River and lake ecological buffer zone carbon sink potential evaluation and repair construction system

Abstract

The invention relates to the technical field of ecological restoration and carbon sink evaluation, and discloses a river and lake ecological buffer zone carbon sink potential evaluation and restoration construction system which comprises a carbon sink potential evaluation module, an intelligent restoration construction module and a collaborative verification module. The carbon sink potential evaluation module is used for collecting soil and vegetation data and calculating the carbon sink potential; the intelligent restoration construction module executes ecological restoration based on a carbon sink potential evaluation result; and the collaborative verification module verifies the repair effect through multi-source data fusion, and returns an optimization instruction to the carbon sink potential evaluation module and the intelligent repair construction module. The vegetation coverage index and the root biomass density are acquired by using a multispectral sensor array, a carbon sink model is constructed by combining the organic carbon content of soil, dynamic weight adjustment is introduced, and the weight is updated according to aerial survey data of an unmanned aerial vehicle in each quarter, so that a more accurate and more economical carbon sink potential evaluation basis can be provided for ecological restoration, and the ecological restoration efficiency is improved. And resource waste and invalid repair are avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological restoration and carbon sink assessment, and in particular to a carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones. Background Art

[0002] As an important part of the terrestrial and hydrological transition zone, river and lake ecological buffer zones play a key role in the ecosystem. Their rich vegetation and diverse soil types give them significant carbon sink functions, which can absorb and fix carbon dioxide in the atmosphere and mitigate climate change. With the increasing demand for ecological restoration, intelligent construction technology has gradually been applied to the field of river and lake ecological restoration. However, existing technologies mostly focus on a single restoration method and lack coordinated optimization with carbon sink potential assessment.

[0003] Currently, assessments of the carbon sequestration potential of river and lake ecological buffer zones primarily rely on costly laboratory analysis or remote sensing technology. While laboratory analysis methods are highly accurate, they require extensive sample collection and analysis, making them expensive and time-consuming, making them difficult to scale up. Remote sensing technology, while covering a wide range, has low resolution and struggles to accurately capture microscopic changes and details within the buffer zones. Furthermore, traditional restoration techniques are generally inefficient, costly, and inflexible, making precise restoration difficult. This results in suboptimal ecological restoration outcomes and prevents the full realization of the carbon sequestration potential of river and lake ecological buffer zones. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a carbon sequestration potential assessment and restoration construction system for river and lake ecological buffer zones, which solves the problems of low efficiency, high cost, poor flexibility and inability to fully tap the carbon sequestration potential of river and lake ecological buffer zones caused by traditional restoration construction technologies.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A carbon sequestration potential assessment and restoration construction system for river and lake ecological buffer zones, including a carbon sequestration potential assessment module, an intelligent restoration construction module, and a collaborative verification module; The carbon sequestration potential assessment module is used to collect soil and vegetation data and calculate carbon sequestration potential, and its output results are transmitted to the smart restoration construction module; The smart restoration construction module performs ecological restoration based on the carbon sequestration potential assessment results, and its operation process data is fed back to the collaborative verification module in real time; The collaborative verification module verifies the restoration effect through multi-source data fusion, and transmits the optimization instructions back to the carbon sink potential assessment module and the intelligent restoration construction module to form a closed-loop control system.

[0006] By adopting the above technical solution: using a multispectral sensor array to collect vegetation cover index (VCI) and root biomass density (RBD), combining soil organic carbon (SOC) content to construct a carbon sink model (CSP), and introducing dynamic weight adjustment, the weights are updated every quarter based on drone aerial survey data, which achieves the effect of reducing the carbon sink assessment error to within 5%, which is significantly better than the traditional method (error is 15%); at the same time, the use of equipment such as portable stainless steel samplers and multispectral sensor arrays optimizes the data collection process, reduces assessment costs, and improves assessment efficiency, thereby providing a more accurate and economical basis for carbon sink potential assessment for ecological restoration, avoiding waste of resources and ineffective restoration.

[0007] Preferably, the carbon sequestration potential assessment module includes: Soil sampling unit: A portable stainless steel sampler equipped with a GNSS positioning module and a mechanism with adjustable depth between 0 cm and 100 cm is used to collect soil samples in a gridded manner with spacing no greater than 50 meters. The soil organic carbon content (SOC) is determined by an elemental analyzer after the soil sampling unit collects the sample. Vegetation monitoring unit: Deploy a multispectral sensor array, set up collection channels in the 550nm visible light, 780nm near infrared and 1650nm shortwave infrared bands, and monitor the vegetation coverage index (VCI) and root biomass density (RBD) at a density of one node per 100 square meters. The root biomass density (RBD) is calculated by combining the root diameter data obtained by the vegetation monitoring unit with a preset biomass-diameter relationship model. The root biomass density (RBD) is calculated as follows: A biomass-diameter relationship model was established based on the historical sample library, satisfying the formula RBD=k·D 2.3 , where D is the average root diameter, unit: mm, k is the vegetation type correction factor, where herbaceous plants k = 0.15, shrubs k = 0.35, and trees k = 0.72; The root diameter distribution map was obtained by inverting data from a multispectral sensor in the 1650nm band, and the RBD value was calculated by inputting the biomass-diameter relationship model. Data processing unit: Construct a carbon sink model CSP = α × SOC + β × VCI + γ × RBD, where α = 0.45 ± 0.05, β = 0.35 ± 0.03, and γ = 0.2 ± 0.02, and output a carbon sink potential spatial distribution map; the vegetation coverage index VCI is calculated by the reflectance difference between the multispectral sensor in the 550nm and 780nm bands, satisfying the following: Among them, NIR is the reflectance value of the 780nm band, and Red is the reflectance value of the 550nm band.

[0008] Preferably, the intelligent repair construction module includes: Tracked repair robot: Equipped with a terrain-adaptive suspension system, a multi-degree-of-freedom robotic arm, and a swappable tool library including a planting shovel, seeder, and cover plate, it generates an obstacle-avoiding path based on LiDAR terrain data and an improved A* algorithm. Distributed environmental sensor network: uses the LoRaWAN protocol for networking, with a node spacing of 50±5 meters, integrating a capacitive soil moisture sensor, a glass electrode pH sensor, and a DS18B20 temperature sensor, with a sampling frequency of 15 minutes per time; Control Center: The core restoration area is divided according to the carbon sequestration potential distribution map. The core restoration area is defined as the area where the carbon sequestration potential index CSP is lower than the 60th percentile value. The planting density is dynamically allocated to 8-12 plants / m 2 , and generate a restoration plan including replanting density and organic fertilizer application amount.

[0009] Preferably, the collaborative verification module includes: UAV-assisted assessment unit: Equipped with a hyperspectral imager and a robotic arm sampling device, it flies at an altitude of 30 meters with a tolerance of ±5 meters and an 80% heading overlap. It generates orthophotos with a 2-centimeter resolution and automatically collects samples when it detects a soil pH value less than 5.5 or greater than 8.5. Data fusion unit: A Kalman filter algorithm is used to fuse the UAV NDVI data with the ground sensor data, with a temporal and spatial matching error of ≤±5%. The implementation of the Kalman filter algorithm includes the following steps: State modeling: Define the state vector x k =[NDVI aav ,NDVI ground ] Τ , where NDVI uav Normalized Difference Vegetation Index (NDVI) obtained by drone ground For ground sensor data; Prediction steps: Among them, F k is the state transfer matrix, constructed according to the spatiotemporal alignment relationship between the UAV and the ground sensor, u k To control the input UAV attitude adjustment parameters, B k is the control matrix; Update steps: Among them, H k is the observation matrix, R k is the observation noise covariance, z k is the actual observed value; Output fusion result: the final NDVI value is taken as the state vector x k The weighted average is calculated by the Kalman gain K kDynamic adjustment; two-level verification unit: short-term verification monitors the soil organic carbon increment ΔSOC, requiring ΔSOC ≥ 0.5% within 3 months after restoration; long-term verification uses drone aerial surveys to measure the annual growth rate of vegetation net primary productivity NPP ≥ 10%. Secondary restoration will be triggered in areas that do not meet the standards.

[0010] Preferably, the vegetation monitoring unit further includes: Disease identification subunit: Using the reflectance data of the multispectral sensor in the 780nm band, the system calculates that when the average reflectance decreases by more than 20% over three consecutive days, it is identified as a leaf disease and the disease type is matched against the vegetation type database. Carbon sequestration prediction subunit: This subunit uses an LSTM neural network model. The model structure includes an input layer, a hidden layer, and an output layer. The number of hidden layer units is 64, and the time step is set to 12 months. The input parameters include the historical vegetation coverage index (VCI), the average monthly temperature (T), and the atmospheric CO2 concentration (C). The output is the predicted carbon sequestration value (FCP) for the next 6 months. The calculation process of the LSTM unit satisfies the following formula: f t =σ(W f ·[h t-1 ,x t ]+b f ) i t =σ(W i ·[h t-1 ,x t ]+b i ) o t =σ(W o ·[h t-1 ,x t ]+b o ) h t =o t ⊙tanh(C t ) Among them, x t =[VCI t ,T t , C t ] is the input vector at time t, VCI t is the vegetation coverage index of the current month, T t is the monthly mean temperature, C t is the atmospheric CO2 concentration; the mean square error loss function is used in model training, the optimizer is Adam, and the learning rate is set to 0.001; Abnormal response subunit: When the monitoring data deviation is greater than 3σ, the drone is called for fixed-point review.

[0011] Preferably, the path planning algorithm of the control center satisfies: Among them, E i is the energy consumption coefficient positively correlated with soil hardness, E i The linear calculation is performed by the measured value of the soil hardness sensor, E i =0.05×H c , H c is the measured value of the soil hardness sensor, unit: kPa; D i is the moving distance; S d is the soil disturbance penalty term negatively correlated with organic carbon content, S d =max(0,1.2-0.01×SOC), SOC is the percentage of soil organic carbon content; Improved A* algorithm execution, real-time receiving of soil hardness and SOC data from distributed environmental sensor network, dynamic update of E i With S d value; When the sensor network detects that the monthly decline rate of the carbon sequestration potential index ΔCSP exceeds 15%, the operating radius of the repair robot is automatically adjusted to within 500 meters.

[0012] Preferably, the repair robot comprises: Adaptive operation subunit: switch to no-till sowing mode when soil moisture > 30% and detect root density > 5kg / m 3 Reduce the downward force of the robot arm by 20%; Blockchain evidence storage sub-unit: uploads operation trajectory data to the blockchain node in real time; Multi-machine collaborative sub-unit: 3 robots form an operating group, maintain a 5-meter distance based on the virtual structure method, and share real-time position data through the mesh network.

[0013] Preferably, the data processing unit further includes: Dynamic weight adjustment subunit: The model weights are updated every quarter based on drone aerial survey data. If the annual NPP growth rate is <10%, the parameters are adjusted according to Δα = +0.02×ln(1+ΔSOC) and Δβ = -0.01×ΔVCI, where ΔSOC is the increment of organic carbon content measured by the soil sampling unit within the quarter, calculated by comparing the historical sampling data of the same area; ΔVCI is the difference in vegetation coverage index between the current quarter and the previous quarter of the vegetation monitoring unit; after the weight adjustment, the carbon sink model recalculates the core restoration area division results; Pollution response subunit: When the soil petroleum hydrocarbon concentration is detected to be >50mg / kg, an attenuation coefficient of 0.7 is applied to the CSP value.

[0014] Preferably, the UAV-assisted evaluation unit is extended to include: Multi-drone formation sub-unit: 3 drones form a triangle formation, maintaining a 10-meter spacing and operating synchronously; Emergency sampling subunit: When a sudden pollution incident is detected, the heading overlap is increased to 90% for intensified monitoring; Manual review subunit: When the deviation between the drone and ground data is greater than 3σ, the manual on-site verification process is triggered.

[0015] Preferably, an extended function module is further included, and the extended function module includes: Digital twin management unit: Builds a 3D model based on BIM technology and synchronizes sensor data every second; Federated learning optimization unit: The local carbon sequestration model parameters of each restoration system are uploaded to the central server through a secure aggregation protocol. The aggregated global model parameters are then distributed to each system, with an update cycle of every 30 days. Knowledge graph database unit: associates and stores soil improvement plans, plant growth characteristics, and historical restoration case data.

[0016] The present invention provides a carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones. It has the following beneficial effects: 1. The present invention uses a multispectral sensor array to collect vegetation cover index (VCI) and root biomass density (RBD), combines it with soil organic carbon (SOC) content to construct a carbon sink model (CSP), and introduces dynamic weight adjustment. The weights are updated quarterly based on drone aerial survey data, achieving the effect of reducing the carbon sink assessment error to within 5%, which is significantly better than the traditional method (error is 15%); at the same time, the use of portable stainless steel samplers and multispectral sensor arrays and other equipment optimizes the data collection process, reduces assessment costs, and improves assessment efficiency, thereby providing a more accurate and economical basis for carbon sink potential assessment for ecological restoration, avoiding waste of resources and ineffective restoration.

[0017] 2. The robot in this invention is equipped with a terrain-adaptive suspension system, a multi-degree-of-freedom robotic arm, and a replaceable tool library. It generates obstacle avoidance paths based on lidar terrain data and an improved A* algorithm, achieving efficient operation. The control center divides the core restoration area according to the carbon sink potential distribution map, dynamically allocates planting density, and generates restoration plans, achieving an average daily restoration area of ​​480m 2 , the operating efficiency is increased by 300% compared with traditional manual restoration, which can quickly and accurately repair ecologically damaged areas, reduce soil disturbance and vegetation damage, and improve restoration quality and ecological recovery speed.

[0018] 3. In the present invention, the drone-assisted evaluation unit and the data fusion unit fuse multi-source data to verify the restoration effect. The digital twin management unit builds a three-dimensional model based on BIM technology, synchronizes sensor data every second, and the federated learning optimization unit updates the global carbon sink model parameters every 30 days, and transmits the optimization instructions back to the evaluation and restoration module, forming a closed-loop control system, which achieves the effect of short-term verification of soil organic carbon increment (ΔSOC ≥ 0.5%) and long-term verification of vegetation net primary productivity (NPP annual growth rate ≥ 10%). Secondary restoration is triggered in areas that do not meet the standards, and the restoration effect can be monitored in real time, and the restoration strategy can be adjusted in time to achieve coordinated optimization of carbon sink potential assessment and restoration construction, thereby improving the overall effect and sustainability of ecological restoration. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a system framework diagram of a carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to the present invention; Figure 2 This is a schematic diagram of a carbon sequestration potential assessment module of a carbon sequestration potential assessment and restoration construction system for river and lake ecological buffer zones according to the present invention; Figure 3 This is a schematic diagram of an intelligent restoration construction module of a carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to the present invention; Figure 4 This is a schematic diagram of a collaborative verification module of a carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to the present invention; Figure 5 This is a schematic diagram of the expanded functional modules of a carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to the present invention; Figure 6 This is a schematic diagram of the system flow for carbon sequestration potential assessment and restoration construction of river and lake ecological buffer zones according to the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present invention.

[0021] Please see the attached Figure 1 -Attached Figure 6 , an embodiment of the present invention provides a carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones, including a carbon sink potential assessment module, an intelligent restoration construction module, and a collaborative verification module; The carbon sequestration potential assessment module is used to collect soil and vegetation data and calculate carbon sequestration potential, and its output is transmitted to the smart restoration construction module; The smart restoration construction module performs ecological restoration based on the carbon sequestration potential assessment results, and its operation process data is fed back to the collaborative verification module in real time; The collaborative verification module verifies the restoration effect through multi-source data fusion, and sends the optimization instructions back to the carbon sink potential assessment module and the intelligent restoration construction module to form a closed-loop control system.

[0022] Specifically, the carbon sequestration potential assessment module serves as the front end of the system, responsible for collecting and analyzing soil and vegetation data in the river and lake ecological buffer zone, and calculating the carbon sequestration potential of the area; the intelligent restoration construction module performs specific ecological restoration construction tasks based on the data and suggestions provided by the carbon sequestration potential assessment module, optimizes the restoration process, and improves restoration efficiency; the collaborative verification module verifies and evaluates the effects of the restoration construction, ensures the reliability and scientific nature of the restoration effect through multi-source data fusion technology, and transmits the optimization instructions back to the evaluation and restoration modules to form a closed-loop control.

[0023] Please see the attached Figure 2 , the carbon sequestration potential assessment module includes: Soil sampling unit: A portable stainless steel sampler equipped with a GNSS positioning module and a depth-adjustable mechanism between 0 cm and 100 cm is used to collect soil samples in a gridded pattern with spacing no greater than 50 m. The soil organic carbon content (SOC) is determined by an elemental analyzer after the soil sampling unit collects the sample. Vegetation Monitoring Unit: Deploy a multispectral sensor array, set up collection channels in the 550nm visible light, 780nm near-infrared, and 1650nm short-wave infrared bands, and monitor the vegetation coverage index (VCI) and root biomass density (RBD) at a density of one node per 100 square meters. The root biomass density (RBD) is calculated by combining the root diameter data obtained by the vegetation monitoring unit with a preset biomass-diameter relationship model. The root biomass density (RBD) is calculated as follows: A biomass-diameter relationship model was established based on the historical sample library, satisfying the formula RBD=k·D 2.3 , where D is the average root diameter, unit: mm, k is the vegetation type correction factor, where herbaceous plants k = 0.15, shrubs k = 0.35, and trees k = 0.72; The root diameter distribution map was obtained by inverting data from a multispectral sensor in the 1650nm band, and the RBD value was calculated by inputting the biomass-diameter relationship model. Data processing unit: Construct a carbon sink model CSP = α × SOC + β × VCI + γ × RBD, where α = 0.45 ± 0.05, β = 0.35 ± 0.03, and γ = 0.2 ± 0.02, and output a carbon sink potential spatial distribution map; the vegetation coverage index VCI is calculated by the reflectance difference between the multispectral sensor in the 550nm and 780nm bands, satisfying the following: Among them, NIR is the reflectance value of the 780nm band, and Red is the reflectance value of the 550nm band.

[0024] Specifically, the soil sampling unit provides basic data for carbon sequestration potential assessment by collecting soil samples and measuring the soil organic carbon (SOC) content; the vegetation monitoring unit assesses the contribution of vegetation to carbon sequestration by monitoring the vegetation cover index (VCI) and root biomass density (RBD); the data processing unit integrates soil and vegetation data by constructing a carbon sequestration model, and outputs a spatial distribution map of carbon sequestration potential.

[0025] Please see the attached Figure 3 , the smart repair construction module includes: Tracked repair robot: Equipped with a terrain-adaptive suspension system, a multi-degree-of-freedom robotic arm, and a swappable tool library including a planting shovel, seeder, and cover plate, it generates an obstacle-avoiding path based on LiDAR terrain data and an improved A* algorithm. Distributed environmental sensor network: uses the LoRaWAN protocol for networking, with a node spacing of 50±5 meters, integrating a capacitive soil moisture sensor, a glass electrode pH sensor, and a DS18B20 temperature sensor, with a sampling frequency of 15 minutes per time; Control Center: The core restoration area is divided according to the carbon sequestration potential distribution map. The core restoration area is defined as the area where the carbon sequestration potential index CSP is lower than the 60th percentile value. The planting density is dynamically allocated to 8-12 plants / m 2 , and generate a restoration plan including replanting density and organic fertilizer application amount.

[0026] Specifically, the tracked restoration robot serves as the core equipment of the intelligent restoration construction module. The tracked restoration robot is responsible for performing specific ecological restoration tasks on complex terrain. The distributed environmental sensor network monitors the environmental parameters during the restoration process in real time, provides data support for the restoration construction, and ensures the scientific nature and effectiveness of the restoration measures. The LoRaWAN protocol is used for networking to ensure stable transmission of sensor data and low-power operation. The sensor node spacing is 50±5 meters, which can cover the entire restoration area and provide high-density monitoring data. The control center serves as the command center of the intelligent restoration construction module. It is responsible for formulating restoration plans based on the carbon sequestration potential assessment results and dynamically adjusting restoration strategies.

[0027] Please see the attached Figure 4 , the collaborative verification module includes: UAV-assisted assessment unit: Equipped with a hyperspectral imager and a robotic arm sampling device, it flies at an altitude of 30 meters with a tolerance of ±5 meters and an 80% heading overlap. It generates orthophotos with a 2-centimeter resolution and automatically collects samples when it detects a soil pH value less than 5.5 or greater than 8.5. Data fusion unit: A Kalman filter algorithm is used to fuse the UAV NDVI data with the ground sensor data, with a temporal and spatial matching error of ≤±5%. The implementation of the Kalman filter algorithm includes the following steps: State modeling: Define the state vector x k =[NDVI aav ,NDVI ground ] Τ , where NDVI uav Normalized Difference Vegetation Index (NDVI) obtained by drone ground For ground sensor data; Prediction steps: Among them, F k is the state transfer matrix, constructed according to the spatiotemporal alignment relationship between the UAV and the ground sensor, u k To control the input UAV attitude adjustment parameters, B k is the control matrix; Update steps: Among them, H k is the observation matrix, R k is the observation noise covariance, z k is the actual observed value; Output fusion result: the final NDVI value is taken as the state vector x k The weighted average is calculated by the Kalman gain K k Dynamic adjustment; two-level verification unit: short-term verification monitors the soil organic carbon increment ΔSOC, requiring ΔSOC ≥ 0.5% within 3 months after restoration; long-term verification uses drone aerial surveys to measure the annual growth rate of vegetation net primary productivity NPP ≥ 10%. Secondary restoration will be triggered in areas that do not meet the standards.

[0028] Specifically, the UAV-assisted assessment unit uses a UAV equipped with a hyperspectral imager and a robotic arm sampling device to collect high-resolution images and soil samples of the restoration area, and monitor the soil and vegetation conditions in real time; the data fusion unit fuses the hyperspectral data (such as NDVI) obtained by the UAV with the real-time data of the ground sensor network to improve the accuracy and reliability of the data; the two-level verification unit comprehensively evaluates the effect of ecological restoration through short-term and long-term verification mechanisms, and conducts secondary restoration of areas that do not meet the standards.

[0029] Please see the attached Figure 2 , the vegetation monitoring unit also includes: Disease identification subunit: Using the reflectance data of the multispectral sensor in the 780nm band, the system calculates that when the average reflectance decreases by more than 20% over three consecutive days, it is identified as a leaf disease and the disease type is matched against the vegetation type database. Carbon sequestration prediction subunit: This subunit uses an LSTM neural network model. The model structure includes an input layer, a hidden layer, and an output layer. The number of hidden layer units is 64, and the time step is set to 12 months. The input parameters include the historical vegetation coverage index (VCI), the average monthly temperature (T), and the atmospheric CO2 concentration (C). The output is the predicted carbon sequestration value (FCP) for the next 6 months. The calculation process of the LSTM unit satisfies the following formula: f t =σ(W f ·[h t-1 ,x t ]+b f ) i t =σ(W i ·[h t-1 ,x t ]+b i ) o t =σ(W o ·[h t-1 ,x t ]+b o ) h t =o t ⊙tanh(C t ) Among them, x t =[VCI t ,T t , C t ] is the input vector at time t, VCI t is the vegetation coverage index of the current month, T t is the monthly mean temperature, C t is the atmospheric CO2 concentration; the mean square error loss function is used in model training, the optimizer is Adam, and the learning rate is set to 0.001; Abnormal response subunit: When the monitoring data deviation is greater than 3σ, the drone is called for fixed-point review.

[0030] Specifically, the disease identification subunit identifies leaf diseases of vegetation by analyzing multispectral sensor data, providing timely early warning for disease prevention and control; the carbon sequestration prediction subunit uses the LSTM neural network model to predict the carbon sequestration amount in the next 6 months, providing a scientific basis for carbon sink management; the function of the abnormal response subunit is to call the drone for fixed-point review when abnormalities occur in the monitoring data, to ensure the accuracy and reliability of the monitoring data.

[0031] Please see the attached Figure 3 , the path planning algorithm of the control center satisfies: Among them, E i is the energy consumption coefficient positively correlated with soil hardness, E i The linear calculation is performed by the measured value of the soil hardness sensor, E i =0.05×H c , H c is the measured value of the soil hardness sensor, unit: kPa; D i is the moving distance; S d is the soil disturbance penalty term negatively correlated with organic carbon content, S d =max(0,1.2-0.01×SOC), SOC is the percentage of soil organic carbon content; Improved A* algorithm execution, real-time receiving of soil hardness and SOC data from distributed environmental sensor network, dynamic update of E i With S d value; When the sensor network detects that the monthly decline rate of the carbon sequestration potential index ΔCSP exceeds 15%, the operating radius of the repair robot is automatically adjusted to within 500 meters.

[0032] Specifically, when the improved A* algorithm is executed, it plays the role of receiving the soil hardness and SOC data of the distributed environmental sensor network in real time, and then dynamically updates E according to the real-time data. i With S d value, ensuring that path planning can adapt to environmental changes; by considering the impact of soil hardness on energy consumption, the algorithm can plan a path with lower energy consumption and reduce the energy consumption of the restoration operation; introducing the soil disturbance penalty term S d When planning the path, the disturbance to the soil with high organic carbon content is minimized to protect the soil's carbon sequestration capacity; at the same time, the algorithm dynamically updates parameters based on real-time soil hardness and SOC data to ensure that the path planning can adapt to environmental changes and improve the flexibility and adaptability of the restoration operation.

[0033] Please refer to Appendix 3, the repair robot includes: Adaptive operation subunit: switch to no-till sowing mode when soil moisture > 30% and detect root density > 5kg / m3 Reduce the downward force of the robot arm by 20%; Blockchain evidence storage sub-unit: uploads operation trajectory data to the blockchain node in real time; Multi-machine collaborative sub-unit: 3 robots form an operating group, maintain a 5-meter distance based on the virtual structure method, and share real-time position data through the mesh network.

[0034] Specifically, the adaptive operation sub-unit automatically adjusts the operation mode according to the real-time monitoring of soil moisture and root density, thereby reducing damage to the soil and vegetation; the blockchain evidence storage sub-unit uploads the operation trajectory data of the repair robot to the blockchain node in real time, thereby ensuring the non-tamperability and traceability of the operation data; the multi-machine collaboration sub-unit improves the repair efficiency and operation quality through the collaborative operation of multiple robots.

[0035] Please see the attached Figure 4 , the data processing unit also includes: Dynamic weight adjustment subunit: Model weights are updated quarterly based on drone aerial survey data. If the annual NPP growth rate is less than 10%, the parameters are adjusted according to Δα = +0.02 × ln(1 + ΔSOC) and Δβ = -0.01 × ΔVCI, where ΔSOC is the increment of organic carbon content measured in the soil sampling unit within the quarter, calculated by comparing historical sampling data in the same area; ΔVCI is the difference in vegetation coverage index between the current quarter and the previous quarter for the vegetation monitoring unit. After the weight adjustment, the carbon sequestration model recalculates the core restoration area demarcation results. Pollution response subunit: When the soil petroleum hydrocarbon concentration is detected to be greater than 50 mg / kg, a 0.7 attenuation coefficient is applied to the CSP value.

[0036] Specifically, the dynamic weight adjustment subunit plays a role in dynamically adjusting the weights of the carbon sink model according to UAV aerial survey data and environmental changes, thereby optimizing the accuracy of the carbon sink potential assessment; the pollution response subunit plays a role in real-time monitoring and responding to soil pollution, adjusting the carbon sink potential assessment results, and ensuring the accuracy of the assessment; through the quarterly update and parameter adjustment of the dynamic weight adjustment subunit, the carbon sink model can be optimized according to the latest environmental data and changing trends, thereby improving the accuracy and adaptability of the carbon sink potential assessment; through the real-time monitoring and attenuation coefficient adjustment of the pollution response subunit, soil pollution can be discovered in a timely manner, and the impact of pollution on carbon sink potential can be reflected by adjusting the CSP value.

[0037] Please see the attached Figure 4 , the UAV-assisted evaluation unit expansion includes: Multi-drone formation sub-unit: 3 drones form a triangle formation, maintaining a 10-meter spacing and operating synchronously; Emergency sampling subunit: When a sudden pollution incident is detected, the heading overlap is increased to 90% for intensified monitoring; Manual review subunit: When the deviation between the drone and ground data is greater than 3σ, the manual on-site verification process is triggered.

[0038] Specifically, the multi-machine formation sub-unit improves monitoring efficiency and data coverage through the coordinated operation of multiple drones. The triangular formation can quickly cover a large area, reduce the time of a single operation, and improve monitoring efficiency; the emergency sampling sub-unit plays the role of quickly responding and encrypting monitoring when a sudden pollution incident is detected. When a sudden pollution incident occurs, it can quickly adjust the monitoring strategy, intensify monitoring, and obtain detailed data in a timely manner; the manual review sub-unit plays the role of triggering a manual on-site verification process when the deviation between the drone and ground data exceeds 3σ, to ensure the accuracy and reliability of the data. Through manual on-site verification, data deviations can be discovered and corrected in a timely manner to ensure the accuracy and reliability of the monitoring data.

[0039] Please see the attached Figure 5 According to the above-mentioned carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones, an extended function module is also included. The extended function module includes: Digital twin management unit: Builds a 3D model based on BIM technology and synchronizes sensor data every second; Federated learning optimization unit: The local carbon sequestration model parameters of each restoration system are uploaded to the central server through a secure aggregation protocol. The aggregated global model parameters are then distributed to each system, with an update cycle of every 30 days. Knowledge graph database unit: associates and stores soil improvement plans, plant growth characteristics, and historical restoration case data.

[0040] Specifically, the digital twin management unit uses BIM technology to build a three-dimensional model, synchronizes sensor data in real time, realizes dynamic monitoring and visual management of river and lake ecological buffer zones, and constructs a three-dimensional model of the river and lake ecological buffer zone based on BIM (Building Information Modeling) technology to intuitively display terrain, vegetation, soil and other information; the federated learning optimization unit uses federated learning technology to optimize the carbon sequestration model to improve the accuracy and generalization ability of the model; the knowledge graph database unit plays the role of storing and associating soil improvement plans, plant growth characteristics and historical restoration case data, providing a scientific basis for ecological restoration.

[0041] Example 1: Dynamic Assessment of Carbon Sink Potential and Weight Optimization Method 1. Technical Solution 1. Soil sampling: Soil samples at a depth of 0-100 cm were collected using a portable stainless steel sampler at a 50-m grid interval, and the soil organic carbon (SOC) content was determined using an elemental analyzer.

[0042] 2. Vegetation monitoring: Deploy a multispectral sensor array to monitor spectral data in the 550nm, 780nm, and 1650nm bands; calculate the vegetation cover index (VCI): The root diameter (D) was inverted by combining the 1650nm band; according to the coefficients of different vegetation types (herbs, shrubs, and trees) (the k values ​​were 0.15, 0.35, and 0.72 respectively), the formula RBD = k·D 2.3 Calculate biomass density (RBD).

[0043] 3. Dynamic Model: Construct a carbon sink potential model: CSP = 0.45 × SOC + 0.35 × VCI + 0.2 × RBD. Dynamically adjust weights quarterly based on drone aerial survey data. If the annual growth rate of vegetation net primary productivity (NPP) is less than 10%, update the model coefficients according to the formulas Δα = +0.02ln(1 + ΔSOC) and Δβ = -0.01ΔVCI.

[0044] 4. Pollution correction: When the concentration of petroleum hydrocarbons in the soil is detected to be greater than 50 mg / kg, the carbon sink potential value (CSP) is multiplied by an attenuation coefficient of 0.7 for correction.

[0045] 2. Parameter Optimization Basis The initial weight coefficients α, β, and γ are determined by regression analysis of 100 sets of historical data, and the model determination coefficient R 2 reached 0.89; The root index of 2.3 is based on the allometric growth law of biomass and diameter in botany. The weight adjustment formula is optimized through gradient descent to ensure that the model error rate is less than 5%.

[0046] 3. Implementation Effect Verification A 10-hectare area was selected in the Taihu Lake buffer zone, and the evaluation results of the traditional laboratory testing method (GB9834-88) and the present embodiment were compared. The verification results are shown in the following table: index Traditional methods This embodiment Test standards Single point assessment time 48 hours 2 hours ISO18400-206:2018 Carbon sink value error rate 15％ 4.2％ RMSE (measured vs predicted) Pollution response sensitivity Not detected Real-time correction HJ605-2011 Conclusion: This example improves assessment efficiency by 24 times, reduces the error rate of carbon sink values ​​by 73%, and realizes the dynamic correction function of pollution impact for the first time.

[0047] Example 2: Terrain-adaptive repair robot collaborative operation system 1. Technical Solution 1. Path planning: The objective function of the improved A* algorithm is The energy consumption coefficient Ei = 0.05Hc (Hc is soil hardness in kPa), and the soil organic carbon (SOC) content correlation coefficient Sd = max(0, 1.2-0.01SOC). The robot receives sensor data in real time to update its parameters to adapt to different terrain and soil conditions.

[0048] 2. Adaptive operation: When the soil moisture exceeds 30%, the robot switches to no-till mode; when the root density exceeds 5kg / m 3 When the robot arm is in the air, the downward force of the robot arm is automatically reduced by 20% to reduce damage to the soil and vegetation.

[0049] 3. Multi-machine collaboration: Three robots form a triangular formation and share position data through the Mesh network to ensure that the formation spacing error is less than 0.3 meters, achieving efficient collaborative operation.

[0050] 2. Parameter Optimization Basis The energy consumption coefficient is 0.05, which is based on 200 sets of soil hardness and energy consumption experimental data. The fitting determination coefficient R 2 reached 0.92; The penalty 1.2 threshold corresponds to a critical organic carbon content of 12% to prevent excessive disturbance of high carbon sink soils; The 5-meter formation spacing was determined through fluid dynamics simulation to reduce airflow interference with robot operations during operation.

[0051] 3. Implementation Effect Verification In the buffer zone of Poyang Lake, traditional manual restoration and robotic restoration were compared (NY / T3343-2018). The verification results are shown in the following table: index Manual repair This embodiment Test standards <![CDATA[Daily working area (m 2 )]]> 120 480 Planting coverage rate per unit time Increase in soil compaction 18％ 5％ GB / T50123-2019 Path planning efficiency - 35% increase Actual path / theoretical optimal path Conclusion: The robot's operating efficiency increased by 300%, soil disturbance was reduced by 72%, and path planning was more in line with the actual terrain, significantly improving the efficiency and quality of ecological restoration.

[0052] Example 3: Restoration Effect Verification System Based on Multi-Source Data Fusion 1. Technical Solution 1. Drone Verification: Three drones were flown in a triangular formation, 10 meters apart and with 90% heading overlap, to generate 2cm resolution orthophotos. Ground sensor data was fused using a Kalman filter to ensure a data fusion error of less than or equal to 5%.

[0053] 2. Two-tier verification: Short-term verification: Within three months after restoration, the change in soil organic carbon (SOC) content (ΔSOC) must reach 0.5% or above (reference standard GB9834-88); Long-term verification: Through drone aerial surveys, the annual growth rate of vegetation net primary productivity (NPP) must reach 10% or above (reference standard MOD17A3H).

[0054] 3. Federated Learning: Aggregate the model parameters of each system through a homomorphic encryption protocol every 30 days and update the global carbon sink model to improve the accuracy and adaptability of the model.

[0055] 2. Parameter Optimization Basis The UAV formation spacing is 10 meters based on image stitching error analysis to ensure that the stitching error is less than 1 pixel offset; The Kalman filter noise covariance matrix was calibrated using 100 sets of measured data to improve the accuracy of data fusion; The federated learning cycle is set to 30 days to balance model timeliness and communication overhead.

[0056] 3. Implementation Effect Verification Comparing traditional manual inspection and drone verification in the Dongting Lake buffer zone (GJB5891-2006), the verification results are shown in the following table: index Manual verification This embodiment Test standards Verification time for 100 hectares 15 days 6 hours Full area coverage time Abnormal soil pH detection rate 82％ 98％ Random sampling of 200 points Accuracy improved after model update - 12.7％ Cross-validation F1 value Conclusion: The verification efficiency was improved by 60 times, the detection rate of soil pH anomalies increased by 16%, and the model adaptive optimization significantly enhanced the generalization ability, providing strong support for the accurate verification of ecological restoration effects.

[0057] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones, characterized in that: It includes carbon sequestration potential assessment module, smart restoration construction module and collaborative verification module; The carbon sequestration potential assessment module is used to collect soil and vegetation data and calculate carbon sequestration potential, and its output results are transmitted to the smart restoration construction module; The smart restoration construction module performs ecological restoration based on the carbon sequestration potential assessment results, and its operation process data is fed back to the collaborative verification module in real time; The collaborative verification module verifies the restoration effect through multi-source data fusion, and transmits the optimization instructions back to the carbon sink potential assessment module and the intelligent restoration construction module to form a closed-loop control system.

2. A river and lake ecological buffer zone carbon sink potential assessment and restoration construction system according to claim 1, characterized in that: The carbon sequestration potential assessment module includes: Soil sampling unit: A portable stainless steel sampler equipped with a GNSS positioning module and a mechanism with adjustable depth between 0 cm and 100 cm is used to collect soil samples in a gridded manner with spacing no greater than 50 meters. The soil organic carbon content (SOC) is determined by an elemental analyzer after the soil sampling unit collects the sample. Vegetation monitoring unit: Deploy a multispectral sensor array, set up collection channels in the 550nm visible light, 780nm near infrared and 1650nm shortwave infrared bands, and monitor the vegetation coverage index (VCI) and root biomass density (RBD) at a density of one node per 100 square meters. The root biomass density (RBD) is calculated by combining the root diameter data obtained by the vegetation monitoring unit with a preset biomass-diameter relationship model. The root biomass density (RBD) is calculated as follows: A biomass-diameter relationship model was established based on the historical sample library, satisfying the formula RBD=k·D 2.3 , where D is the average root diameter, unit: mm, k is the vegetation type correction factor, where herbaceous plants k = 0.15, shrubs k = 0.35, and trees k = 0.72; The root diameter distribution map was obtained by inverting data from a multispectral sensor in the 1650nm band, and the RBD value was calculated by inputting the biomass-diameter relationship model. Data processing unit: Construct a carbon sink model CSP = α × SOC + β × VCI + γ × RBD, where α = 0.45 ± 0.05, β = 0.35 ± 0.03, and γ = 0.2 ± 0.02, and output a carbon sink potential spatial distribution map; the vegetation coverage index VCI is calculated by the reflectance difference between the multispectral sensor in the 550nm and 780nm bands, satisfying the following: Among them, NIR is the reflectance value of the 780nm band, and Red is the reflectance value of the 550nm band.

3. A carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to claim 1, characterized in that: The smart repair construction module includes: Tracked repair robot: Equipped with a terrain-adaptive suspension system, a multi-degree-of-freedom robotic arm, and a replaceable tool library including a planting shovel, seeder, and cover plate, it generates obstacle avoidance paths based on LiDAR terrain data and an improved A* algorithm. Distributed environmental sensor network: Using the LoRaWAN protocol, with a node spacing of 50±5 meters, it integrates a capacitive soil moisture sensor, a glass electrode pH sensor, and a DS18B20 temperature sensor, with a sampling frequency of 15 minutes. Control Center: The core restoration area is divided according to the carbon sequestration potential distribution map. The core restoration area is defined as the area where the carbon sequestration potential index CSP is lower than the 60th percentile value. The planting density is dynamically allocated to 8-12 plants / m 2 , and generate a restoration plan including replanting density and amount of organic fertilizer application.

4. A carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to claim 1, characterized in that: The collaborative verification module includes: UAV-assisted assessment unit: Equipped with a hyperspectral imager and a robotic arm sampling device, it flies at an altitude of 30 meters with a tolerance of ±5 meters and an 80% heading overlap. It generates orthophotos with a 2-centimeter resolution and automatically collects samples when it detects a soil pH value less than 5.5 or greater than 8.

5. Data fusion unit: A Kalman filter algorithm is used to fuse the UAV NDVI data with the ground sensor data, with a temporal and spatial matching error of ≤±5%. The implementation of the Kalman filter algorithm includes the following steps: State modeling: Define the state vector x k =[NDVI aav ,NDVI ground ] Τ , where NDVI uav Normalized Difference Vegetation Index (NDVI) obtained by drone ground For ground sensor data; Prediction steps: Among them, F k is the state transfer matrix, constructed according to the spatiotemporal alignment relationship between the UAV and the ground sensor, u k To control the input UAV attitude adjustment parameters, B k is the control matrix; Update steps: Among them, H k is the observation matrix, R k is the observation noise covariance, z k is the actual observed value; Output fusion result: the final NDVI value is taken as the state vector x k The weighted average is calculated by the Kalman gain K k Dynamic adjustment; Two-level verification unit: short-term verification monitors the soil organic carbon increment ΔSOC, requiring ΔSOC ≥ 0.5% within 3 months after restoration; long-term verification uses drone aerial surveys to measure the annual growth rate of vegetation net primary productivity NPP ≥ 10%. Secondary restoration will be triggered in areas that do not meet the standards.

5. The carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to claim 2 is characterized in that: The vegetation monitoring unit also includes: Disease identification subunit: Using the reflectance data of the multispectral sensor in the 780nm band, the system calculates that when the average reflectance decreases by more than 20% over three consecutive days, it is identified as a leaf disease and the disease type is matched against the vegetation type database. Carbon sequestration prediction subunit: This subunit uses an LSTM neural network model. The model structure includes an input layer, a hidden layer, and an output layer. The number of hidden layer units is 64, and the time step is set to 12 months. The input parameters include the historical vegetation coverage index (VCI), the average monthly temperature (T), and the atmospheric CO2 concentration (C). The output is the predicted carbon sequestration value (FCP) for the next 6 months. The calculation process of the LSTM unit satisfies the following formula: f t =σ(W f ·[h t-1 ,x t ]+b f ) i t =σ(W i ·[h t-1 ,x t ]+b i ) the t =σ(W o ·[h t-1 ,x t ]+b o ) h t =o t ⊙tanh(C t ) Among them, x t =[VCI t ,T t , C t ] is the input vector at time t, VCI t is the vegetation coverage index of the current month, T t is the monthly mean temperature, C t is the atmospheric CO2 concentration; the mean square error loss function is used in model training, the optimizer is Adam, and the learning rate is set to 0.001; Abnormal response subunit: When the monitoring data deviation is greater than 3σ, the drone is called for fixed-point review.

6. A carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to claim 3, characterized in that: The path planning algorithm of the control center satisfies: Among them, E i is the energy consumption coefficient positively correlated with soil hardness, E i The linear calculation is performed by the measured value of the soil hardness sensor, E i =0.05×H c , H c is the measured value of the soil hardness sensor, unit: kPa; D i is the moving distance; S d is the soil disturbance penalty term negatively correlated with organic carbon content, S d =max(0,1.2-0.01×SOC), SOC is the percentage of soil organic carbon content; Improved A* algorithm execution, real-time receiving of soil hardness and SOC data from distributed environmental sensor network, dynamic update of E i With S d value; When the sensor network detects that the monthly decline rate of the carbon sequestration potential index ΔCSP exceeds 15%, the operating radius of the repair robot is automatically adjusted to within 500 meters.

7. The carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to claim 3 is characterized in that: The repair robot comprises: Adaptive operation subunit: switch to no-till sowing mode when soil moisture > 30% and detect root density > 5kg / m 3 Reduce the downward force of the robot arm by 20%; Blockchain evidence storage sub-unit: uploads operation trajectory data to the blockchain node in real time; Multi-machine collaborative sub-unit: 3 robots form an operating group, maintain a 5-meter distance based on the virtual structure method, and share real-time position data through the mesh network.

8. The carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to claim 2 is characterized in that: The data processing unit further includes: Dynamic weight adjustment subunit: The model weights are updated every quarter based on drone aerial survey data. If the annual NPP growth rate is <10%, the parameters are adjusted according to Δα = +0.02×ln(1+ΔSOC) and Δβ = -0.01×ΔVCI, where ΔSOC is the increment of organic carbon content measured by the soil sampling unit within the quarter, calculated by comparing the historical sampling data of the same area; ΔVCI is the difference in vegetation coverage index between the current quarter and the previous quarter of the vegetation monitoring unit; after the weight adjustment, the carbon sink model recalculates the core restoration area division results; Pollution response subunit: When the soil petroleum hydrocarbon concentration is detected to be >50mg / kg, an attenuation coefficient of 0.7 is applied to the CSP value.

9. The carbon sink potential assessment and restoration construction system for river and lake ecological buffer zones according to claim 4 is characterized in that: The UAV-assisted evaluation unit expansion includes: Multi-drone formation sub-unit: 3 drones form a triangle formation, maintaining a 10-meter spacing and operating synchronously; Emergency sampling subunit: When a sudden pollution incident is detected, the heading overlap is increased to 90% for intensified monitoring; Manual review subunit: When the deviation between the drone and ground data is greater than 3σ, the manual on-site verification process is triggered.

10. A system for carbon sink potential assessment and restoration construction of river and lake ecological buffer zones according to any one of claims 1 to 9, characterized in that: It also includes an extended function module, which includes: Digital twin management unit: Builds a 3D model based on BIM technology and synchronizes sensor data every second; Federated learning optimization unit: The local carbon sequestration model parameters of each restoration system are uploaded to the central server through a secure aggregation protocol. The aggregated global model parameters are then distributed to each system, with an update cycle of every 30 days. Knowledge graph database unit: associates and stores soil improvement plans, plant growth characteristics, and historical restoration case data.

Images