Forestry carbon sink dynamic accounting method, device, equipment, storage medium and program product

Abstract

The application discloses a forestry carbon sink dynamic accounting method, device, equipment, storage medium and program product, relates to the technical field of forestry carbon sink accounting, and in the application, project information of a forestry carbon sink accounting project to be implemented in a current accounting period is collected from project declaration materials, and project characteristics of the forestry carbon sink accounting project are identified therefrom, the actual content of a key accounting factor of the forestry carbon sink accounting project is further determined based on the project characteristics, and finally, the forestry carbon sink accounting is performed according to an accounting scheme obtained by using the actual content of the key accounting factor. In this way, compared with manual maintenance which is difficult to accurately and comprehensively determine the accounting core factors of different accounting periods, the actual content of the comprehensive and accurate key accounting factor is determined by identifying the project characteristics of the forestry carbon sink accounting project in the current accounting period, so that the accurate forestry carbon sink accounting is performed according to the accounting scheme obtained by using the actual content of the key accounting factor.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of forestry carbon sink accounting, and particularly relates to a forestry carbon sink dynamic accounting method based on machine learning, a forestry carbon sink dynamic accounting device based on machine learning, a forestry carbon sink dynamic accounting equipment based on machine learning, a storage medium and a computer program product. BACKGROUND

[0002] When long-term accounting of forestry carbon sink is performed, the core accounting factors of different accounting periods change, and when dynamic accounting of forestry carbon sink is performed, many key factors need to be considered to achieve accurate accounting. These factors are interrelated and jointly affect the accuracy and reliability of the final accounting result. At present, manual maintenance of the core accounting factors of different accounting periods often has the problem of inaccuracy and incompleteness, which makes it difficult to guarantee the accuracy of the final actual accounting scheme.

[0003] The above content is only used to assist in understanding the technical solutions of the present application and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0004] The main purpose of the present application is to provide a forestry carbon sink dynamic accounting method based on machine learning, a forestry carbon sink dynamic accounting device based on machine learning, a forestry carbon sink dynamic accounting equipment based on machine learning, a storage medium and a computer program product, which aims to solve the technical problem of inaccurate accounting when dynamic accounting of forestry carbon sink is performed.

[0005] To achieve the above-mentioned purpose, the present application provides a forestry carbon sink dynamic accounting method based on machine learning, which comprises:

[0006] determining a forestry carbon sink accounting project to be implemented in the current accounting period, collecting project information of the forestry carbon sink accounting project from project declaration materials of the forestry carbon sink accounting project;

[0007] identifying project characteristics of the forestry carbon sink accounting project from the project information based on machine learning; wherein the project characteristics include project type, project scale, and regional terrain and environmental characteristics of the forestry carbon sink accounting project;

[0008] determining actual content of key accounting factors of the forestry carbon sink accounting project according to the project characteristics, and performing forestry carbon sink accounting according to an accounting scheme obtained by using the actual content of the key accounting factors.

[0009] In an embodiment, the step of determining the actual content of the key accounting factors of the forestry carbon sink accounting project according to the project characteristics comprises:

[0010] determining a first actual content of carbon pool, baseline and disturbance type in the key accounting factors of the forestry carbon sink accounting project according to the project type of the forestry carbon sink accounting project;

[0011] determining a second actual content of monitoring target in the key accounting factors of the forestry carbon sink accounting project according to the project scale and the regional topography where the project is located;

[0012] determining a third actual content of accounting model and parameters in the key accounting factors of the forestry carbon sink accounting project according to the environmental characteristics of the region where the project is located.

[0013] In an embodiment, the step of determining the actual content of the key accounting factors of the forestry carbon sink accounting project according to the project characteristics comprises:

[0014] determining the carbon pool range and priority of the forestry carbon sink accounting project according to the first actual content;

[0015] determining the monitoring method and accuracy and frequency of the forestry carbon sink accounting project according to the second actual content;

[0016] determining the biomass model and accounting parameters of the forestry carbon sink accounting project according to the third actual content;

[0017] developing a baseline scenario scheme and taking the baseline scenario scheme after removing uncertainty sources as an accounting scheme according to the carbon pool range and priority, the monitoring method and accuracy and frequency, the biomass model and accounting parameters of the forestry carbon sink accounting project.

[0018] In an embodiment, the step of determining the actual content of the key accounting factors of the forestry carbon sink accounting project according to the project characteristics further comprises:

[0019] taking the project characteristics as input characteristics, and inferring the output label corresponding to the input characteristics based on a pre-trained selection model, wherein the output label is a label value corresponding to the actual content of the key accounting factors of the forestry carbon sink accounting project.

[0020] In an embodiment, the step of taking the project characteristics as input characteristics, and inferring the output label corresponding to the input characteristics based on a pre-trained selection model comprises:

[0021] obtaining historical project information and historical accounting schemes of historical forestry carbon sink accounting projects, and determining historical actual content of the key accounting factors of the forestry carbon sink accounting project from the historical accounting schemes;

[0022] A training set is constructed based on historical project information and historical actual content of the historical forestry carbon sink accounting project, and a selected model is trained based on the training set.

[0023] In an embodiment, the method further comprises:

[0024] In the forestry carbon sink accounting, a target forest area to be accounted for and a forest growth stage in which the target forest area is located are determined.

[0025] Multi-modal carbon sink related data of the target forest area are obtained through a three-dimensional monitoring network of the target forest area.

[0026] The carbon sink amount of the target forest area is estimated based on the forest growth stage in which the target forest area is located and the multi-modal carbon sink related data of the target forest area.

[0027] In addition, to achieve the above-mentioned purpose, the present application also proposes a forestry carbon sink dynamic accounting device based on machine learning, which comprises:

[0028] The preparation module is configured to determine a forestry carbon sink accounting project to be implemented in a current accounting period, and collect project information of the forestry carbon sink accounting project from project declaration materials of the forestry carbon sink accounting project.

[0029] The identification module is configured to identify project features of the forestry carbon sink accounting project from the project information based on machine learning, wherein the project features include a project type, a project scale, a regional terrain where the project is located, and an environmental feature of the project.

[0030] The generation module is configured to determine actual content of a key accounting factor of the forestry carbon sink accounting project according to the project features, and perform forestry carbon sink accounting according to an accounting scheme obtained by using the actual content of the key accounting factor.

[0031] In addition, to achieve the above-mentioned purpose, the present application also proposes a forestry carbon sink dynamic accounting device based on machine learning, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the above-mentioned forestry carbon sink dynamic accounting method based on machine learning.

[0032] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer readable storage medium, and a computer program is stored on the storage medium, wherein the computer program is executed by a processor to implement the steps of the above-mentioned forestry carbon sink dynamic accounting method based on machine learning.

[0033] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which comprises a computer program, and the computer program realizes the steps of the forestry carbon sink dynamic accounting method based on machine learning when executed by a processor.

[0034] The one or more technical solutions provided in the present application have at least the following technical effects:

[0035] In the present application, the project information of the forestry carbon sink accounting project to be implemented in the current accounting period is collected from the project declaration materials, and the project characteristics of the forestry carbon sink accounting project are identified therefrom. Further, the actual content of the key accounting factors of the forestry carbon sink accounting project is determined based on the project characteristics, and finally the forestry carbon sink accounting is performed according to the accounting scheme obtained by using the actual content of the key accounting factors. In this way, compared with manually maintaining the accounting core factors of different accounting periods which is difficult to accurately and comprehensively determine, the actual content of the key accounting factors is determined by identifying the project characteristics of the forestry carbon sink accounting project in the current accounting period, so that the accurate forestry carbon sink accounting is performed according to the accounting scheme obtained by using the actual content of the key accounting factors. BRIEF DESCRIPTION OF DRAWINGS

[0036] The accompanying drawings incorporated in and forming a part of the specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the application.

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without creative labor.

[0038] Figure 1 A flowchart is provided for the first embodiment of the forestry carbon sink dynamic accounting method based on machine learning of the present application;

[0039] Figure 2 An application diagram is provided for the first embodiment of the forestry carbon sink dynamic accounting method based on machine learning of the present application;

[0040] Figure 3 A flowchart is provided for the second embodiment of the forestry carbon sink dynamic accounting method based on machine learning of the present application;

[0041] Figure 4 A module structure diagram of the forestry carbon sink dynamic accounting device based on machine learning of the present application is provided;

[0042] Figure 5A device structure schematic diagram of a hardware running environment involved in a forestry carbon sink dynamic accounting method based on machine learning in the embodiments of the present application.

[0043] The object implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0044] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and do not limit the present application.

[0045] In order to better understand the technical solutions of the present application, the following will be described in detail in combination with the drawings and specific embodiments of the specification.

[0046] It should be noted that the execution subject of the present embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a forestry carbon sink dynamic accounting device based on machine learning, etc. The following takes the forestry carbon sink dynamic accounting device based on machine learning as an example to describe the present embodiment and each of the following embodiments.

[0047] Based on this, the present embodiment provides a forestry carbon sink dynamic accounting method based on machine learning, referring to Figure 1 and Figure 2 , Figure 1 A flowchart of a first embodiment of the forestry carbon sink dynamic accounting method based on machine learning of the present application, Figure 2 An application schematic diagram provided by the first embodiment of the forestry carbon sink dynamic accounting method based on machine learning of the present application.

[0048] In the present embodiment, the forestry carbon sink dynamic accounting method based on machine learning comprises steps S10-S30:

[0049] Step S10, determining a forestry carbon sink accounting project to be implemented in the current accounting period, collecting project information of the forestry carbon sink accounting project from project declaration materials of the forestry carbon sink accounting project;

[0050] Forestry carbon sink refers to the process that forests in forest land absorb CO2 through photosynthesis and store it in the form of biomass (trees, soil, etc.) for a long time, thereby reducing the concentration of greenhouse gases. The accounting goal is to determine the increment of carbon sink in a certain period of time (such as how many tons of CO2 are absorbed per year) in a specific area or project.

[0051] Since the dynamic changes of carbon sinks need to be continuously monitored and evaluated, in order to ensure the carbon sink effect and accuracy of the project, the forestry carbon sink accounting project usually involves multiple accounting periods. The accounting periods include 1, annual accounting: suitable for projects that need high-frequency monitoring, such as high-risk areas or high-interference projects. 2, interval accounting such as every 3-5 years: suitable for most forest projects, which can balance the monitoring cost and data accuracy. 3, long-term accounting: some projects may need longer accounting periods, such as long-term forest management projects or carbon sequestration projects.

[0052] The project information of the forestry carbon sink accounting project refers to a set of data used to describe the key dimensions of the nature of the project, the implementation conditions, the ecological environment and the management strategy, which can be collected from the project declaration materials of the forestry carbon sink accounting project. In the project declaration materials, there are generally project design documents, feasibility study reports, geographic spatial data, field survey data, historical and real-time monitoring data, etc. The project information of the forestry carbon sink accounting project can be read from the project declaration materials. The project declaration materials can also include monitoring reports, land use maps, satellite remote sensing images, meteorological data, soil distribution maps, operation plans, and budget tables. In this embodiment, the method for obtaining the project information of the forestry carbon sink accounting project is not limited.

[0053] In step S20, the project characteristics of the forestry carbon sink accounting project are identified from the project information based on machine learning; wherein the project characteristics include the project type, the project scale and the topography of the region where the project is located, and the environmental characteristics of the region where the project is located.

[0054] In identifying the project characteristics of the forestry carbon sink accounting project from the project information based on machine learning: text classification methods such as using pre-trained language models (such as BERT, RoBERTa) or traditional machine learning models (such as SVM, random forest) can be used to classify the project description to obtain text keywords such as afforestation, forest management, REDD+, and project goals such as carbon trading and ecological protection. Regression analysis. Linear regression, random forest regression or neural network models can be used to predict the project area, including keywords in the project description such as large-scale and small-area, forest coverage area in remote sensing images, and regional area in geographic data. Spatial analysis can be performed using GIS tools to extract topographic features such as elevation, slope, aspect and relief from DEM data, and finally output topographic feature indicators such as flat land, mountainous land and complex terrain. Environmental factor extraction can be performed using environmental data such as climate data and soil data to extract features such as climate zone, annual precipitation, annual mean temperature and soil type, and finally output environmental feature indicators such as humid climate, arid climate and fertile soil. In this embodiment, the method for identifying the project characteristics of the forestry carbon sink accounting project from the project information based on machine learning is not limited.

[0055] In step S30, the actual content of the key accounting factors of the forestry carbon sink accounting project is determined according to the project characteristics, and the forestry carbon sink accounting is performed according to the accounting scheme obtained by using the actual content of the key accounting factors.

[0056] Since the project characteristics contain the key information of the forestry carbon sink accounting project, the actual content of the key accounting factors of the forestry carbon sink accounting project can be determined according to the project characteristics, that is, the actual content of the key accounting factors of the project is determined, which covers all aspects of the forestry carbon sink accounting and can fully and accurately reflect the carbon sink capacity of the project. For example, according to the project characteristics, the project type and applicable methodology in the key accounting factors of the forestry carbon sink accounting project can be determined, the monitoring means and plot design in the key accounting factors of the forestry carbon sink accounting project can be determined according to the geographical characteristics and area of the project, the buffer mechanism and uncertainty management in the key accounting factors of the forestry carbon sink accounting project can be determined according to the historical disturbance and risk factor design of the project area, and so on.

[0057] In a feasible implementation, step S30 can include steps S30A10-S30A30:

[0058] In step S30A10, the first actual content of the carbon pool, baseline and disturbance type in the key accounting factors of the forestry carbon sink accounting project is determined according to the project type of the forestry carbon sink accounting project.

[0059] In step S30A20, the second actual content of the monitoring target in the key accounting factors of the forestry carbon sink accounting project is determined according to the project scale and the terrain of the region where the project is located.

[0060] In step S30A30, the third actual content of the accounting model and parameters in the key accounting factors of the forestry carbon sink accounting project is determined according to the environmental characteristics of the region where the project is located.

[0061] When conducting dynamic accounting of forest carbon sinks, key accounting factors include 1. the scope and completeness of carbon pools. That is, which carbon pools are included in the accounting (aboveground biomass, belowground biomass, litter, dead wood, soil organic carbon). Ignoring important carbon pools (such as soil organic carbon or dead wood) will underestimate or overestimate the net carbon sink of the project / activity (if some pools are net sources). For example, soil carbon loss after harvesting or fire may be ignored; accounting only for the aboveground part of trees will ignore the carbon in roots and litter accumulation. It is crucial to choose an accounting scope that is consistent with the project goals, ecosystem type, and carbon pools affected by the activity. 2. Project boundaries. That is, clear and accurate, fixed and verifiable geographical boundaries (geographical coordinates, area). Changes in boundaries over time (such as natural expansion, addition of newly planted areas, reduction of areas due to land use change). Unclear or inaccurate boundaries (such as low-resolution remote sensing images, inaccurate field surveys) can lead to errors in the area of the accounting area. Failure to track and record boundary changes in a timely manner can lead to the inclusion of non-project areas in the accounting or the omission of areas that should be included, directly affecting the attribution and total amount of carbon storage estimates. 3. Baseline scenario. That is, the trajectory of changes in carbon storage over time without the implementation of project activities. The baseline is the reference point for calculating the additional emission reduction of the project. Setting too optimistic (high baseline) or too pessimistic (low baseline) will distort the true increment of carbon sink benefits. Underestimating the baseline means overestimating the project's emission reduction; overestimating the baseline underestimates the value of the project. A reasonable baseline needs to be based on historical evidence (land use history, management practices), regional average level, model prediction and clear and reasonable assumptions. The uncertainty of the baseline is usually the main source of uncertainty of the entire accounting. 4. Monitoring methods, frequency and accuracy. The choice of method is ground measurement (plot), remote sensing (such as optical, radar), model simulation or a combination of the above methods. The monitoring frequency is the time interval at which the carbon pool changes are measured (such as 1 year, 3 years, 5 years). The measurement accuracy includes the tool accuracy of field measurement, operator level, sample analysis accuracy (laboratory) and the spatial, spectral, temporal resolution and interpretation accuracy of remote sensing. For the monitoring methods used for data collection, low-cost remote sensing or sparse plot coverage may result in serious bias (failure to capture small-scale heterogeneity). Too low a measurement frequency will miss changes in disturbance events (fire, windfall) or rapid growth stages, resulting in distortion of the carbon storage change curve. Therefore, high-precision, appropriate frequency and reasonable method monitoring is the key to obtaining reliable carbon storage change data. 5. Biomass models and expansion / conversion factors. Including biomass equations, root-shoot ratios, carbon contents, basic densities, etc. used to convert measured values (such as diameter at breast height, tree height) to biomass and carbon storage. Mismatched models / parameters (such as using temperate models to estimate tropical tree species), outdated parameters (not updated in a timely manner), and regional differences in parameters (using universal parameters instead of localized parameters) can result in systematic errors. For example, overestimating growth models will make the accounting results larger, and it is necessary to use localized models for specific regions, specific tree species or forest ages reasonably.6. Land use, land-use change, and disturbance. Natural or human disturbances occurring within and outside the project area (wildfires, pests and diseases, windstorms, droughts, logging, encroachment, etc.) and their impacts on carbon stocks (leakage or project-wide losses) can lead to overestimation of actual carbon stocks if disturbances are not monitored or their losses are underestimated, seriously compromising the accuracy of the accounting results. Clear and conservative disturbance monitoring and reporting mechanisms and leakage estimation methods are required, and corresponding losses should be deducted from the total carbon sink or deposited into a buffer pool. 7. Quantifying uncertainty. Identify the sources of uncertainty in each link (sampling error, measurement error, model error, baseline uncertainty) and use appropriate methods (such as Monte Carlo simulation and error propagation analysis) to quantify the total uncertainty. Ignoring uncertainty will produce erroneous and distorted precision. Quantifying uncertainty is a key indicator of the reliability of the results. Accurately estimating uncertainty facilitates decision-making (such as setting the buffer pool size) and guides how to optimize subsequent monitoring to reduce key sources of uncertainty. 8. Temporal dynamics and baseline consistency. Carbon stock changes due to project activities (tree planting, fertilization, reduced logging) and natural processes (growth and aging) are non-linear. Furthermore, baseline scenarios need to simulate dynamic changes over time (e.g., the succession of secondary forests). Using static models or ignoring the dynamic effects of project activities (e.g., the short-term accelerating effects of fertilization) can lead to unrealistic estimates for a specific time period. Baseline scenarios need to reflect both natural trends over time and anticipated human activities (e.g., potential conversion of forestland to other uses under land market pressure). Inconsistent baselines can introduce additional systematic biases. 9. Persistence Risk and Accounting Period Selection. Forestry carbon sinks carry the potential risk of reversal (the release of sequestered carbon back into the atmosphere due to disturbance). The length of the accounting period (e.g., 20, 40, 100 years) influences the assessment of carbon sink effectiveness. Ignoring reversal risks or failing to consider maintenance costs / risks after the end of the accounting period may overestimate the long-term climate benefits of carbon sinks. Dynamic accounting should continuously assess risks (through monitoring and buffer pool mechanisms) and clearly state the time constraints when reporting net carbon sinks. An accounting period that is too short may fail to capture the potential and risks of the project throughout its life cycle.

[0062] In this embodiment, the carbon pool, baseline, and disturbance type among the key accounting factors of the forestry carbon sink accounting project are determined based on the project type of the forestry carbon sink accounting project and used as the first actual content. For example, when the project type is new afforestation / reforestation, the carbon pool among the key accounting factors is full carbon pool accounting, the baseline is unforested land / degraded forest land, and the disturbance type is to focus on the impact of afforestation activities; when the project type is forest management / improved management, the carbon pool among the key accounting factors is aboveground / belowground biomass and soil carbon, the baseline is the existing forest stand natural growth baseline, and the disturbance type is conventional management measures and natural disturbance; when the project type is to reduce deforestation and degradation, the carbon pool among the key accounting factors is the key aboveground biomass, the baseline is the historical deforestation rate baseline, and the disturbance type is human encroachment and fire risk.

[0063] In this embodiment, the monitoring objectives of the key accounting factors for the forestry carbon sink accounting project are determined based on the project scale and the terrain of the project area, and are used as the second actual content. For example, when the project scale and the terrain of the project area are small and simple, the monitoring objectives of the key accounting factors are mainly based on high-precision ground sampling. When the project scale and the terrain of the project area are large and complex, the monitoring objectives of the key accounting factors are based on a combination of remote sensing and ground sampling, with the boundaries accurately surveyed by GIS.

[0064] In this embodiment, the accounting model and parameters for the key accounting factors of the forestry carbon sink accounting project are determined based on the environmental characteristics of the project's location, and serve as the third actual content. For example, if the project's location is a high-interference risk zone, the key accounting factor's accounting model is a high-frequency monitoring and high-buffer pool model, with anti-interference parameters. If the project's location is a stable climate zone, the key accounting factor's accounting model is a general growth model, with default parameters.

[0065] The purpose of dynamic accounting is to reflect as closely as possible the changes in carbon stocks in the project / activity area relative to the baseline scenario due to changes in project implementation or management. On this basis, the above-mentioned key accounting factors are not isolated, they interact with each other and cumulatively affect the final accounting results. For example, errors in the biomass model will be introduced into the remote sensing estimation based on the model, which will ultimately affect the carbon stock calculation. Therefore, it is necessary to fully cover the carbon pool; accurately define and monitor; adopt a transparent, verifiable, conservative and reasonable baseline setting method; adopt the best feasible monitoring technology combination, use locally verified models and parameters that match the tree species / age class; strictly monitor and quantify internal and external interference and leakage; conduct uncertainty assessments for all key steps, and report the total uncertainty range.

[0066] In another feasible implementation, step S30 may include:

[0067] Based on the first practical content, determine the carbon pool scope and priority of forestry carbon sink accounting projects;

[0068] According to the second practical content, determine the monitoring method, accuracy and frequency of the forestry carbon sink accounting project;

[0069] According to the third practical content, determine the biomass model and accounting parameters of the forestry carbon sink accounting project;

[0070] Based on the carbon pool scope and priority, monitoring methods and accuracy, frequency, biomass model and accounting parameters of the forestry carbon sink accounting project, a baseline scenario plan is formulated and the baseline scenario plan after removing the uncertainty sources is used as the accounting plan.

[0071] After determining the first, second and third actual contents, further, the carbon pool range and priority of the forestry carbon sink accounting project can be determined according to the first actual content. Specifically, the project goal of the new afforestation / reafforestation project usually includes restoring forests and increasing carbon sinks, the project area is usually non-forest land or degraded forest land, the carbon pool range includes aboveground biomass, underground biomass, litter, dead wood and soil organic carbon, and the priority is soil organic carbon > aboveground biomass > underground biomass > litter > dead wood. The project goal of the forest management / improved management project can include optimizing forest resource management and improving carbon sink potential, the project area is usually an existing forest area, the carbon pool range includes aboveground biomass, underground biomass and soil organic carbon, and the priority is aboveground biomass > underground biomass > soil organic carbon. The project goal of the REDD+ project can include optimizing forest resource management and improving carbon sink potential, the project area is usually an existing forest area, the carbon pool range includes aboveground biomass, litter and dead wood, and the priority is aboveground biomass > litter > dead wood.

[0072] After determining the first, second and third actual contents, further, the monitoring method and precision and frequency of the forestry carbon sink accounting project can also be determined according to the second actual content, that is, the optimal monitoring technology and frequency selection strategy is determined according to the project scale (small area / large area) and the terrain complexity (simple / complex). Specifically, according to the project scale and the terrain of the region where the project is located, that is, the project scale and the terrain complexity, the scale category and the terrain complexity category are determined. In an embodiment, if the project scale is small area and the terrain is simple: the monitoring technology is mainly ground plot method (such as setting fixed sample plots, using high-precision measuring equipment), according to the project type and the growth characteristics of the tree species, a medium frequency (such as once every 2-3 years) monitoring frequency can be usually set, wherein the ground plot method includes setting fixed sample plots, using tools such as dendrometers and growth cones for field measurement. If the project scale is small area but the terrain is complex, the monitoring technology adopts ground plot method (but the terrain influence needs to be considered, and the number of sample plots may need to be increased or auxiliary low-altitude aerial photography by unmanned aerial vehicle is adopted), and a medium frequency (once every 2-3 years) monitoring frequency is set. If the project scale is medium area, the monitoring technology is set to combine remote sensing (such as high-resolution satellite images or unmanned aerial vehicle images) and ground sample plot verification (stratified sampling), and a medium frequency (once every 2-3 years) monitoring frequency is adopted. If the project scale is large area and the terrain is simple (which can be regarded as a kind of E2), the monitoring technology is mainly remote sensing (satellite remote sensing, such as Sentinel-2, Landsat, etc.), and a small amount of ground sample plot verification is adopted, because the cost of large-area ground investigation is high, so a lower frequency (such as once every 3-5 years) monitoring frequency is set. If the project scale is large area and the terrain is complex, the monitoring technology: high-precision remote sensing (such as LiDAR combined with multispectral and high-resolution images) combined with ground sample plot verification (using GIS technology to accurately arrange sample plots to ensure covering different terrains and vegetation types), boundary processing adopts GIS technology to accurately survey the boundary, including using high-precision GPS or Beidou to recheck the boundary, and the monitoring frequency is determined according to the availability and cost of remote sensing data and the requirements of the project: usually once every 2-4 years. If the project is in a high-interference risk area, the frequency can be appropriately increased.

[0073] After determining the first, second and third actual contents, further, the biomass model and accounting parameters of the forestry carbon sink accounting project can also be determined according to the third actual content. Specifically, for the high-interference risk area determined by the environmental characteristics of the region where the project is located, a dynamic biomass model and corresponding parameters are selected considering the influence of interference events on tree growth, and a model and corresponding parameters capable of simulating rapid tree growth, mortality and the influence of interference events are used; for the stable climate area determined by the environmental characteristics of the region where the project is located, a static or semi-static model suitable for stable environment and corresponding parameters are selected, and the model assumes less interference events and makes prediction based on stable climate conditions.

[0074] Further, according to the carbon pool scope and priority of the forestry carbon sink accounting project, the monitoring method and accuracy and frequency, the biomass model and accounting parameters, a baseline scenario scheme can be formulated, and the baseline scenario scheme after removing the uncertainty sources can be taken as the accounting scheme. Specifically, the carbon pool scope needs to be determined to account for the carbon pool (aboveground / underground / soil, etc.), and then the carbon pool required to construct the baseline scenario scheme is determined. The monitoring capability such as remote sensing accuracy, ground sample density and monitoring frequency determines the selection of data sources of the baseline scenario scheme such as historical remote sensing or field monitoring. The model accuracy level and regional adaptability of the biological model affect the reliability of growth prediction in the baseline scenario scheme. After the baseline scenario scheme is formulated, the uncertainty sources can include measurement uncertainty, model parameter uncertainty and baseline scenario uncertainty, and the baseline scenario scheme after removing the uncertainty sources can be taken as the accounting scheme.

[0075] In another possible implementation, step S30 can include step S30B:

[0076] In step S30B, the project characteristics are taken as input characteristics, and the output label corresponding to the input characteristics is inferred based on the pre-trained selection model, wherein the output label is a label value corresponding to the actual content of the key accounting factor of the forestry carbon sink accounting project.

[0077] In the present embodiment, another method for determining the actual content of the key accounting factor of the forestry carbon sink accounting project according to the project characteristics is provided. The key core factor is defined as the output label of the selection model, such as the scope and integrity of the carbon pool, the project boundary, the baseline scenario setting, the monitoring method and frequency, etc. The project characteristics are collected, including numerical and categorical types, such as the categorical type of project type (new afforestation / forest management / REDD+ / agroforestry, etc.), and the numerical type of project scale, etc. Then, data preprocessing is performed, such as processing missing values: filling numerical characteristics with mean / median, filling missing values of categorical characteristics with new categories; using One-Hot Encoding or Embedding method to encode categorical characteristics; and using Z-score standardization or Min-Max normalization method to standardize numerical characteristics. Further, the project characteristics after data preprocessing are taken as input characteristics, and the output label corresponding to the input characteristics is inferred based on the pre-trained selection model, i.e., the label value corresponding to the actual content of the key accounting factor of the forestry carbon sink accounting project.

[0078] In another possible implementation, step S30B can include:

[0079] The historical project information and historical accounting scheme of the historical forestry carbon sink accounting project are obtained, and the historical actual content of the key accounting factor of the forestry carbon sink accounting project is determined from the historical accounting scheme;

[0080] The training set is constructed by historical project information and historical actual content of the historical forestry carbon sink accounting project, and the selected model is trained based on the training set.

[0081] In selecting the machine learning model, a commonly used model for multi-label classification can be selected, multi-labels are converted into multiple binary classifications or label combinations, a multi-label algorithm such as ML-kNN (multi-label k-nearest neighbor), MLP (multi-label decision tree) is used for algorithm adaptation, a neural network is constructed, multiple sigmoid activation functions are used in the output layer, and each label is independently binary classified. In an embodiment, a deep neural network (DNN) or a tree model (such as a multi-label version of a random forest or XGBoost combined with a problem conversion method) that captures complex nonlinear relationships and feature interactions is used. For the input layer, the number of project features is used. Several fully connected layers (such as 256, 128, and 64 neurons) are set in the hidden layer, followed by an activation function (such as ReLU) and a Dropout layer to prevent overfitting. The output layer includes neurons corresponding to the number of key core factors, each neuron uses a sigmoid activation function and is independently binary classified. A loss function and an optimizer are defined, and since each label is an independent binary classification, a binary cross entropy (Binary Cross Entropy) loss function can be used, the loss of each label can be calculated and then averaged (or weighted). The Adam optimizer is used as the optimizer. During training, the model is trained on the training set, and the weights are updated for each batch. In the validation process, the validation set is used to monitor the model performance, and early stopping (Early Stopping) and hyperparameter adjustment are performed according to the validation loss or evaluation index, for example, the validation set is used to adjust hyperparameters such as the number of network layers and neurons, learning rate, Dropout ratio, and batch size. The trained model is finally deployed, the project features are input, and the core factor selection result is output.

[0082] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar content as the above first embodiment can refer to the above introduction, and the following will not be repeated. On this basis, please refer to Figure 3 , the method further comprises steps T10-T30:

[0083] Step T10, when performing forestry carbon sink accounting, determining the target forest area to be accounted and the forest growth stage of the target forest area;

[0084] The forest growth stage affects the amount of carbon sink, and the ability of different forest growth stages to absorb and release carbon dioxide is different. Therefore, the growth stage can be determined by field investigation, satellite image analysis and historical data. Specifically, the target forest area can be determined by using tree age, height, density and other indicators. In this embodiment, the method for determining the target forest area to be calculated and the forest growth stage of the target forest area is not limited. For example, age structure analysis can be performed: according to the age distribution of trees, the growth stage of the forest area is determined. Generally, the forest growth stage can be divided into young forest, middle-aged forest, near-mature forest, mature forest and over-mature forest. The growth stage of the forest area can also be evaluated by measuring the diameter at breast height, height, crown width and other indicators of the trees through field investigation. Remote sensing image analysis can also be used: high-resolution images obtained by satellites or drones can be used to analyze the coverage and canopy density of the forest to assist in determining the growth stage.

[0085] Step T20, obtaining multi-modal carbon sink related data of the target forest area through the stereoscopic monitoring network of the target forest area;

[0086] The data sources of the multi-modal carbon sink related data of the target forest area monitored by the stereoscopic monitoring network include satellite remote sensing, ground sensors, drones, etc., covering coverage area, tree height, biomass, soil condition, meteorological data. Further, the non-uniform format and time resolution difference of each modal carbon sink related data are processed. In an embodiment, the established stereoscopic monitoring network includes satellite remote sensing: using multi-spectral and high-resolution satellite images, the coverage area, tree height, biomass and other information of the forest area are obtained. Commonly used satellite data includes Landsat, Sentinel-2, etc. Ground sensors: sensors are arranged inside the forest area to monitor environmental parameters such as soil moisture, temperature, precipitation, and tree growth conditions. Unmanned aerial vehicle monitoring: using multi-spectral or LiDAR equipment carried by the unmanned aerial vehicle, high-precision terrain and tree structure data are obtained. Meteorological data: collect meteorological data of the target forest area, including temperature, precipitation, wind speed, etc., for evaluating the influence of climate on carbon sink. Further, the multi-modal carbon sink related data is cleaned to remove noise and outliers; and data fusion is performed to fuse data from different sources to ensure the unity of time and space. For example, satellite images are combined with ground sensor data to improve data accuracy and resolution. Finally, the processed data is stored in a database for subsequent analysis and modeling.

[0087] Step T30, estimating the carbon sink amount of the target forest area based on the forest growth stage of the target forest area and the multi-modal carbon sink related data of the target forest area.

[0088] In selecting a carbon sink model, one can choose a biomass model that estimates carbon sink based on the relationship between tree biomass and carbon storage, including forest biomass models (such as the CASA model), tree growth models (such as the Cohort model), etc. One can also choose a carbon cycle model that considers the input and output of carbon in forest ecosystems, including processes such as photosynthesis, respiration, litter decomposition, etc., including the Carnegie-Ames-Stanford Approach (CASA) model or the CENTURY model, etc. Input growth stage data and multi-modal data, i.e., growth stages determined according to step T10, adjust model parameters. For example, trees in different growth stages have different responses to photosynthesis and respiration. Input multi-modal data obtained in step T20 into the model, including biomass, environmental parameters, meteorological data, etc. Finally, run the carbon sink model to simulate the carbon sink at different time points. Dynamic estimation means that the model can adjust the estimated results of carbon sink according to real-time or regularly updated data. And through field measurement or other independent data sources, verify the accuracy of the model's estimated results. For example, compare the model's estimated carbon storage with the actually measured tree biomass. Further, the results can be visualized: visualize the estimated results of carbon sink, such as generating maps or charts, to facilitate management and decision-making. And, according to the output results of the model, dynamically adjust the management strategies of the forest area, such as optimizing afforestation, tending or harvesting plans, to improve carbon sink capacity.

[0089] In an embodiment, step T30 includes:

[0090] Step T301, obtain historical forest growth stages and historical multi-modal carbon sink related data of the calculated forest area, and construct a training set of historical forest growth stages;

[0091] Collect historical forest growth stage data. Obtain historical data from the calculated forest area, including forest growth stages in different years (such as young forest, middle-aged forest, mature forest, etc.). Record the growth stages of the forest area at different time points, usually based on tree age, biomass, coverage, etc. Collect historical multi-modal carbon sink related data. Including satellite remote sensing data (such as Landsat, Sentinel-2), ground sensor data (such as soil moisture, temperature, precipitation), unmanned aerial vehicle data (such as LiDAR, multispectral image), etc. Record the carbon sink related indicators of the historical forest area, such as tree biomass, carbon storage, net primary productivity (NPP), environmental factors (such as temperature, precipitation, radiation, etc.). Data preprocessing. Including 1, data cleaning: remove outliers, fill in missing values. 2, data standardization: standardize different sources and formats of data to ensure data consistency. 3, feature extraction: extract carbon sink related features, such as tree height, crown width, biomass density, environmental factors, etc.

[0092] Step T302, based on the training set of historical forest growth stages, the prediction model for dynamically estimating the carbon sink capacity of the forest area according to the multi-modal carbon sink related data of the historical forest growth stage is trained;

[0093] Define the structure of the training set. Input features: historical multi-modal carbon sink related data (such as biomass, environmental factors, tree height, etc.). Output label: historical forest growth stage (such as young forest, middle-aged forest, mature forest, etc.). Data annotation: according to the historical data that has been calculated, the corresponding forest growth stage is annotated for each time point of the forest area. Data division: divide the historical data into training set and validation set, usually with a ratio of 70% training and 30% validation. Data storage: store the processed data in the database or file for subsequent model training and validation.

[0094] Select appropriate models according to data size and complexity. For example, LSTM or RNN is more suitable for time series data, and random forest or gradient boosting model is more effective for nonlinear relationship. Among them, the regression models that can be selected include linear regression, random forest regression, gradient boosting regression (such as XGBoost, LightGBM). Machine learning models that can be selected include support vector machine (SVM), artificial neural network (ANN). Deep learning models that can be selected include recurrent neural network (RNN), long short-term memory network (LSTM), convolutional neural network (CNN). Train the model using the training set data, input features are multi-modal carbon sink related data, and output is the target forest growth stage. Optimize model parameters through cross-validation (such as grid search, random search) to improve model performance. Evaluate model performance using validation set data, calculate accuracy, mean square error (MSE), coefficient of determination (R²), etc. If the model performance is not ideal, adjust the model structure or reselect the algorithm.

[0095] Step T303, dynamically estimate the carbon sink capacity of the target forest area through the target prediction model and the multi-modal carbon sink related data of the target forest area.

[0096] First, obtain real-time or multi-modal data of the target forest area. Data source: obtain real-time data of the target forest area through stereo monitoring network (such as satellite, unmanned aerial vehicle, ground sensor). Data content: including tree biomass, environmental factors, terrain information, etc. Data preprocessing: preprocess the real-time data of the target forest area, including cleaning, standardization, feature extraction, so that it is consistent with the training set data format. Then model inference, input the preprocessed target forest area data into the trained prediction model, dynamically estimate the carbon sink capacity of the target forest area. Finally, result output and visualization. Output the estimated results in the form of charts or maps for analysis and display. In addition, according to the update frequency of real-time data (such as daily, weekly, monthly), dynamically adjust the estimated results of carbon sink capacity.

[0097] It should be noted that the above examples are only for understanding the present application and do not constitute a limitation on the machine learning-based forestry carbon sink dynamic accounting method of the present application. Further simple transformations based on this technical concept are within the scope of protection of the present application.

[0098] The present application also provides a machine learning-based forestry carbon sink dynamic accounting device, which will be described in detail with reference to Figure 4 The machine learning-based forestry carbon sink dynamic accounting device comprises:

[0099] The preparation module 10 is configured to determine a forestry carbon sink accounting project to be implemented in a current accounting period, and collect project information of the forestry carbon sink accounting project from project declaration materials of the forestry carbon sink accounting project.

[0100] The identification module 20 is configured to identify project features of the forestry carbon sink accounting project from the project information based on machine learning; wherein the project features include a project type, a project scale, a regional terrain where the project is located, and environmental features of the region where the project is located.

[0101] The generation module 30 is configured to determine actual contents of key accounting factors of the forestry carbon sink accounting project according to the project features, and perform forestry carbon sink accounting according to an accounting scheme obtained by using the actual contents of the key accounting factors.

[0102] In an embodiment, the generation module 30 is further configured to:

[0103] determine first actual contents of a carbon pool, a baseline, and a disturbance type in the key accounting factors of the forestry carbon sink accounting project according to the project type of the forestry carbon sink accounting project;

[0104] determine second actual contents of a monitoring target in the key accounting factors of the forestry carbon sink accounting project according to the project scale and the regional terrain where the project is located;

[0105] determine third actual contents of an accounting model and parameters in the key accounting factors of the forestry carbon sink accounting project according to the environmental features of the region where the project is located.

[0106] In an embodiment, the generation module 30 is further configured to:

[0107] determine a carbon pool range and priority of the forestry carbon sink accounting project according to the first actual contents;

[0108] determine a monitoring method, accuracy, and frequency of the forestry carbon sink accounting project according to the second actual contents;

[0109] determine a biomass model and accounting parameters of the forestry carbon sink accounting project according to the third actual contents.

[0110] According to the carbon pool scope and priority of the forestry carbon sink accounting project, the monitoring method and precision, and the frequency, the biomass model and accounting parameters, a baseline scenario scheme is formulated, and the baseline scenario scheme after removing the uncertainty sources is taken as the accounting scheme.

[0111] In an embodiment, the generating module 30 is further configured to:

[0112] The project characteristics are taken as input features, and the output label corresponding to the input features is inferred based on the pre-trained selection model, wherein the output label is a label value corresponding to the actual content of the key accounting factor of the forestry carbon sink accounting project.

[0113] In an embodiment, the generating module 30 is further configured to:

[0114] The historical project information and the historical accounting scheme of the historical forestry carbon sink accounting project are obtained, and the historical actual content of the key accounting factor of the forestry carbon sink accounting project is determined from the historical accounting scheme.

[0115] The training set is constructed based on the historical project information and the historical actual content of the historical forestry carbon sink accounting project, and the selection model is trained based on the training set.

[0116] In an embodiment, the device further comprises an application module configured to:

[0117] When performing forestry carbon sink accounting, the target forest area to be accounted and the forest growth stage in which the target forest area is located are determined.

[0118] The multi-modal carbon sink related data of the target forest area are obtained through the stereoscopic monitoring network of the target forest area.

[0119] The carbon sink amount of the target forest area is estimated based on the forest growth stage in which the target forest area is located and the multi-modal carbon sink related data of the target forest area.

[0120] The forestry carbon sink dynamic accounting device based on machine learning provided in the present application adopts the forestry carbon sink dynamic accounting method based on machine learning in the above embodiments, and can solve the technical problem of inaccurate accounting when performing forestry carbon sink dynamic accounting. Compared with the prior art, the beneficial effects of the forestry carbon sink dynamic accounting device based on machine learning provided in the present application are the same as those of the forestry carbon sink dynamic accounting method based on machine learning provided in the above embodiments, and other technical features in the forestry carbon sink dynamic accounting device based on machine learning are the same as those disclosed in the above embodiments, which will not be repeated here.

[0121] The present application provides a dynamic accounting device for forestry carbon sinks based on machine learning. The dynamic accounting device for forestry carbon sinks based on machine learning includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the dynamic accounting method for forestry carbon sinks based on machine learning in the above-mentioned embodiment one.

[0122] Reference below Figure 5 , which shows a schematic diagram of the structure of a machine learning-based dynamic accounting device for forestry carbon sinks suitable for implementing the embodiments of the present application. The machine learning-based dynamic accounting device for forestry carbon sinks in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The forestry carbon sink dynamic accounting device based on machine learning shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of this application.

[0123] like Figure 5As shown, the machine learning based forestry carbon sink dynamic accounting device can include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to programs stored in a read-only memory 1002 or loaded from a storage device 1003 into a random access memory 1004. In the random access memory 1004, various programs and data required for operation of the machine learning based forestry carbon sink dynamic accounting device are also stored. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are connected to each other through a bus 1005. An input / output interface 1006 is also connected to the bus. Generally, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; the storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the machine learning based forestry carbon sink dynamic accounting device to communicate with other devices wirelessly or by wire to exchange data. Although the machine learning based forestry carbon sink dynamic accounting device with various systems is shown in the figure, it should be understood that all the systems shown are not required to be implemented or possessed. More or fewer systems can be alternatively implemented or possessed.

[0124] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program codes for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through the communication device, or installed from the storage device 1003, or installed from the read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the methods of the embodiments disclosed in the present application are performed.

[0125] The machine learning based forestry carbon sink dynamic accounting device provided by the present application adopts the machine learning based forestry carbon sink dynamic accounting method in the above-mentioned embodiments, and can solve the technical problem of inaccurate accounting in forestry carbon sink dynamic accounting. Compared with the prior art, the machine learning based forestry carbon sink dynamic accounting device provided by the present application has the same beneficial effects as the machine learning based forestry carbon sink dynamic accounting method provided by the above-mentioned embodiments, and other technical features in the machine learning based forestry carbon sink dynamic accounting device are the same as the features disclosed in the previous embodiment method, which will not be repeated here.

[0126] It should be understood that various parts of the present application can be realized by hardware, software, firmware, or a combination thereof. In the above-described embodiments, specific features, structures, materials or characteristics can be combined in any appropriate manner in any one or more embodiments or examples.

[0127] The above description is merely that of a specific implementation of the present application, and the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, and all such changes or replacements should be encompassed within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0128] The present application provides a computer readable storage medium having stored thereon computer readable program instructions (i.e., computer programs) for performing the machine learning based forestry carbon sink dynamic accounting method in the above-described embodiments.

[0129] The computer readable storage medium provided by the present application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system or device, or any combination thereof. More specific examples of the computer readable storage medium can include, but are not limited to, an electric connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM), a read only memory (ROM), an erasable programmable read only memory (EPROM or flash memory), an optical fiber, a portable compact disk read only memory (CD-ROM), an optical storage device, a magnetic storage device, or any appropriate combination thereof. In the present embodiment, the computer readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer readable storage medium can be transmitted by any appropriate medium, including but not limited to an electric wire, an optical cable, an RF (Radio Frequency), etc., or any appropriate combination thereof.

[0130] The above-described computer readable storage medium can be included in the machine learning based forestry carbon sink dynamic accounting device; or can exist separately and not be assembled into the machine learning based forestry carbon sink dynamic accounting device.

[0131] The computer readable storage medium described above carries one or more programs, when the one or more programs are executed by the forest carbon sink dynamic accounting device based on machine learning, the forest carbon sink dynamic accounting device based on machine learning is caused to: determine a forest carbon sink accounting project to be implemented in a current accounting period, collect project information of the forest carbon sink accounting project from project declaration materials of the forest carbon sink accounting project; based on machine learning, identify project characteristics of the forest carbon sink accounting project from the project information; wherein the project characteristics include a project type, a project scale, and a regional terrain and environmental characteristics of the forest carbon sink accounting project; according to the project characteristics, determine actual content of a key accounting factor of the forest carbon sink accounting project, and perform forest carbon sink accounting according to an accounting scheme obtained by using the actual content of the key accounting factor.

[0132] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0133] The flow diagrams and the block diagrams in the drawings are illustrations of architectures, functionalities, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of present application. In this regard, each block in the flow diagrams or block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may

[0134] The modules described in the embodiments of the present application can be implemented in the form of software or in the form of hardware. In some cases, the name of the module does not constitute a limitation on the module itself.

[0135] The readable storage medium provided by the present application is a computer readable storage medium, which stores computer readable program instructions (i.e., a computer program) for executing the above-mentioned machine learning-based forestry carbon sink dynamic accounting method, and can solve the technical problem of inaccurate accounting in forestry carbon sink dynamic accounting. Compared with the prior art, the computer readable storage medium provided by the present application has the same beneficial effects as the machine learning-based forestry carbon sink dynamic accounting method provided in the above-mentioned embodiments, and will not be described here.

[0136] The present application also provides a computer program product comprising a computer program, which, when executed by a processor, implements the steps of the machine learning-based forestry carbon sink dynamic accounting method as described above.

[0137] The computer program product provided by the present application can solve the technical problem of inaccurate accounting in forestry carbon sink dynamic accounting. Compared with the prior art, the computer program product provided by the present application has the same beneficial effects as the machine learning-based forestry carbon sink dynamic accounting method provided in the above-mentioned embodiments, and will not be described here.

[0138] The above only describes some embodiments of the present application, and does not limit the patent scope of the present application. Any equivalent structural transformation, direct / indirect application in other related technical fields based on the technical concept of the present application, and the contents of the present application specification and drawings are included in the patent protection scope of the present application.

Claims

1. A method for dynamic accounting of forestry carbon sinks based on machine learning, characterized in that: The forestry carbon sink dynamic accounting method based on machine learning includes: Determine the forestry carbon sink accounting projects to be implemented in the current accounting cycle, and collect project information of the forestry carbon sink accounting projects from the project application materials of the forestry carbon sink accounting projects; Identifying project characteristics of the forestry carbon sink accounting project from the project information based on machine learning; wherein the project characteristics include the project type, project scale, topography of the project area, and environmental characteristics of the project area; Determine the first actual content of the carbon pool, baseline and disturbance type among the key accounting factors of the forestry carbon sink accounting project based on the project type of the forestry carbon sink accounting project; Determine the second actual content of the monitoring target in the key accounting factors of the forestry carbon sink accounting project based on the project scale and the topography of the project area; Determine the third actual content of the accounting model and parameters in the key accounting factors of the forestry carbon sink accounting project based on the environmental characteristics of the project area where the forestry carbon sink accounting project is located; Determine the carbon pool scope and priority of the forestry carbon sink accounting project based on the first actual content; Determine the monitoring method, accuracy, and frequency of the forestry carbon sink accounting project based on the second practical content; Determine the biomass model and accounting parameters of the forestry carbon sink accounting project based on the third actual content; Based on the carbon pool scope and priority, monitoring methods, accuracy and frequency, biomass model and accounting parameters of the forestry carbon sink accounting project, a baseline scenario plan is developed and used as the accounting plan after removing sources of uncertainty; sources of uncertainty include measurement uncertainty, model parameter uncertainty and baseline scenario uncertainty; Forestry carbon sink accounting is carried out according to the accounting scheme obtained by adopting the actual content of the key accounting factors.

2. The method for dynamic forestry carbon sink accounting based on machine learning according to claim 1, characterized in that: The step of determining the actual content of the key accounting factors of the forestry carbon sink accounting project based on the project characteristics further includes: The project features are used as input features, and the output labels corresponding to the input features are inferred based on the pre-trained selection model, wherein the output labels are label values ​​corresponding to the actual content of the key accounting factors of the forestry carbon sink accounting project.

3. The method for dynamic forestry carbon sink accounting based on machine learning according to claim 2, characterized in that: The step of using the project features as input features and obtaining output labels corresponding to the input features by inference based on the pre-trained selection model includes: Obtain historical project information and historical accounting plans for historical forestry carbon sink accounting projects, and determine the historical actual content of key accounting factors of the forestry carbon sink accounting projects from the historical accounting plans; A training set is constructed through the historical project information and historical actual content of the historical forestry carbon sink accounting project, and a selection model is obtained by training based on the training set.

4. The method for dynamic forestry carbon sink accounting based on machine learning according to claim 1, characterized in that: The method further comprises: When conducting forestry carbon sink accounting, determine the target forest area to be accounted for and the forest growth stage of the target forest area; Acquiring multimodal carbon sink-related data of the target forest area through a three-dimensional monitoring network of the target forest area; The carbon sequestration amount of the target forest area is estimated based on the forest growth stage of the target forest area and the multimodal carbon sequestration-related data of the target forest area.

5. A dynamic accounting device for forestry carbon sinks based on machine learning, characterized in that: The forestry carbon sink dynamic accounting device based on machine learning includes: A preparation module, configured to determine a forestry carbon sink accounting project to be implemented in the current accounting cycle, and to collect project information of the forestry carbon sink accounting project from project application materials of the forestry carbon sink accounting project; an identification module, configured to identify project characteristics of the forestry carbon sink accounting project from the project information based on machine learning; wherein the project characteristics include the project type, project scale, and topography and environmental characteristics of the project area; A generation module is used to determine the first actual content of the carbon pool, baseline and interference type in the key accounting factors of the forestry carbon sink accounting project according to the project type of the forestry carbon sink accounting project; determine the second actual content of the monitoring target in the key accounting factors of the forestry carbon sink accounting project according to the project scale of the forestry carbon sink accounting project and the terrain of the project area; determine the third actual content of the accounting model and parameters in the key accounting factors of the forestry carbon sink accounting project according to the environmental characteristics of the project area of ​​the forestry carbon sink accounting project; determine the carbon pool scope and priority of the forestry carbon sink accounting project according to the first actual content; and determine the third actual content of the accounting model and parameters in the key accounting factors of the forestry carbon sink accounting project according to the environmental characteristics of the project area of ​​the forestry carbon sink accounting project. The second actual content determines the monitoring method, accuracy and frequency of the forestry carbon sink accounting project; based on the third actual content, determines the biomass model and accounting parameters of the forestry carbon sink accounting project; based on the carbon pool scope and priority, monitoring method, accuracy and frequency, biomass model and accounting parameters of the forestry carbon sink accounting project, formulates a baseline scenario plan and uses the baseline scenario plan after removing the sources of uncertainty as the accounting plan; wherein, the sources of uncertainty include measurement uncertainty, model parameter uncertainty and baseline scenario uncertainty; forestry carbon sink accounting is carried out according to the accounting plan obtained by adopting the actual content of the key accounting factors.

6. A forestry carbon sink dynamic accounting device based on machine learning, characterized in that: The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the dynamic accounting method for forestry carbon sinks based on machine learning as described in any one of claims 1 to 4.

7. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the dynamic accounting method of forestry carbon sinks based on machine learning as described in any one of claims 1 to 4 are implemented.

8. A computer program product, characterized in that The computer program product includes a computer program, which, when executed by a processor, implements the steps of the dynamic accounting method for forestry carbon sinks based on machine learning as described in any one of claims 1 to 4.

Images