carbon sink volume, analyzes the contribution of each part to the total carbon sink volume through bidirectional multi-head information flow, and simultaneously generates the GIS coordinates of the abnormal area.

[0056] Dynamic Monitoring and Regulation The intelligent monitoring module uses the prediction results of the comprehensive analysis unit to make a dynamic judgment on the monitoring area.

[0057] The sampling frequency adjustment unit automatically increases the sensor sampling frequency for the abnormal area and decreases the sampling frequency for the stable area to ensure the accuracy of data collection and the reasonable utilization of resources.

[0058] Meanwhile, the early warning suggestion generation unit combines environmental variable analysis to generate specific early warning suggestions and timely provides managers with response plans for abnormal reasons such as drought, pests and diseases, or insufficient soil nutrients.

[0059] Multi-dimensional Data Calculation and Comprehensive Statistics In the data calculation module, the newly added sample plot carbon sink calculation unit, forest carbon layer calculation unit, and regional carbon sink statistics unit expand and statistically analyze the collected data to ensure the accurate calculation of the carbon sink volume from single-plant sample data to the entire region.

[0060] Data Display and Interaction Finally, all prediction results, calculation data, and early warning suggestions are displayed on the remote terminal in the form of 3D heat maps, tree species distribution maps, and time series animations through the data transmission and display module.

[0061] Users can zoom in on the data, switch views, and replay the carbon sink change animation through the interaction interface to achieve intuitive monitoring and timely adjustment of the forest carbon sink status.

[0062] By implementing the multi-head attention mechanism data fusion through the comprehensive analysis unit and adopting residual connection and layer normalization techniques, the accuracy and robustness of carbon sink data prediction are greatly improved, supporting reliable prediction of short-term and long-term change trends.

[0063] The intelligent monitoring module dynamically adjusts the sensor sampling frequency according to signal prediction, realizes high-density monitoring of abnormal areas and low-frequency sampling of stable areas, effectively improves the system operation efficiency, reduces energy consumption, and simultaneously ensures data quality.

[0064] Through plot carbon sink extrapolation, forest carbon layer calculation, and regional carbon sink statistical units, the monitoring is extended from sample data to the whole-region carbon sink, providing a more comprehensive and detailed carbon sink assessment and scientific basis for forest carbon sink management.

[0065] The data visualization unit and data transmission and display module present complex carbon sink data in the form of 3D heat maps, tree species distribution maps, and time-series animations, enabling users to intuitively identify abnormal regions and carbon sink change trends, facilitating remote monitoring and timely decision-making.

[0066] The early warning and suggestion generation unit generates targeted early warnings in combination with environmental variables. It can not only detect abnormalities in real time but also provide cause analysis and countermeasures for problems such as drought and pests and diseases, providing effective risk early warning and decision-making support for forest carbon sink managers.

[0067] In summary, on the basis of Embodiment 1, Embodiment 2 realizes the full-process intelligent monitoring from data collection to intelligent prediction, abnormal early warning, and dynamic adjustment through intelligent data comprehensive analysis, dynamic regulation, and interactive 3D visualization, greatly improving the application effect and management level of the forest carbon sink monitoring system.

[0068] Embodiment 3 As Figures 1 to 3 shown, this embodiment proposes a forest carbon sink intelligent monitoring method based on Embodiment 1 and Embodiment 2, realizing the whole-process intelligent monitoring and prediction of the forest carbon sink status through data collection, in-depth analysis, dynamic regulation, and 3D visualization display.

[0069] Data Collection and Preliminary Processing (Step S1 and Its Sub-steps) Full inspection and sample selection: In the pre-set plots, conduct a comprehensive census of all trees, classify the trees according to the diameter at breast height (diameter class), use the diameter class division unit to screen out normal-growing and representative sample trees, select monitoring samples for each diameter class, and install high-precision sensors.

[0070] Tree Species Classification and Environmental Data Collection Using the tree species classification unit, classify the trees in the forest into categories such as coniferous trees, broad-leaved trees, and specific tree species according to the growth characteristics and carbon sink capabilities of the trees. At the same time, collect the growth data and key environmental variables (such as temperature, rainfall, soil humidity, etc.) of each tree.

[0071] Grid Layout of Sensors and Data Collection Through the sensor layout unit, arrange monitoring points in a grid pattern in each plot. The high-precision sensors collect the data of the diameter change rate and biomass increment of the trees in real time, and organize the collection results into a multi-dimensional data set as the basis for subsequent analysis.

[0072] Deep data analysis and prediction (Step S2 and its sub-steps) Multi-head attention mechanism analysis Input the multi-dimensional data set sorted out in Step S1 into the large model analysis module, and this module uses the following sub-units to work together: Classification attention unit: Assign independent attention heads to different tree species and diameter classes, extract time series features and calculate the classified carbon sink; Global attention unit: Integrate the classification results of each category, use the global attention mechanism to analyze the synergistic effect between different tree species and diameter classes, calculate the total carbon sink of the entire forest and generate a change trend curve; Regional attention unit: Divide the forest into several sub-regions based on GIS data, use another group of multi-head attention mechanisms to focus on the carbon sink changes in each sub-region, and mark the areas with abnormal carbon sinks; Comprehensive analysis unit: Through residual connection and layer normalization techniques, fuse the above multi-head data to form a two-way information flow, and output the prediction results of the final classified carbon sink, regional carbon sink and total carbon sink.

[0073] Dynamic regulation and early warning (Step S3 and its sub-steps) Dynamically adjust the sampling frequency. The sampling frequency adjustment unit in the intelligent monitoring module makes an intelligent judgment on the monitoring area according to the prediction results after information fusion: For the areas where carbon sink anomalies are detected, automatically increase the sampling frequency of the sensors to obtain higher-resolution data; For the areas with stable states, reduce the sampling frequency to achieve resource optimization.

[0074] Abnormal analysis and early warning generation. The early warning suggestion generation unit combines real-time environmental variables to deeply analyze the detected reasons for carbon sink anomalies (such as drought, pests and diseases, or insufficient soil nutrients, etc.), and generates specific and targeted early warning suggestions. At the same time, the data visualization unit converts the collected carbon sink data into a three-dimensional view to provide an intuitive basis for early warning decision-making.

[0075] Data transmission and interactive display (Step S4) Three-dimensional visualization display and remote interaction. Use the data transmission and display module to transmit the prediction results and early warning suggestions of the large model analysis and intelligent monitoring modules to the remote terminal through the wireless network. The display content includes: Three-dimensional heat map; Tree species distribution map; Time series animation; Users can perform interactive operations through the remote terminal, such as zooming in on the abnormal area, switching different tree species views, and playing back the carbon sink change process, which is convenient for real-time monitoring and decision-making.

[0076] Step S1: Data Collection and Sorting Conduct a full survey within the sample plot, screen out normal-growing sample trees according to diameter class grading, and install high-precision sensors; Collect growth data and environmental variables of various forest trees through tree species classification units; Use the sensor layout unit to layout monitoring points in a grid pattern, collect data on diameter changes and biomass increments in real time, and organize them into a multi-dimensional data set.

[0077] Step S2: Data Input and Depth Prediction Input the sorted multi-dimensional data set into the large model analysis module; The classification attention unit, global attention unit, and regional attention unit respectively extract features of different dimensions, and fuse multi-head data through the comprehensive analysis unit. Use residual connection and layer normalization techniques to output the final prediction results of classification carbon sink volume, regional carbon sink volume, and total carbon sink volume; At the same time, generate a forest carbon sink change trend curve and GIS coordinate marks for abnormal areas.

[0078] Step S3: Intelligent Regulation and Abnormal Warning The intelligent monitoring module dynamically adjusts the sampling frequency of monitoring points through the sampling frequency adjustment unit according to the prediction results, realizing resource optimization of high-frequency monitoring of abnormal areas and low-frequency sampling of stable areas; The warning suggestion generation unit combines environmental data analysis to analyze the reasons for abnormalities and generates targeted warning suggestions in aspects such as drought, pests and diseases, or insufficient soil nutrients; The data visualization unit converts these carbon sink data and warning information into intuitive three-dimensional views.

[0079] Step S4: Data Transmission and Remote Display All prediction results, warning suggestions, and three-dimensional visualization graphics are transmitted to the remote terminal in a wireless network manner through the data transmission and display module; Users can adjust the view through the terminal interface, such as zooming in on the abnormal area, switching the tree species distribution map, and playing back the time series animation, to obtain forest carbon sink monitoring and warning information in a timely manner.

[0080] This method realizes the full-process automated monitoring from full survey sampling, data sorting, depth prediction to dynamic regulation, warning prompt, and remote interactive display, ensuring the scientificity and systematicness of forest carbon sink monitoring work.

[0081] Through the residual connection and layer normalization techniques of the multi-head attention mechanism and the comprehensive analysis unit, the carbon sink data of different tree species, diameter classes, and regions are deeply fused, significantly improving the prediction accuracy and supporting reliable predictions of short-term and long-term change trends.

[0082] The intelligent monitoring module can automatically adjust the sampling frequency according to the signal prediction results, achieving high-density monitoring of abnormal areas and low-frequency sampling of stable areas, which not only improves the accuracy of data acquisition but also effectively reduces energy consumption and system resource consumption.

[0083] Through the display methods of three-dimensional heat maps, tree species distribution maps, and time-series animations, users can intuitively understand the carbon sink status and change trends of various forest areas, facilitating the timely discovery of problems and the adoption of countermeasures. At the same time, it supports remote monitoring and decision-making assistance.

[0084] The early warning and suggestion generation unit provides targeted early warnings by combining real-time environmental variables, timely identifies abnormal situations such as droughts and pests and diseases, provides scientific risk warnings and decision-making bases for forest managers, and ensures the healthy and stable development of the forest ecosystem.

[0085] In summary, through a series of systematic and automated operation steps, Example 3 upgrades the traditional data acquisition and monitoring methods to a full-process, intelligent, and dynamically adjustable forest carbon sink monitoring method, providing comprehensive, accurate, and efficient technical support for forest carbon sink management.

[0086] The above are only the preferred embodiments of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein, should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications, and environments, and can be changed within the scope of the concept described herein through the above teachings or the techniques or knowledge in related fields. As long as the changes and variations made by those skilled in the art do not depart from the spirit and scope of the present invention, they should all be within the protection scope of the appended claims of the present invention.

Claims

1. An intelligent monitoring system for forest carbon sinks, characterized in that, It includes a monitoring level division module, a classified sampling monitoring module, a tree species classification unit, a diameter class division unit, a sensor deployment unit, a data calculation module, a large model analysis module, an intelligent monitoring module, and a data transmission and display module; The monitoring level division module is used to sequentially divide the forest carbon layer, plots or sample points, and sample tree monitoring levels within a preset area range; The diameter class division unit is used to classify the trees of each tree species in the plot according to the diameter at breast height, and select the trees with normal growth and representativeness from each diameter class as monitoring samples; The sensor deployment unit is used to deploy monitoring points in the forest plot in a typical sampling or grid pattern, deploy high-precision sensors to collect the growth change rate and biomass increment data of the trees, and support dynamically increasing the monitoring point density according to the initial data anomalies; The data calculation module includes a sample tree carbon sink calculation unit, which is used to calculate the carbon sink amount of a single sample tree based on the collected sample tree data; The large model analysis module includes: The classification attention unit is used to assign independent attention heads to each tree species and diameter class through the multi-head attention mechanism. Each attention head extracts the carbon sink change law based on the time series characteristics of the input data and calculates the carbon sink amount of each type of tree; The global attention unit is used to receive the output of the classification attention unit, analyze the carbon sink synergy effect between different tree species and diameter classes through the global attention mechanism, calculate the total carbon sink amount of the entire forest, and generate a carbon sink change trend curve; The regional attention unit is used to divide the forest into sub-regions differentiated by site type, dominant tree species, and main forest stand factors based on GIS data of geographic information systems, and use another set of multi-head attention mechanisms to focus on the carbon sink changes in each region and mark the carbon sink anomaly regions.

2. The intelligent monitoring system for forest carbon sink according to claim 1, wherein: The classified sampling monitoring module is used to collect the carbon sink data of trees of different tree species and different diameter classes through the sensors deployed in the forest, and classify and organize them into a multi-dimensional data set according to the tree species and diameter class; The tree species classification unit is used to classify the trees into coniferous trees, broad-leaved trees, and specific tree species according to the growth characteristics and carbon sink capacity of the tree species in the forest patch, and collect the growth data and environmental variables of each tree species, including temperature, rainfall, and soil humidity; the high-precision sensors in the sensor deployment unit are used to monitor the data of the tree diameter change rate in real time and upload the data to the classified sampling monitoring module through the wireless communication module.

3. The intelligent monitoring system for forest carbon sinks according to claim 1, wherein: The comprehensive analysis unit is used to fuse the outputs of the classification attention unit, the global attention unit, and the regional attention unit, realize the interactive analysis of multi-head data through residual connection and layer normalization technologies, output the monitoring calculation results of the classified carbon sink amount, regional carbon sink amount, and total carbon sink amount, and train the model based on historical data and real-time data to support short-term and long-term carbon sink change prediction; The comprehensive analysis unit analyzes the contribution rate of each classified and regional carbon sink amount to the total carbon sink amount through the bidirectional information flow of the classification head, regional head, and global head of the multi-head attention mechanism, identifies the key influencing factors, and generates the GIS coordinate marks of the anomaly regions.

4. The intelligent monitoring system for forest carbon sinks according to claim 1, wherein: The intelligent monitoring module is used to dynamically adjust the sensor sampling frequency according to the prediction results of the large model analysis module, identify carbon sink anomalies and generate targeted early warning suggestions; The intelligent monitoring module includes: The sampling frequency adjustment unit is used to judge carbon sink anomalies according to the prediction results of the comprehensive analysis unit, increase the sampling frequency for abnormal areas and decrease the sampling frequency for stable areas; The early warning suggestion generation unit is used to analyze the reasons for carbon sink anomalies in combination with environmental variables, including drought, pests and diseases, or insufficient soil nutrients, and generate specific suggestions; The data calculation module further includes: a plot carbon sink calculation unit for calculating or estimating the carbon sink of a plot or a sample point based on the carbon sink of sample trees; a forest carbon layer calculation unit for statistically calculating the carbon sink of multiple plots in combination with sampling theory to obtain the carbon sink of each forest carbon layer; a regional carbon sink statistics unit for summarizing and calculating the carbon sink of each forest carbon layer to calculate the total carbon sink of the entire region; The classified sampling monitoring module supports adjusting the standards of tree species classification and diameter class division according to forest types and monitoring requirements, and realizes the dynamic optimal layout of monitoring points through the sensor layout unit.

5. The intelligent monitoring system for forest carbon sink according to claim 1, wherein: The data visualization unit is used to convert carbon sink data into three-dimensional heat maps, tree species distribution maps and time series animations, and support users to view interactively through a remote terminal; The data transmission and display module is used to transmit the classified carbon sink, regional carbon sink, total carbon sink and early warning suggestions to the remote terminal through a wireless network and present them in a three-dimensional visualization form; The early warning suggestion generation unit of the intelligent monitoring module can generate hierarchical early warning suggestions according to the duration and influence range of carbon sink anomalies, and push them to the remote terminal through the data transmission and display module.

6. The intelligent monitoring system for forest carbon sinks according to claim 1, wherein: The large model analysis module is used to receive the multi-dimensional data set of the classified sampling monitoring module, analyze the carbon sink changes of different tree species, diameter classes and regions respectively based on the multi-head attention mechanism, and predict the classified carbon sink, regional carbon sink and total carbon sink of the entire forest; The data transmission and display module supports users to adjust the three-dimensional visualization perspective through the remote terminal, including zooming in on the abnormal area, switching the tree species distribution view and playing back the carbon sink change animation.

7. An intelligent monitoring method for forest carbon sinks, characterized in that: Including the following steps: S1: On the basis of a full inspection of the trees in the plot, select normal-growing sample trees according to different diameter classes, install high-precision sensors, collect the carbon sink data of trees of different tree species and diameter classes in the forest through the classified sampling monitoring module, and organize them into a multi-dimensional data set; S2: Input the multi-dimensional data set into the large model analysis module, and use the multi-head attention mechanism to analyze and predict the classified carbon sink, regional carbon sink and total carbon sink; S3: Adjust the sampling frequency according to the prediction results through the intelligent monitoring module, identify anomalies and generate early warning suggestions; S4: Use the data transmission and display module to transmit the prediction results and suggestions to the remote terminal in a three-dimensional visualization form.

8. The intelligent monitoring method for forest carbon sink according to claim 7, characterized in that: The step S1 includes the following sub-steps: Classify forest trees according to tree species classification units based on tree species growth characteristics and carbon sequestration capabilities, and collect growth data and environmental variables; classify by diameter class unit according to breast height diameter, and screen representative forest trees; deploy monitoring points in a typical representative or grid manner through the sensor deployment unit, and collect diameter change and biomass data.

9. The intelligent monitoring method for forest carbon sinks according to claim 7, wherein: The step S2 includes the following sub-steps: Allocate multi-head attention to tree species and diameter classes through the classification attention unit, extract time series features, and calculate the classified carbon sequestration amount; integrate the classification results through the global attention unit, analyze the synergy effect, and calculate the total carbon sequestration amount and change curve; divide sub-regions based on GIS through the regional attention unit, focus on regional carbon sequestration changes, and mark abnormal regions; fuse multi-head data through the comprehensive analysis unit and output the final prediction result.

10. A method for intelligent monitoring of forest carbon sinks according to claim 7, characterized in that: The step S3 includes the following sub-steps: Adjust the sampling frequency according to the prediction result through the sampling frequency adjustment unit; analyze the cause of the anomaly and generate targeted suggestions through the early warning suggestion generation unit; convert the carbon sequestration data into a three-dimensional view through the data visualization unit and push it to the remote terminal.

Images