Collaborative inversion method, system and equipment for carbon cycle component of global terrestrial ecosystem
By preprocessing ground station observation data and image data and training a random forest model, the co-retrieval of carbon cycle components in global terrestrial ecosystems was achieved. This solved the problem of insufficient mass conservation constraints on carbon cycle components in existing technologies, and improved the understanding of dynamic changes in the carbon cycle and the ability to perform carbon accounting.
Patent Information
- Application Number
- CN202511245748.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-02
- Publication Date
- 2026-01-23
AI Technical Summary
Current technologies for estimating carbon cycles in global terrestrial ecosystems lack a comprehensive consideration of the mass conservation constraints among the components of the carbon cycle. This makes it difficult to achieve a physically consistent representation of the carbon cycle process at the system level, affecting the accurate assessment of the carbon sink capacity of ecosystems and the scientific understanding of carbon-climate feedback mechanisms.
By acquiring global-scale ground station observation data and image data, performing preprocessing and carbon balance correction, a random forest model is established. The random forest method is used for training to construct a global terrestrial ecosystem carbon cycle component collaborative inversion model, realizing refined component separation and matching of carbon cycle components and satisfying mass conservation constraints.
It provides remote sensing products for the carbon cycle components of global terrestrial ecosystems that simultaneously meet the constraints of mass conservation and have high precision, enhancing our understanding of the dynamic changes in the carbon cycle of terrestrial ecosystems and providing a solid foundation for global carbon accounting and carbon trading.
Smart Images
Figure CN121390503A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of remote sensing estimation of global terrestrial ecosystem carbon cycle components, in particular to a global terrestrial ecosystem carbon cycle component collaborative inversion method, system and device. BACKGROUND
[0002] Land carbon cycle plays a key role in regulating atmospheric carbon dioxide concentration, maintaining climate system stability and supporting ecosystem functions. Accurate acquisition of global terrestrial ecosystem carbon cycle key components is of great significance for revealing carbon source and sink patterns, assessing climate change impacts and developing carbon neutralization policies.
[0003] In recent decades, the rapid accumulation of satellite remote sensing observation data has greatly promoted the development and application of remote sensing-based terrestrial ecosystem carbon flux estimation models, providing key support for regional to global scale carbon budget assessment. However, existing methods generally focus on the estimation of a single carbon flux, lack comprehensive consideration of mass conservation constraints between carbon cycle components, and are difficult to achieve physical consistency expression of carbon cycle processes at the system level, thereby affecting accurate assessment of ecosystem carbon sink capacity and scientific understanding of carbon-climate feedback mechanisms.
[0004] Therefore, it is urgent to develop a global terrestrial ecosystem carbon cycle component collaborative inversion method with carbon mass conservation constraints, which has important theoretical significance and application value for improving the quantification ability of key carbon cycle processes, building a high-quality carbon source and sink assessment system, and supporting global climate change scientific research and policy making. SUMMARY
[0005] The purpose of the present application is to provide a global terrestrial ecosystem carbon cycle component collaborative inversion method, system and device to solve the technical problems pointed out in the background art.
[0006] The present application is implemented by the following technical solution: a global terrestrial ecosystem carbon cycle component collaborative inversion method, comprising the following steps: Obtain global-scale ground station observation data and image data, the ground station observation data containing original carbon flux data, and the image data including original remote sensing image data representing ground vegetation and terrain information and original meteorological reanalysis data representing meteorological and soil conditions; Preprocess the ground station observation data and image data respectively, wherein the preprocessing of the ground station observation data includes carbon balance correction processing to obtain carbon flux data satisfying the mass conservation constraint; Establish a database based on the preprocessed observation data and image data; carrying out carbon component separation on the carbon flux data satisfying the mass conservation constraint at a site scale to obtain a carbon cycle component of a refined component; According to geographical position information of the site, image data corresponding to a pixel value of each site is extracted at a pixel scale, and is matched with the carbon cycle component to construct a data set; Taking the carbon cycle component as a model dependent variable and the image data as a model independent variable, a random forest model is established by using a random forest method, and the random forest model is trained by using the data set to obtain a global terrestrial ecosystem carbon cycle component collaborative inversion model; The image data is obtained as an input of the global terrestrial ecosystem carbon cycle component collaborative inversion model to obtain a carbon cycle component inversion result.
[0007] According to a preferred embodiment, the original carbon flux data includes total primary productivity, ecosystem respiration, and ecosystem net exchange; The remote sensing image data includes a normalized vegetation index, a leaf area index, a sunlight-induced chlorophyll fluorescence, and elevation data; The meteorological reanalysis data includes downwelling shortwave radiation, near-surface atmospheric temperature, dew point temperature, soil temperature, soil moisture, and near-surface wind speed.
[0008] According to a preferred embodiment, the step of carbon balance correction is as follows: It is judged whether the original carbon flux data satisfies the mass conservation constraint, and the expression of the mass conservation constraint is as follows:
[0009] In the above formula, represents total primary productivity, represents ecosystem respiration, represents ecosystem net exchange; The carbon non-conservation amount and the carbon non-conservation ratio are calculated for the original carbon flux data not satisfying the mass conservation constraint, and the expressions are as follows:
[0010]
[0011] In the above formula, represents the carbon non-conservation amount, represents the carbon non-conservation ratio; The original carbon flux data that and satisfy the set threshold value at the same time is determined as the original carbon flux data to be processed; The carbon imbalance is proportionally distributed to gross primary productivity and ecosystem respiration respectively based on the proportion of gross primary productivity and ecosystem respiration in the original carbon flux data to be processed, and the carbon balance corrected gross primary productivity and ecosystem respiration are obtained, and the expression is as follows:
[0012]
[0013] In the above formula, represents the gross primary productivity in the original carbon flux data to be processed, represents the ecosystem respiration in the original carbon flux data to be processed, represents the carbon balance corrected gross primary productivity, represents the carbon balance corrected ecosystem respiration.
[0014] According to a preferred embodiment, the preprocessing of the ground station observation data further includes quality control and scale conversion; wherein the quality control is to filter out measurement data and high-precision interpolation data from the original carbon flux data according to a quality control symbol to obtain high-quality data, and the high-quality data is subjected to carbon balance correction; The scale conversion is to aggregate the original carbon flux data of the hour scale to the required time scale.
[0015] According to a preferred embodiment, the preprocessing of the original image data includes quality control and spatio-temporal resolution unification processing; wherein the quality control is to retain high-quality pixel values according to a quality control symbol; The spatio-temporal resolution unification processing is to unify the original image data to the same spatio-temporal resolution by using resampling or interpolation method to obtain image data with unified spatio-temporal resolution.
[0016] According to a preferred embodiment, the carbon cycle component of the refined component includes gross primary productivity, net primary productivity, ecosystem total respiration, ecosystem autotrophic respiration, ecosystem heterotrophic respiration and ecosystem net carbon exchange.
[0017] According to a preferred embodiment, the carbon flux data satisfying the mass conservation constraint is subjected to carbon component separation at the site scale, and the specific steps are as follows: The net primary productivity and the ecosystem autotrophic respiration are separated from the carbon balance corrected gross primary productivity according to the proportion of the net primary productivity and the gross primary productivity of different vegetation types at the site, and the expression is as follows:
[0018]
[0019] NPP represents net primary productivity, NPP represents net primary productivity, NPP represents net primary productivity and the ratio of total primary productivity of different vegetation types, NPP represents ecosystem autotrophic respiration; The ecosystem heterotrophic respiration is separated from the ecosystem autotrophic respiration, and the expression is as follows:
[0020] NPP represents net primary productivity and the ratio of total primary productivity of different vegetation types, NPP represents net primary productivity and the ratio of total primary productivity of different vegetation types,
[0021] According to a preferred embodiment, the random forest model is trained using the data set, specifically including: model parameter tuning is performed by random search method and grid search method to determine the optimal parameter configuration The application also provides a global terrestrial ecosystem carbon cycle component collaborative inversion system, which is applied to the global terrestrial ecosystem carbon cycle component collaborative inversion method described above, and the system comprises: A data acquisition unit is configured to acquire global-scale ground station observation data and image data, wherein the ground station observation data includes original carbon flux data, and the image data includes original remote sensing image data representing ground vegetation and terrain information and original meteorological reanalysis data representing meteorological and soil conditions; A data preprocessing unit is configured to preprocess the ground station observation data and image data respectively, wherein the preprocessing of the ground station observation data includes carbon balance correction processing to obtain carbon flux data satisfying the mass conservation constraint; A database construction unit is configured to construct a database based on the preprocessed observation data and image data; A refined carbon component calculation unit is configured to separate carbon components at a site scale for the carbon flux data satisfying the mass conservation constraint to obtain carbon cycle components of refined components; A data matching unit is configured to extract pixel values corresponding to each site from the image data at a pixel scale according to geographical location information of the site, and match the carbon cycle components to construct a data set, wherein the data set includes a training set and a test set; A data set training unit is configured to use the carbon cycle components as model dependent variables, the image data as model independent variables, establish a random forest model by using a random forest method, train the random forest model by using the training set, perform model parameter tuning by using a random search method and a grid search method, obtain a plurality of parameter-optimized random forest models, evaluate each parameter-optimized random forest model by using the test set, determine an optimal parameter configuration from the plurality of parameter-optimized random forest models based on an evaluation result, and obtain a global terrestrial ecosystem carbon cycle component collaborative inversion model. A model production unit is configured to obtain image data as input of a global terrestrial ecosystem carbon cycle component collaborative inversion model, and obtain carbon cycle component inversion results.
[0022] The application further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the global terrestrial ecosystem carbon cycle component collaborative inversion method when executing the computer program.
[0023] The global terrestrial ecosystem carbon cycle component collaborative inversion method, system and device provided by the application have at least the following advantages and beneficial effects: the application can provide global terrestrial ecosystem carbon cycle component remote sensing products that meet the mass conservation constraint and high precision at the same time, which helps to improve the understanding of the dynamic changes of the terrestrial ecosystem carbon cycle and provides a solid foundation for global carbon accounting and carbon trading. BRIEF DESCRIPTION OF DRAWINGS
[0024] Figure 1 A flowchart of the global terrestrial ecosystem carbon cycle component collaborative inversion method provided for the embodiment 1 of the application is shown in the figure. Figure 2 A structure block diagram of the global terrestrial ecosystem carbon cycle component collaborative inversion system provided for the embodiment 2 of the application is shown in the figure. DETAILED DESCRIPTION
[0025] To make the objectives, technical solutions and advantages of the embodiments of the application clearer, the technical solutions in the embodiments of the application will be described below in connection with the drawings of the embodiments of the application. Obviously, the described embodiments are some embodiments of the application but not all the embodiments of the application. The components of the embodiments of the application described and shown in the drawings can be arranged and designed in various different configurations.
[0026] Embodiment 1 The embodiment of the application provides a global terrestrial ecosystem carbon cycle component collaborative inversion method, Figure 1 A flowchart of the global terrestrial ecosystem carbon cycle component collaborative inversion method is shown in the figure. Figure 1 The global terrestrial ecosystem carbon cycle component collaborative inversion method includes the following steps: Step S01, obtaining global-scale ground station observation data and image data; The ground station observation data includes original carbon flux data, and the main sources are global / regional flux networks, such as the global flux observation total network FLUXNET and the AmeriFlux network in the Americas region; the variables to be extracted include total primary productivity, ecosystem respiration and ecosystem net exchange, and the time resolution is half an hour or one hour.
[0027] The image data includes original remote sensing image data representing ground vegetation and terrain information and original meteorological reanalysis data representing meteorological and soil conditions; wherein the original remote sensing image data mainly comes from MODIS reflectivity products, MODIS leaf area index products, sunlight-induced chlorophyll fluorescence GOSIF products and normalized vegetation index, leaf area index, sunlight-induced chlorophyll fluorescence and elevation data of GMTED2010; the meteorological reanalysis data mainly comes from downlink shortwave radiation, near-surface atmospheric temperature, dew point temperature, soil temperature, soil humidity and near-surface wind speed data of ERA5-land product. It should be noted that the number and type of sources of ground station observation data and image data are not limited in the embodiments of the present application.
[0028] Step S02, the ground station observation data and image data are respectively preprocessed; In this embodiment, the preprocessing of the ground station observation data includes quality control, carbon balance correction processing and scale conversion, to obtain carbon flux data satisfying the quality conservation constraint; Specifically, in this embodiment, the quality control is to filter out measured data and high-precision interpolated data from the original carbon flux data according to the quality control symbol to obtain high-quality data, and the high-quality data is subjected to carbon balance correction; The specific steps of carbon balance correction are as follows: First, it is judged whether the original carbon flux data satisfies the quality conservation constraint, and the expression of the quality conservation constraint is as follows:
[0029] In the above formula, represents the total primary productivity, represents the ecosystem respiration, represents the ecosystem net exchange; Then, the carbon non-conservation amount and the carbon non-conservation ratio of the original carbon flux data not satisfying the quality conservation constraint are calculated respectively, and the expressions are as follows:
[0030]
[0031] In the above formula, represents the carbon non-conservation amount, represents the carbon non-conservation ratio; Further, the screening obtains and The original carbon flux data that meets the set threshold value at the same time is determined as the to-be-processed original carbon flux data; in some embodiments, the to-be-processed original carbon flux data is respectively and A threshold standard is formulated, for example, the screening threshold standard is strictly set as and 5%, only the original carbon flux data that meets the threshold standard at the same time is reserved, and the expression is as follows:
[0032]
[0033] It should be noted that, because the original carbon flux data that exceeds the corresponding screening threshold standard is too high in non-conservation degree and is unreliable, only the original carbon flux data that meets the threshold standard at the same time is reserved, the non-conservation degree of such original carbon flux data is small enough, the influence can be ignored, and the carbon balance correction can be carried out; Finally, based on the ratio of the total primary productivity to the ecosystem respiration in the to-be-processed original carbon flux data, the carbon non-conservation amount is proportionally distributed to the total primary productivity and the ecosystem respiration respectively, and the total primary productivity and the ecosystem respiration after the carbon balance correction are obtained, and the expression is as follows:
[0034]
[0035] In the above formula, represents the total primary productivity in the to-be-processed original carbon flux data, represents the ecosystem respiration in the to-be-processed original carbon flux data, represents the total primary productivity after the carbon balance correction, represents the ecosystem respiration after the carbon balance correction.
[0036] Specifically in this embodiment, the scale conversion is to aggregate the original carbon flux data of the half-hour / hour scale to the required time scale, for example, to aggregate the original carbon flux data of the half-hour / hour scale to the daily scale.
[0037] In some embodiments of the present embodiment, the preprocessing of the original image data includes quality control and spatio-temporal resolution unification processing; wherein the quality control is to reserve high-quality pixel values according to the quality control symbol; the spatio-temporal resolution unification processing is to unify the original image data to the same spatio-temporal resolution by using resampling or interpolation method to obtain the spatio-temporally unified image data, for example, to unify it to the spatio-temporal resolution of 500 meters per day, which will not be described in detail here.
[0038] Step S03, database establishment; In the present embodiment, a database is established using the observation data and the image data pre-processed in step S02.
[0039] Step S04, refined carbon component calculation; The carbon flux data satisfying the mass conservation constraint obtained in step S02 is combined with site-scale vegetation type information and a lookup table of net primary productivity and total primary productivity ratio to carry out site-scale carbon component separation, and refined component carbon cycle components are obtained. In the present embodiment, the refined component carbon cycle components include total primary productivity, net primary productivity, ecosystem total respiration, ecosystem autotrophic respiration, ecosystem heterotrophic respiration, and ecosystem net carbon exchange.
[0040] The site-scale carbon component separation is carried out on the carbon flux data satisfying the mass conservation constraint, and the specific steps are as follows: First, the net primary productivity and the ecosystem autotrophic respiration are separated from the total primary productivity corrected by carbon balance according to the ratio of net primary productivity to total primary productivity of different vegetation types, and the expression is as follows:
[0041]
[0042] In the above formula, represents the net primary productivity, represents the ratio of net primary productivity to total primary productivity of different vegetation types, represents the ecosystem autotrophic respiration; in some embodiments, the ratio of net primary productivity to total primary productivity of different vegetation types is 0.43; The ecosystem heterotrophic respiration is separated from the ecosystem autotrophic respiration, and the expression is as follows:
[0043] In the above formula, represents the ecosystem heterotrophic respiration.
[0044] Step S05, data matching; In the present embodiment, according to the geographical position information of the site, such as latitude and longitude data, the image data corresponding to each site is extracted on the pixel scale, and is matched with the carbon cycle components to construct a data set; thus, the data set includes three types of data, which are site-scale carbon flux data corrected by carbon balance, remote sensing data for representing ground vegetation and terrain information, and meteorological reanalysis data for representing meteorological and soil conditions.
[0045] Further, the three types of data are matched in space-time, and are split into training data and test data by random sampling method, and the ratio of training and test data can be set as 8:2, so as to form a training set and a test set.
[0046] Step S06, training of the data set; In the embodiment, the random forest model is established by taking the carbon cycle component as the model dependent variable and the image data as the model independent variable, and the random forest model is trained by using the training set. The model is tuned by random search method and grid search method, a plurality of parameter-optimized random forest models are obtained, each parameter-optimized random forest model is evaluated by using the test set, and the optimal parameter configuration is determined in the plurality of parameter-optimized random forest models based on the evaluation result, so as to obtain the global terrestrial ecosystem carbon cycle component collaborative inversion model.
[0047] In some embodiments, the global terrestrial ecosystem carbon cycle component collaborative inversion model can be expressed as:
[0048] In the above formula, represents the global terrestrial ecosystem carbon cycle component collaborative inversion model, represents a mapping function, represents a normalized vegetation index, represents a leaf area index, represents sunlight-induced chlorophyll fluorescence, represents air humidity, represents downward shortwave radiation, represents soil temperature, in addition to which, the mapping function further includes elevation data, near-surface air temperature, dew point temperature, soil moisture, near-surface wind speed and the like, which are not specifically limited here.
[0049] Step S07, model production; The image data is obtained as the input of the global terrestrial ecosystem carbon cycle component collaborative inversion model, the carbon cycle component inversion result is obtained, and the global-scale terrestrial ecosystem carbon cycle component product is produced.
[0050] In summary, the embodiment of the present application can provide a global terrestrial ecosystem carbon cycle component remote sensing product which meets the quality conservation constraint and high precision at the same time, which helps to improve the understanding of the dynamic changes of the terrestrial ecosystem carbon cycle, and can provide a solid foundation for global carbon accounting and carbon trading.
[0051] Embodiment 2 The embodiment is based on the technical solution provided in Embodiment 1, and provides a global terrestrial ecosystem carbon cycle component co-inversion system. The global terrestrial ecosystem carbon cycle component co-inversion system is applied to the global terrestrial ecosystem carbon cycle component co-inversion method provided in Embodiment 1. Referring to Figure 2 As shown in the figure, the system comprises: A data acquisition unit is configured to acquire global-scale ground station observation data and image data. The ground station observation data includes original carbon flux data, and the image data includes original remote sensing image data representing ground vegetation and terrain information and original meteorological reanalysis data representing meteorological and soil conditions. A data preprocessing unit is configured to preprocess the ground station observation data and the image data respectively. The preprocessing of the ground station observation data includes carbon balance correction processing to obtain carbon flux data satisfying the mass conservation constraint. A database construction unit is configured to construct a database based on the preprocessed observation data and image data. A refined carbon component calculation unit is configured to perform carbon component separation at a station scale on the carbon flux data satisfying the mass conservation constraint to obtain carbon cycle components of refined components. A data matching unit is configured to extract pixel values corresponding to each station from the image data at a pixel scale according to the geographical position information of the station, and match the carbon cycle components to construct a data set including a training set and a test set. A data set training unit is configured to use the carbon cycle components as model dependent variables and the image data as model independent variables, to establish a random forest model using a random forest method, to train the random forest model using the data set, to obtain a plurality of parameter-optimized random forest models, to evaluate each parameter-optimized random forest model using the test set, and to determine a global terrestrial ecosystem carbon cycle component co-inversion model from the plurality of parameter-optimized random forest models based on the evaluation results. A model production unit is configured to obtain image data as input of the global terrestrial ecosystem carbon cycle component co-inversion model to obtain carbon cycle component inversion results.
[0052] The functions of the modules of the global terrestrial ecosystem carbon cycle component co-inversion system of the embodiment are the same as the explanation of the embodiment of the global terrestrial ecosystem carbon cycle component co-inversion method, and the technical effects are the same. Therefore, the repeated explanation is omitted here.
[0053] Embodiment 3 The embodiment provides an electronic device based on the technical scheme provided in the embodiment 1, and the electronic device comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, the global land ecosystem carbon cycle component collaborative inversion method is realized.
[0054] The above merely provides the preferred embodiments of the present application but not for limiting the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A method for synergistic inversion of components of global terrestrial ecosystem carbon cycle, characterized in that, The method comprises the following steps: Obtaining global-scale ground station observation data and image data, wherein the ground station observation data comprises original carbon flux data, and the image data comprises original remote sensing image data representing ground vegetation and terrain information and original meteorological reanalysis data representing meteorological and soil conditions; Preprocessing the ground station observation data and the image data respectively, wherein the preprocessing of the ground station observation data comprises carbon balance correction processing to obtain carbon flux data satisfying a mass conservation constraint; Building a database based on the preprocessed observation data and image data; Carrying out site-scale carbon component separation on the carbon flux data satisfying the mass conservation constraint to obtain carbon cycle components of refined components; According to the geographical location information of the site, extracting the pixel value corresponding to each site from the image data at the pixel scale and matching the pixel value with the carbon cycle components to build a data set; Taking the carbon cycle components as model dependent variables and the image data as model independent variables, building a random forest model by using a random forest method, training the random forest model by using the data set, and obtaining a global terrestrial ecosystem carbon cycle component collaborative inversion model; Obtaining image data as input of the global terrestrial ecosystem carbon cycle component collaborative inversion model to obtain carbon cycle component inversion results.
2. The method for synergistic inversion of components of global terrestrial ecosystem carbon cycle according to claim 1, wherein, The original carbon flux data comprises gross primary productivity, ecosystem respiration and ecosystem net exchange; The remote sensing image data comprises normalized vegetation index, leaf area index, sunlight-induced chlorophyll fluorescence and elevation data; The meteorological reanalysis data comprises downwelling shortwave radiation, near-surface air temperature, dew point temperature, soil temperature, soil moisture and near-surface wind speed.
3. The method for synergistic inversion of components of global terrestrial ecosystem carbon cycle according to claim 2, wherein, The steps of carbon balance correction are as follows: Judging whether the original carbon flux data satisfies the mass conservation constraint, wherein the expression of the mass conservation constraint is as follows: In the above formulae, represents the total primary productivity, represents the ecosystem respiration, represents the net exchange of the ecosystem; Calculating the carbon non-conservation amount and the carbon non-conservation ratio of the original carbon flux data not satisfying the mass conservation constraint respectively, wherein the expressions are as follows: In the above formulae, represents the amount of carbon non-conservation, represents the carbon non-conservation ratio; Screening acquisition And The original carbon flux data meeting the set threshold value at the same time, determine the original carbon flux data to be processed; According to the proportion of gross primary productivity and ecosystem respiration in the original carbon flux data to be processed, distributing the carbon non-conservation amount to the gross primary productivity and the ecosystem respiration respectively in proportion to obtain carbon balance corrected gross primary productivity and ecosystem respiration, wherein the expression is as follows: In the above formulae, denotes the total primary productivity in the raw carbon flux data to be processed, denotes the ecosystem respiration in the raw carbon flux data to be processed, denotes the total primary productivity after carbon balance correction, denotes the ecosystem respiration after carbon balance correction.
4. The method for synergistic inversion of components of global terrestrial ecosystem carbon cycle of claim 1, wherein, The preprocessing of the ground station observation data further comprises quality control and scale conversion; The quality control is to filter out measured data and high-precision interpolated data from the original carbon flux data according to a quality control symbol to obtain high-quality data, and to perform carbon balance correction on the high-quality data; The scale conversion is to aggregate the original carbon flux data at the hourly scale to the required time scale.
5. The method for synergistic inversion of components of global terrestrial ecosystem carbon cycle of claim 1, wherein, The preprocessing of the original image data comprises quality control and spatio-temporal resolution unification processing; The quality control is to retain high-quality pixel values according to a quality control symbol; The spatio-temporal resolution unification processing is to unify the original image data to the same spatio-temporal resolution by using a resampling or interpolation method to obtain spatio-temporally unified image data.
6. The method for synergistic inversion of components of global terrestrial ecosystem carbon cycle of claim 3, wherein, The carbon cycle component of the refined component includes total primary productivity, net primary productivity, ecosystem total respiration, ecosystem autotrophic respiration, ecosystem heterotrophic respiration, and ecosystem net carbon exchange.
7. The method for synergistic inversion of components of global terrestrial ecosystem carbon cycle of claim 6, wherein, The carbon flux data satisfying the mass conservation constraint is subjected to site-scale carbon component separation, and the specific steps are as follows: According to the proportion of net primary productivity and total primary productivity of different vegetation types of the site, the net primary productivity and the ecosystem autotrophic respiration are separated from the total primary productivity corrected by the carbon balance, and the expression is as follows: In the above formulae, NPP represents net primary productivity, NPP represents the ratio of net primary productivity and gross primary productivity of different vegetation types, NPP represents autotrophic respiration of the ecosystem; The ecosystem heterotrophic respiration is separated from the ecosystem autotrophic respiration, and the expression is as follows: In the above formulae, represents the ecosystem heterotrophic respiration.
8. The method for synergistic inversion of components of global terrestrial ecosystem carbon cycle of claim 1, wherein, The random forest model is trained using the data set, specifically including: model parameter tuning is performed by random search method and grid search method to determine the optimal parameter configuration.
9. A system for synergistic inversion of components of global terrestrial ecosystem carbon cycle, characterized in that, The system comprises the collaborative inversion method of global terrestrial ecosystem carbon cycle components according to any one of claims 1 to 8. A data acquisition unit is configured to acquire global-scale ground station observation data and image data, wherein the ground station observation data includes original carbon flux data, and the image data includes original remote sensing image data representing ground vegetation and terrain information and original meteorological reanalysis data representing meteorological and soil conditions. A data preprocessing unit is configured to preprocess the ground station observation data and image data, wherein the preprocessing of the ground station observation data includes carbon balance correction processing to obtain carbon flux data satisfying the mass conservation constraint. A database construction unit is configured to construct a database based on the preprocessed observation data and image data. A refined carbon component calculation unit is configured to separate carbon components at a site scale from the carbon flux data satisfying the mass conservation constraint to obtain carbon cycle components of refined components. A data matching unit is configured to extract pixel values corresponding to each site from image data at a pixel scale according to geographical location information of the site, and match the carbon cycle components to construct a data set, wherein the data set includes a training set and a test set. A data set training unit is configured to use the carbon cycle components as model dependent variables and the image data as model independent variables to establish a random forest model using a random forest method, train the random forest model using the training set, perform model parameter tuning by random search method and grid search method, obtain a plurality of parameter-optimized random forest models, evaluate each parameter-optimized random forest model using the test set, determine an optimal parameter configuration from the plurality of parameter-optimized random forest models based on the evaluation results, and obtain a global terrestrial ecosystem carbon cycle component collaborative inversion model. A model production unit is configured to obtain image data as input of the global terrestrial ecosystem carbon cycle component collaborative inversion model to obtain carbon cycle component inversion results.
10. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the collaborative inversion method of global terrestrial ecosystem carbon cycle components according to any one of claims 1 to 8.