Forest carbon sink accurate monitoring system and method based on multi-source data fusion

Abstract

The invention discloses a forest carbon sink accurate monitoring system and method based on multi-source data fusion, and relates to the technical field of carbon sink monitoring. In order to solve the problems of low precision and poor timeliness of the traditional monitoring means, the adopted scheme comprises a multi-source data acquisition module used for acquiring satellite remote sensing, unmanned aerial vehicle aerial photography, ground monitoring sensor and meteorological data; the data pre-processing and fusion module is used for pre-processing the collected data and extracting and combining related features through a two-layer fusion method to serve as input data of a carbon sink estimation model; the model construction and optimization module is used for constructing and optimizing a carbon sink estimation model; the result visualization and analysis module is used for displaying a carbon sink estimation result, researching influence factors of forest carbon sinks and predicting a future change trend; and the integration and management module is used for integrating hardware equipment to construct a monitoring platform, developing a monitoring system based on cloud computing and the Internet of Things, and establishing and perfecting a management system. The method is used for monitoring the forest carbon sink.

Description

Technical Field

[0001] This invention relates to the field of carbon sink monitoring technology, specifically a precise monitoring system and method for forest carbon sinks based on multi-source data fusion. Background Technology

[0002] Against the backdrop of global warming, forest carbon sinks, as an important component of carbon sequestration in terrestrial ecosystems, play an irreplaceable role in accurately monitoring their value for quantifying ecosystem services, formulating carbon reduction strategies, and promoting the development of carbon trading markets.

[0003] Traditional forest carbon sequestration monitoring methods are mainly divided into ground survey methods and remote sensing monitoring methods.

[0004] The ground survey method involves professionals measuring parameters such as diameter at breast height (DBH), tree height, and crown width of trees in the field, and then calculating carbon storage using a biomass conversion model. Taking a provincial forest resource inventory as an example, a comprehensive ground survey requires hundreds of people and 3-6 months of work, costing millions of yuan. Moreover, due to the limited number of sampling points, it is difficult to cover complex terrain areas and remote forest areas, resulting in significant limitations in the spatial representativeness of the monitoring results.

[0005] Remote sensing monitoring utilizes satellite imagery to acquire forest vegetation information and subsequently estimate carbon sequestration. However, limitations in the spatial resolution of satellite imagery (e.g., Landsat 8 imagery has a resolution of 30 meters) make it difficult to identify individual trees. Furthermore, factors such as clouds, fog, and atmospheric aerosols severely interfere with the quality of remote sensing data, leading to deviations in vegetation index calculations and affecting the accuracy of carbon sequestration estimation. According to relevant research, in the cloudy and rainy forest areas of southern China, the carbon sequestration estimation error using traditional remote sensing monitoring methods can reach as high as 25%-35%.

[0006] Furthermore, most existing monitoring methods rely on a single data source, such as satellite remote sensing data or ground survey data alone. A single data source cannot comprehensively capture the dynamic changes in forest ecosystems, such as vegetation carbon loss caused by forest pests and diseases, and changes in understory carbon storage, making the monitoring results unable to accurately reflect the actual situation of forest carbon sinks. With the gradual improvement of the carbon trading market, higher requirements are placed on the accuracy, timeliness, and reliability of forest carbon sink monitoring. Therefore, there is an urgent need for a technical solution that can integrate multi-source data to achieve precise monitoring. Summary of the Invention

[0007] This invention addresses the problems of low accuracy, poor timeliness, and high cost of traditional monitoring methods by providing a precise monitoring system and method for forest carbon sequestration based on multi-source data fusion.

[0008] In a first aspect, the present invention provides a forest carbon sequestration precision monitoring system based on multi-source data fusion, and the technical solution adopted to solve the above-mentioned technical problems is as follows:

[0009] A precise monitoring system for forest carbon sequestration based on multi-source data fusion, comprising:

[0010] The multi-source data acquisition module is used to collect satellite remote sensing data, UAV aerial photography data, ground monitoring sensor data, and meteorological data;

[0011] The data preprocessing and fusion module is used to preprocess the collected data and extract and combine relevant features through the feature layer and decision layer fusion method, which are used as input data for the carbon sink estimation model.

[0012] The model building and optimization module is used to build carbon sink estimation models by integrating convolutional neural networks, long short-term memory networks and random forest algorithms for different data characteristics, and to continuously optimize model performance through a dynamic data update mechanism.

[0013] The carbon sink estimation model is used to take the output of the data preprocessing and fusion module as input and output the carbon sink estimation result.

[0014] The results visualization and analysis module is used to display carbon sink estimation results in various visualization formats, and to study the impact of different factors on forest carbon sinks and predict future trends using sensitivity analysis and scenario simulation techniques.

[0015] The integration and management module is used to integrate hardware devices to build a monitoring platform, develop monitoring systems based on cloud computing and the Internet of Things, establish sound management systems, and ensure the operation of the monitoring system and data security.

[0016] Optionally, the multi-source data acquisition modules involved specifically include:

[0017] The satellite remote sensing data acquisition unit is used to acquire satellite remote sensing data using high-resolution optical satellites and radar satellites, and to obtain multispectral, hyperspectral and microwave remote sensing images.

[0018] The drone aerial photography data acquisition unit is used to acquire drone aerial photography data and obtain local images of forests using drones equipped with multispectral cameras, thermal infrared cameras and lidar.

[0019] The ground sensor data acquisition unit is used to collect ground monitoring sensor data using tree diameter growth and tree height measurement equipment, soil temperature and humidity sensors, soil carbon flux sensors, and weather stations.

[0020] Meteorological data acquisition unit, used to collect meteorological data in cooperation with meteorological departments or by using meteorological satellites.

[0021] Optionally, the data preprocessing and fusion modules involved specifically include:

[0022] The data preprocessing unit is used to perform targeted processing on different types of collected data, including radiometric correction, geometric correction and atmospheric correction for satellite remote sensing data, mosaicking and color correction for UAV aerial photography data, filtering and noise reduction and outlier processing for ground monitoring sensor data, and data cleaning and format conversion for meteorological data.

[0023] The data fusion unit employs a two-level fusion method: a feature layer and a decision layer. At the feature layer, various relevant features are extracted and combined from the preprocessed data, such as normalized difference vegetation index, enhanced vegetation index, texture features, single tree canopy area, canopy closure, three-dimensional structural features, tree growth rate, soil carbon content, and meteorological factors. At the decision layer, a carbon sink estimation model based on satellite remote sensing data, UAV data, and ground monitoring data is used to synthesize the estimation results of the carbon sink estimation model through a weighted average method, ultimately obtaining the carbon sink estimation result.

[0024] Optionally, the model building and optimization modules involved specifically include:

[0025] The model building unit is used to extract spatial features from satellite remote sensing data and UAV aerial photography data using convolutional neural networks, and to capture the temporal patterns of carbon sink changes using long short-term memory networks for ground monitoring sensor data and meteorological data. Then, the features of multi-source data are integrated through random forest algorithm to build a carbon sink estimation model.

[0026] The data preparation unit is used to collect ground-based measured carbon sink data and remote sensing / monitoring data from different climate zones and forest types, and is divided into training set, validation set and test set in a 7:2:1 ratio;

[0027] The model optimization unit is used to employ the Adam optimization algorithm with an initial learning rate of 0.001. It dynamically adjusts the batch size and number of iterations, evaluates the performance of the carbon sink estimation model in real time through the validation set, and saves the optimal model parameters.

[0028] The model monitoring unit is used to continuously monitor the operating status of the carbon sink estimation model, save the optimal model parameters, and ensure the accuracy and timeliness of the estimation results.

[0029] The timed update unit is used to periodically call the data preparation unit to collect ground-based measured carbon sink data and remote sensing / monitoring data from different climate zones and forest types, and to incrementally update the parameters of the carbon sink estimation model to ensure that the carbon sink estimation model adapts to the dynamic changes of the forest ecosystem.

[0030] Alternatively, the convolutional neural network used in the model building units may be an improved ResNet-50 network with added residual connections and attention mechanisms.

[0031] Optionally, the visualization and analysis modules involved specifically include:

[0032] The visualization unit provides a visual interface that presents carbon sequestration estimation results in the form of heat maps, 3D scenes, and time series charts. It intuitively displays the spatial distribution, vertical distribution characteristics, time-varying trends, and correlation with meteorological factors of carbon sequestration, allowing users to clearly and comprehensively understand carbon sequestration-related data.

[0033] The analysis and prediction unit is used to conduct in-depth research on the impact of different factors on forest carbon sinks using sensitivity analysis and scenario simulation techniques, and to predict future trends in forest carbon sinks. Through in-depth data mining and analysis, it assesses the dynamic changes and impacts of carbon sinks.

[0034] Alternatively, the visualization unit can provide a visualization interface developed based on the ArcGIS Online geographic information system platform, allowing users to view carbon sequestration-related data by clicking on any area on an interactive map.

[0035] The analysis and prediction unit uses sensitivity analysis to study the impact of afforestation area, forest tending measures, and climate change factors on forest carbon sinks. Based on historical data and combined with future climate prediction scenarios of RCP4.5 and RCP8.5, it simulates the changing trend of forest carbon sinks over a specified future time period.

[0036] Optionally, the integration and management modules involved specifically include:

[0037] The hardware integration unit is used to integrate data acquisition devices, edge computing nodes, data servers, and visualization terminals to build a hardware architecture for forest carbon sequestration monitoring.

[0038] The software development unit is used to develop a forest carbon sequestration monitoring software system based on cloud computing technology and using the Spring Boot + Vue.js technology stack. It implements data interaction between various components through RESTful API interfaces.

[0039] The management and maintenance unit is used to formulate data security management systems, establish regular equipment maintenance plans, and set up system log monitoring functions to provide dual protection for the stable operation of the monitoring system and the security of data from both institutional and technical perspectives.

[0040] Secondly, this invention provides a method for precise monitoring of forest carbon sequestration based on multi-source data fusion. The technical solution adopted to solve the above-mentioned technical problems is as follows:

[0041] A method for precise monitoring of forest carbon sequestration based on multi-source data fusion, which is based on the system described in the first aspect, includes the following steps:

[0042] The system utilizes a multi-source data acquisition module to collect satellite remote sensing data, UAV aerial photography data, ground monitoring sensor data, and meteorological data.

[0043] The collected data is preprocessed using a data preprocessing and fusion module, and relevant features are extracted and combined using a feature layer and decision layer fusion method to provide input data for the carbon sink estimation model.

[0044] The model building and optimization module integrates convolutional neural networks, long short-term memory networks, and random forest algorithms to construct carbon sink estimation models for different data characteristics, and continuously optimizes model performance through a dynamic data update mechanism; the carbon sink estimation model obtains carbon sink estimation results based on the feature data output by the data preprocessing and fusion module.

[0045] The results visualization and analysis module presents carbon sink estimation results in various visualization formats, and uses sensitivity analysis and scenario simulation techniques to study the impact of different factors on forest carbon sinks and predict future trends.

[0046] The integration and management module integrates hardware devices to build a monitoring platform, develops a monitoring system based on cloud computing and the Internet of Things, establishes a sound management system, ensures the operation of the monitoring system and data security, and provides support for the entire process.

[0047] The forest carbon sequestration precision monitoring system and method based on multi-source data fusion of the present invention has the following advantages compared with the prior art:

[0048] 1. This invention integrates multi-source data such as satellite remote sensing, UAVs, ground sensors and meteorology, and uses data processing technology and machine learning algorithms to make up for the shortcomings of traditional monitoring methods, so as to achieve high-precision, real-time monitoring and dynamic assessment of forest carbon sinks, providing accurate carbon sink data support for forest resource management departments to formulate scientific afforestation and forest cultivation policies and for enterprises to participate in carbon trading.

[0049] 2. Compared with traditional single-data source monitoring methods, this invention can improve monitoring accuracy by more than 30%, shorten the monitoring cycle from several months to less than 7 days, and reduce monitoring costs by about 40%, effectively solving the problems of low accuracy, poor timeliness, and high cost of traditional monitoring methods;

[0050] 3. This invention effectively compensates for the shortcomings of single data sources by using multi-source data fusion and machine learning algorithms, significantly improving monitoring accuracy. By leveraging periodic observations from satellites and drones, as well as real-time data acquisition from ground sensors, the monitoring cycle is shortened from several months in traditional methods to less than 7 days, facilitating real-time dynamic monitoring. It reduces the manpower and material resources required for ground surveys, while optimizing data processing workflows improves monitoring efficiency and lowers monitoring costs. It not only monitors forest vegetation carbon sinks but also comprehensively monitors understory carbon storage, soil carbon flux, and the response of forest ecosystems to climate change. Through collaborative analysis of multi-source data, it provides in-depth understanding of forest carbon cycle processes, offering comprehensive information for the scientific management of forest ecosystems. Through visualization and in-depth data analysis, it provides precise data support for relevant departments in formulating forestry development plans and for enterprises participating in carbon trading. Attached Figure Description

[0051] Appendix Figure 1 This is a system architecture flowchart of Embodiment 1 of the present invention;

[0052] Appendix Figure 2 This is a system module connection block diagram of Embodiment 1 of the present invention;

[0053] Appendix Figure 3 This is a schematic diagram of the implementation process of the data preprocessing and fusion module in Embodiment 1 of the present invention;

[0054] Appendix Figure 4 This is a flowchart of the method in Embodiment 2 of the present invention. Detailed Implementation

[0055] To make the technical solution, the technical problem solved, and the technical effect of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with specific embodiments.

[0056] Example 1:

[0057] Combined with appendix Figure 1 , 2 3. This embodiment proposes a precise monitoring system for forest carbon sequestration based on multi-source data fusion, which includes:

[0058] The multi-source data acquisition module is used to collect satellite remote sensing data, UAV aerial photography data, ground monitoring sensor data, and meteorological data;

[0059] The data preprocessing and fusion module is used to preprocess the collected data and extract and combine relevant features through the feature layer and decision layer fusion method, which are used as input data for the carbon sink estimation model.

[0060] The model building and optimization module is used to build carbon sink estimation models by integrating convolutional neural networks, long short-term memory networks and random forest algorithms for different data characteristics, and to continuously optimize model performance through a dynamic data update mechanism.

[0061] The carbon sink estimation model is used to take the output of the data preprocessing and fusion module as input and output the carbon sink estimation result.

[0062] The results visualization and analysis module is used to display carbon sink estimation results in various visualization formats, and to study the impact of different factors on forest carbon sinks and predict future trends using sensitivity analysis and scenario simulation techniques.

[0063] The integration and management module is used to integrate hardware devices to build a monitoring platform, develop monitoring systems based on cloud computing and the Internet of Things, establish sound management systems, and ensure the operation of the monitoring system and data security.

[0064] In this embodiment, the multi-source data acquisition module specifically includes:

[0065] The satellite remote sensing data acquisition unit is used to acquire satellite remote sensing data using high-resolution optical satellites and radar satellites, and to obtain multispectral, hyperspectral and microwave remote sensing images.

[0066] The drone aerial photography data acquisition unit is used to acquire drone aerial photography data and obtain local images of forests using drones equipped with multispectral cameras, thermal infrared cameras and lidar.

[0067] The ground sensor data acquisition unit is used to collect ground monitoring sensor data using tree diameter growth and tree height measurement equipment, soil temperature and humidity sensors, soil carbon flux sensors, and weather stations.

[0068] Meteorological data acquisition unit, used to collect meteorological data in cooperation with meteorological departments or by using meteorological satellites.

[0069] Specifically, the satellite remote sensing data acquisition unit uses the WorldView-3 high-resolution optical satellite to acquire high-resolution forest texture information with a spatial resolution of 0.31 meters, uses multispectral data from 13 spectral bands of the Sentinel-2 satellite to extract vegetation indices, and uses RADARSAT-2 radar satellite data to acquire forest vertical structure parameters in cloudy and foggy weather. Satellite image data is acquired through data receiving stations or commercial remote sensing data service platforms, and the Sentinel-2 satellite data update cycle is 5 days.

[0070] For key monitoring areas and complex terrain areas, the UAV aerial photography data acquisition unit uses UAVs equipped with Micasense RedEdge-MX multispectral cameras, FLIRVueProR thermal infrared cameras and Velodyne VLP-16 lidar to conduct low-altitude operations at heights of 50-300 meters. The flight path is set using Pix4Dmapper flight path planning software to acquire image data with centimeter-level resolution.

[0071] In this embodiment, the data preprocessing and fusion module specifically includes:

[0072] The data preprocessing unit performs targeted processing on different types of acquired data, including radiometric, geometric, and atmospheric corrections for satellite remote sensing data; mosaicking and color correction for UAV aerial data; filtering, denoising, and outlier handling for ground monitoring sensor data; and data cleaning and format conversion for meteorological data. Specifically, it uses ENVI software for radiometric correction, ground control points for geometric correction with errors controlled within one pixel, and a 6S model for atmospheric correction; Pix4Dmapper software for image mosaicking to generate orthophoto maps and digital surface models, followed by color balancing and light equalization; Kalman filtering for data smoothing; and linear interpolation or a random forest-based missing value prediction algorithm for filling in missing data.

[0073] The data fusion unit employs a two-level fusion method: a feature layer and a decision layer. At the feature layer, various relevant features are extracted and combined from the preprocessed data, such as normalized difference vegetation index, enhanced vegetation index, texture features, single tree canopy area, canopy closure, three-dimensional structural features, tree growth rate, soil carbon content, and meteorological factors. At the decision layer, a carbon sink estimation model based on satellite remote sensing data, UAV data, and ground monitoring data is used to synthesize the estimation results of the carbon sink estimation model through a weighted average method, ultimately obtaining the carbon sink estimation result.

[0074] In this embodiment, the model building and optimization module specifically includes:

[0075] The model building unit is used to extract spatial features from satellite remote sensing data and UAV aerial photography data using convolutional neural networks, and to capture the temporal patterns of carbon sink changes using long short-term memory networks for ground monitoring sensor data and meteorological data. Then, the features of multi-source data are integrated through random forest algorithm to build a carbon sink estimation model. Specifically, the convolutional neural network is an improved ResNet-50 network with added residual connections and attention mechanisms.

[0076] The data preparation unit is used to collect ground-based measured carbon sink data and remote sensing / monitoring data from different climate zones and forest types, and is divided into training set, validation set and test set in a 7:2:1 ratio;

[0077] The model optimization unit is used to employ the Adam optimization algorithm with an initial learning rate of 0.001. It dynamically adjusts the batch size and number of iterations, evaluates the performance of the carbon sink estimation model in real time through the validation set, and saves the optimal model parameters.

[0078] The model monitoring unit is used to continuously monitor the operating status of the carbon sink estimation model, save the optimal model parameters, and ensure the accuracy and timeliness of the estimation results.

[0079] The timed update unit is used to periodically call the data preparation unit to collect ground-based measured carbon sink data and remote sensing / monitoring data from different climate zones and forest types, and to incrementally update the parameters of the carbon sink estimation model to ensure that the carbon sink estimation model adapts to the dynamic changes of the forest ecosystem.

[0080] In this embodiment, the result visualization and analysis module specifically includes:

[0081] The visualization unit provides a visual interface that presents carbon sequestration estimation results in the form of heat maps, 3D scenes, and time series charts. It intuitively displays the spatial distribution, vertical distribution characteristics, time-varying trends, and correlation with meteorological factors of carbon sequestration, allowing users to clearly and comprehensively understand carbon sequestration-related data.

[0082] The analysis and prediction unit is used to conduct in-depth research on the impact of different factors on forest carbon sinks using sensitivity analysis and scenario simulation techniques, and to predict future trends in forest carbon sinks. Through in-depth data mining and analysis, it assesses the dynamic changes and impacts of carbon sinks.

[0083] Specifically, the visualization unit provides a visualization interface developed based on the ArcGIS Online geographic information system platform. Users can view carbon sink-related data by clicking on any area on the interactive map. The analysis and prediction unit uses sensitivity analysis to study the impact of afforestation area, forest tending measures, and climate change factors on forest carbon sinks. Based on historical data and combined with future climate prediction scenarios of RCP4.5 and RCP8.5, it simulates the changing trend of forest carbon sinks over a specified future time period (e.g., 20 years).

[0084] In this embodiment, the integration and management module specifically includes:

[0085] The hardware integration unit is used to integrate data acquisition equipment, edge computing nodes, data servers, and visualization terminals to build a hardware architecture for forest carbon sink monitoring. Specifically, the data acquisition equipment includes a satellite signal receiving antenna, a UAV ground station, and a ground sensor gateway. The edge computing nodes use NVIDIA Jetson Xavier NX development boards, and the data servers use Dell PowerEdge R740xd, which are configured with 24 CPU cores, 128GB of memory, and 48TB of storage capacity.

[0086] The software development unit is used to develop a forest carbon sequestration monitoring software system based on cloud computing technology and using the Spring Boot + Vue.js technology stack. It implements data interaction between various components through RESTful API interfaces.

[0087] The management and maintenance unit is used to formulate data security management systems, establish regular equipment maintenance plans, and set up system log monitoring functions to provide dual protection for the stable operation of the monitoring system and the security of data from both institutional and technical perspectives.

[0088] Example 2:

[0089] Reference Appendix Figure 4 This embodiment proposes a precise monitoring method for forest carbon sinks based on multi-source data fusion, which is based on the system described in Embodiment 1 and includes the following steps:

[0090] S1. Collect satellite remote sensing data, UAV aerial photography data, ground monitoring sensor data, and meteorological data using a multi-source data acquisition module;

[0091] S2. The collected data is preprocessed using the data preprocessing and fusion module, and relevant features are extracted and combined using the feature layer and decision layer fusion method to provide input data for the carbon sink estimation model.

[0092] S3, the model building and optimization module, integrates convolutional neural networks, long short-term memory networks and random forest algorithms to build carbon sink estimation models for different data characteristics, and continuously optimizes model performance through a dynamic data update mechanism; the carbon sink estimation model obtains carbon sink estimation results based on the feature data output by the data preprocessing and fusion module.

[0093] S4. The Results Visualization and Analysis module displays the carbon sink estimation results in various visualization formats, and uses sensitivity analysis and scenario simulation techniques to study the impact of different factors on forest carbon sinks and predict future trends.

[0094] S5. Integration and Management Module: Integrates hardware devices to build a monitoring platform, develops a monitoring system based on cloud computing and the Internet of Things, establishes a sound management system, ensures the operation of the monitoring system and data security, and provides support for the entire process.

[0095] In summary, the forest carbon sequestration precision monitoring system and method based on multi-source data fusion of this invention integrates multi-source data such as satellite remote sensing, UAVs, ground sensors, and meteorological data. By utilizing data processing technology and machine learning algorithms, it overcomes the shortcomings of traditional monitoring methods, achieving high-precision, real-time monitoring and dynamic assessment of forest carbon sequestration. This provides accurate carbon sequestration data support for forest resource management departments to formulate scientific afforestation and forest cultivation policies, and for enterprises to participate in carbon trading.

[0096] The above specific examples illustrate the principles and implementation methods of the present invention in detail. These embodiments are merely for the purpose of helping to understand the core technical content of the present invention. Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principles of the present invention should fall within the patent protection scope of the present invention.

Claims

1. A precise monitoring system for forest carbon sequestration based on multi-source data fusion, characterized in that, It includes: The multi-source data acquisition module is used to collect satellite remote sensing data, UAV aerial photography data, ground monitoring sensor data, and meteorological data; The data preprocessing and fusion module is used to preprocess the collected data and extract and combine relevant features through the feature layer and decision layer fusion method, which are used as input data for the carbon sink estimation model. The model building and optimization module is used to build carbon sink estimation models by integrating convolutional neural networks, long short-term memory networks and random forest algorithms for different data characteristics, and to continuously optimize model performance through a dynamic data update mechanism. The carbon sink estimation model is used to take the output of the data preprocessing and fusion module as input and output the carbon sink estimation result. The results visualization and analysis module is used to display carbon sink estimation results in various visualization formats, and to study the impact of different factors on forest carbon sinks and predict future trends using sensitivity analysis and scenario simulation techniques. The integration and management module is used to integrate hardware devices to build a monitoring platform, develop monitoring systems based on cloud computing and the Internet of Things, establish sound management systems, and ensure the operation of the monitoring system and data security.

2. The forest carbon sequestration precision monitoring system based on multi-source data fusion according to claim 1, characterized in that, The multi-source data acquisition module specifically includes: The satellite remote sensing data acquisition unit is used to acquire satellite remote sensing data using high-resolution optical satellites and radar satellites, and to obtain multispectral, hyperspectral and microwave remote sensing images. The drone aerial photography data acquisition unit is used to acquire drone aerial photography data and obtain local images of forests using drones equipped with multispectral cameras, thermal infrared cameras and lidar. The ground sensor data acquisition unit is used to collect ground monitoring sensor data using tree diameter growth and tree height measurement equipment, soil temperature and humidity sensors, soil carbon flux sensors, and weather stations. Meteorological data acquisition unit, used to collect meteorological data in cooperation with meteorological departments or by using meteorological satellites.

3. The forest carbon sequestration precision monitoring system based on multi-source data fusion according to claim 1, characterized in that, The data preprocessing and fusion module specifically includes: The data preprocessing unit is used to perform targeted processing on different types of collected data, including radiometric correction, geometric correction and atmospheric correction for satellite remote sensing data, mosaicking and color correction for UAV aerial photography data, filtering and noise reduction and outlier processing for ground monitoring sensor data, and data cleaning and format conversion for meteorological data. The data fusion unit employs a two-level fusion method involving a feature layer and a decision layer. At the feature layer, various relevant features are extracted and combined from the preprocessed data. At the decision layer, a carbon sink estimation model based on satellite remote sensing data, UAV data, and ground monitoring data is used to synthesize the estimation results of the carbon sink estimation model through a weighted average method, ultimately yielding the carbon sink estimation result.

4. The forest carbon sequestration precision monitoring system based on multi-source data fusion according to claim 1, characterized in that, The model construction and optimization module specifically includes: The model building unit is used to extract spatial features from satellite remote sensing data and UAV aerial photography data using convolutional neural networks, and to capture the temporal patterns of carbon sink changes using long short-term memory networks for ground monitoring sensor data and meteorological data. Then, the features of multi-source data are integrated through random forest algorithm to build a carbon sink estimation model. The data preparation unit is used to collect ground-based measured carbon sink data and remote sensing / monitoring data from different climate zones and forest types, and is divided into training set, validation set and test set in a 7:2:1 ratio; The model optimization unit is used to employ the Adam optimization algorithm with an initial learning rate of 0.

001. It dynamically adjusts the batch size and number of iterations, evaluates the performance of the carbon sink estimation model in real time through the validation set, and saves the optimal model parameters. The model monitoring unit is used to continuously monitor the operating status of the carbon sink estimation model, save the optimal model parameters, and ensure the accuracy and timeliness of the estimation results. The timed update unit is used to periodically call the data preparation unit to collect ground-based measured carbon sink data and remote sensing / monitoring data from different climate zones and forest types, and to incrementally update the parameters of the carbon sink estimation model to ensure that the carbon sink estimation model adapts to the dynamic changes of the forest ecosystem.

5. A forest carbon sequestration precision monitoring system based on multi-source data fusion according to claim 4, characterized in that, The convolutional neural network used in the model building unit is an improved ResNet-50 network with added residual connections and attention mechanisms.

6. The forest carbon sequestration precision monitoring system based on multi-source data fusion according to claim 1, characterized in that, The results visualization and analysis module specifically includes: The visualization unit provides a visual interface that presents carbon sequestration estimation results in the form of heat maps, 3D scenes, and time series charts. It intuitively displays the spatial distribution, vertical distribution characteristics, time-varying trends, and correlation with meteorological factors of carbon sequestration, allowing users to clearly and comprehensively understand carbon sequestration-related data. The analysis and prediction unit is used to conduct in-depth research on the impact of different factors on forest carbon sinks using sensitivity analysis and scenario simulation techniques, and to predict future trends in forest carbon sinks. Through in-depth data mining and analysis, it assesses the dynamic changes and impacts of carbon sinks.

7. A forest carbon sequestration precision monitoring system based on multi-source data fusion according to claim 6, characterized in that, The visualization unit provides a visualization interface developed based on the ArcGIS Online geographic information system platform. Users can view carbon sequestration-related data by clicking on any area on the interactive map. The analysis and prediction unit uses sensitivity analysis to study the impact of afforestation area, forest tending measures, and climate change factors on forest carbon sinks. Based on historical data and combined with future climate prediction scenarios of RCP4.5 and RCP8.5, it simulates the changing trend of forest carbon sinks over a specified future time period.

8. A forest carbon sequestration precision monitoring system based on multi-source data fusion according to claim 1, characterized in that, The integration and management module specifically includes: The hardware integration unit is used to integrate data acquisition devices, edge computing nodes, data servers, and visualization terminals to build a hardware architecture for forest carbon sequestration monitoring. The software development unit is used to develop a forest carbon sequestration monitoring software system based on cloud computing technology and using the Spring Boot + Vue.js technology stack. It implements data interaction between various components through RESTful API interfaces. The management and maintenance unit is used to formulate data security management systems, establish regular equipment maintenance plans, and set up system log monitoring functions to provide dual protection for the stable operation of the monitoring system and the security of data from both institutional and technical perspectives.

9. A method for precise monitoring of forest carbon sequestration based on multi-source data fusion, characterized in that, Based on the system as described in claim 1, it includes the following steps: The system utilizes a multi-source data acquisition module to collect satellite remote sensing data, UAV aerial photography data, ground monitoring sensor data, and meteorological data. The collected data is preprocessed using a data preprocessing and fusion module, and relevant features are extracted and combined using a feature layer and decision layer fusion method to provide input data for the carbon sink estimation model. The model building and optimization module integrates convolutional neural networks, long short-term memory networks, and random forest algorithms to construct carbon sink estimation models for different data characteristics, and continuously optimizes model performance through a dynamic data update mechanism; the carbon sink estimation model obtains carbon sink estimation results based on the feature data output by the data preprocessing and fusion module. The results visualization and analysis module presents carbon sink estimation results in various visualization formats, and uses sensitivity analysis and scenario simulation techniques to study the impact of different factors on forest carbon sinks and predict future trends. The integration and management module integrates hardware devices to build a monitoring platform, develops a monitoring system based on cloud computing and the Internet of Things, establishes a sound management system, ensures the operation of the monitoring system and data security, and provides support for the entire process.

Images