A data-driven collaborative inversion method and system for key carbon-water elements

Abstract

This invention relates to the field of carbon and water flux remote sensing inversion technology, and provides a data-driven method and system for the co-inversion of key carbon and water elements. The method includes: acquiring the data required for the data-driven co-inversion of key carbon and water elements, and constructing a co-inversion database of key carbon and water elements; preprocessing the co-inversion database of key carbon and water elements to obtain a driving dataset for the co-inversion of key carbon and water elements; constructing a co-inversion model of key carbon and water elements; training the co-inversion model of key carbon and water elements to obtain the optimal co-inversion model and co-inverting global evapotranspiration, gross primary productivity, and ecosystem water use efficiency. This solves the problem that existing data-driven models fail to effectively consider the coupling relationship between gross primary productivity and evapotranspiration, resulting in inaccurate results and significant discrepancies with observational data, which brings considerable uncertainty to the study of changes and attributions in global ecosystem water use efficiency.

Description

Technical Field

[0001] This invention relates to the field of carbon-water flux remote sensing inversion technology, and more specifically, to a data-driven collaborative inversion method and system for key carbon-water elements. Background Technology

[0002] Key carbon-water elements include gross primary productivity (GMP), evapotranspiration (evapotranspiration), and ecosystem water use efficiency (EBQ), and their accurate estimation is crucial for water resource management and food security. EBQ, the ratio of GMP to evapotranspiration, links the carbon-water cycle. Remote sensing inversion methods for evapotranspiration and GMP at global and regional scales are quite mature, while EBQ is primarily obtained by estimating the ratio of GMP to evapotranspiration at global and regional scales. Existing models for estimating GMP, evapotranspiration, and EBQ fall into two main categories: non-carbon-water coupled products and carbon-water coupled products. Non-carbon-water coupled products, such as MODIS, estimate GMP and evapotranspiration separately and then solve for EBQ. Carbon-water coupled products, such as PML_V2, use a stomatal conductance model to couple vegetation photosynthesis and transpiration, simultaneously estimating evapotranspiration and GMP, and then solving for EBQ. Although existing models can obtain reliable evapotranspiration and gross primary productivity, the results of ecosystem water use efficiency obtained from these two are not very accurate. The inversion errors of gross primary productivity and evapotranspiration will increase the error and uncertainty of the ecosystem water use efficiency estimation results.

[0003] While the emergence of data-driven models such as machine learning and deep learning has improved the accuracy of estimating gross primary productivity and evapotranspiration, it still involves independently modeling and estimating the two and then calculating the ratio to obtain ecosystem water use efficiency. This fails to effectively consider the coupling relationship between gross primary productivity and evapotranspiration, resulting in inaccurate ecosystem water use efficiency results that differ significantly from observational results. This introduces considerable uncertainty into research on changes and attributions in global ecosystem water use efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide a data-driven method and system for the synergistic inversion of key carbon-water elements. This aims to address the problem that existing data-driven models fail to effectively consider the coupling relationship between total primary productivity and evapotranspiration, resulting in inaccurate ecosystem water use efficiency results that differ significantly from observational results, thus introducing considerable uncertainty into the study of changes and attributions in global ecosystem water use efficiency.

[0005] The embodiments of the present invention are achieved through the following technical solutions: A data-driven method for the synergistic inversion of key carbon-water elements includes the following steps: Acquire the data required for data-driven co-inversion of key carbon-water elements and construct a corresponding co-inversion database of key carbon-water elements; wherein, the key carbon-water elements include ecosystem water use efficiency, gross primary productivity and evapotranspiration. The data in the carbon-water key element co-inversion database are preprocessed to obtain the co-inversion driving dataset; A data-driven co-inversion model for key carbon and water elements is constructed, with the co-inversion driving dataset serving as the input to the model. The expression of the data-driven co-inversion model for key carbon and water elements is as follows (1): (1) in, For ecosystem water use efficiency. For total primary productivity, For evaporation, For wind speed, This refers to the relative humidity of the air. This is the saturated water vapor pressure reverse. For air temperature, Atmospheric pressure. Net radiation, For incident shortwave radiation, For incident long-wave radiation, It is chlorophyll fluorescence. For surface temperature, Normalized Difference Vegetation Index (NDVI) The proportion of photosynthetically active radiation absorbed. For vegetation coverage, Leaf area index, Albedo, For elevation, For soil temperature, Soil moisture content; The data-driven co-inversion model of carbon-water key elements was trained based on the co-inversion driven dataset. During training, a cost function was set, and the model parameters with the minimum cost function were obtained by grid search to obtain the optimal data-driven co-inversion model of carbon-water key elements. The expression of the cost function is shown in the following equation (2): (2) in, Represents the cost function, This is an estimate of the ecosystem's water use efficiency. Ecosystem water use efficiency calculated from site observations. The coefficient to be calibrated for the ecosystem water use efficiency term. This is an estimate of total primary productivity. The total primary productivity is obtained by separating the total primary productivity observed at the site or the total primary productivity obtained by separating the net ecosystem carbon exchange observed at the site. The coefficient to be determined for the total primary productivity item. This is an estimated value for evapotranspiration. Evapotranspiration observed at the station The coefficient to be determined for the evaporation term; By using the optimal data-driven co-inversion model for key carbon-water elements, and based on relevant remote sensing products and meteorological reanalysis products, global ecosystem water use efficiency, total primary productivity, and evapotranspiration are co-inverted.

[0006] Optionally, the data required for the data-driven collaborative inversion of key carbon-water elements includes: observation data from global flux network sites, remote sensing products, and meteorological reanalysis products.

[0007] Optionally, the specific process of preprocessing the carbon-water key element co-inversion database to obtain the co-inversion driving dataset is as follows: Based on the total primary productivity and evapotranspiration observed at global flux network stations, the ecosystem water use efficiency observed at the stations is obtained, and the expression for the ecosystem water use efficiency observed at the stations is shown in the following equation (3):

[0008] in, This represents the ecosystem water use efficiency calculated from site observations. This represents the total primary productivity observed at the site. This represents the evapotranspiration observed at the station; Based on the latitude and longitude of the Global Flux Network stations, the data values ​​of the meteorological reanalysis grid and remote sensing pixels corresponding to each Global Flux Network station are extracted. The obtained data values ​​of the corresponding meteorological reanalysis grid and remote sensing pixels are combined with the observation data of the Global Flux Network stations to obtain the collaborative inversion driving dataset.

[0009] Optionally, the total primary productivity can be obtained directly from observation data of the Global Flux Network sites or obtained by solving for net ecosystem carbon exchange data from the observation data of the Global Flux Network sites.

[0010] Optionally, the specific process of using the collaborative inversion driving dataset as input to the data-driven collaborative inversion model of key carbon-water elements is as follows: All available data in the collaborative inversion driving dataset were randomly sampled and divided into an 80% model training set and a 20% model validation set for each vegetation type. For each vegetation type, the data-driven collaborative inversion model of carbon-water key elements was trained using the 80% model training set with ten-fold cross-validation, and the ecological water use efficiency inversion model was validated using the 20% model validation set.

[0011] This invention also provides a data-driven collaborative inversion system for key carbon-water elements, comprising: The data acquisition module is used to acquire global flux station observation data, remote sensing products, and reanalysis products; The data preprocessing module is used to preprocess the collected global flux network site observation data, remote sensing products, and meteorological reanalysis products. The model building and training module is used to build and train a data-driven carbon-water key element co-inversion model; the expression of the data-driven carbon-water key element co-inversion model is shown in equation (1). The inversion calculation module is used to perform the co-inversion of ecosystem water use efficiency, total primary productivity and evapotranspiration based on the corresponding global flux station observation data, remote sensing products and reanalysis products, using the optimal data-driven carbon-water key element co-inversion model; a cost function is set during training, and the expression of the cost function is shown in equation (2). The results verification and correction module is used to drive the verification and correction of the dataset by using the collaborative inversion of key carbon-water elements.

[0012] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the above-described data-driven co-inversion method for key carbon-water elements.

[0013] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described data-driven co-inversion method for key carbon-water elements.

[0014] The technical solutions of the embodiments of the present invention have at least the following advantages and beneficial effects: This invention provides a data-driven method for the co-inversion of key carbon-water elements, comprising the following steps: acquiring the data required for the data-driven co-inversion of key carbon-water elements and constructing a corresponding co-inversion database; wherein the key carbon-water elements include ecosystem water use efficiency, gross primary productivity, and evapotranspiration; preprocessing the data in the co-inversion database to obtain a co-inversion driving dataset; constructing a data-driven co-inversion model for key carbon-water elements, using the co-inversion driving dataset as input to the model; training the model based on the co-inversion driving dataset, setting a cost function during training, and obtaining the cost function through a grid search method. The optimal data-driven co-inversion model for key carbon-water elements was obtained by minimizing the model parameters of the function. Using this optimal model, global ecosystem water use efficiency, gross primary productivity (GMP), and evapotranspiration (evapotranspiration) were co-inverted based on relevant remote sensing and meteorological reanalysis products. This approach ensured high inversion accuracy for evapotranspiration, GMP, and ecosystem water use efficiency, while also addressing the problem of inaccurate ecosystem water use efficiency estimations caused by existing data-driven models that independently modeled evapotranspiration and GMP and failed to effectively consider the coupling relationship between them. This further improved the reliability of studies on changes and attributions in global GMP, evapotranspiration, and ecosystem water use efficiency. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the data-driven collaborative inversion method for key carbon-water elements in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram of the data-driven carbon-water key element collaborative inversion system in Embodiment 2 of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0017] Example 1 Reference Figure 1 A data-driven method for the synergistic inversion of key carbon-water elements includes the following steps: Step 1: Obtain the data required for data-driven co-inversion of key carbon-water elements and construct the corresponding co-inversion database of key carbon-water elements; wherein, the key carbon-water elements include ecosystem water use efficiency, gross primary productivity and evapotranspiration. In this embodiment, the data required for data-driven collaborative inversion of key carbon-water elements includes: observation data from global flux network sites, remote sensing products, and meteorological reanalysis products; Global flux network sites include: Fluxnet, Ameriflux, OzFlux, EuroFlux, HiWATER (Heihe Eco-hydrological Remote Sensing Experiment), and CARN (Lanzhou University Cold and Arid Regions Scientific Observation Network), among other international and domestic flux observation network sites. Global flux network station observation data includes: station latitude and longitude data, automatic weather station observation data, eddy covariance instrument observation data, and ground vegetation height measurement data, etc. Automatic weather station observation data include: wind speed, air humidity, soil moisture, air temperature, soil temperature, atmospheric pressure, water vapor pressure deficit, net radiation, incident shortwave radiation, and incident longwave radiation, etc. Eddy covariance data include: friction velocity, evapotranspiration, gross primary productivity, and net ecosystem carbon exchange. Remote sensing products include: GOSIF chlorophyll fluorescence, MOD11A1 land surface temperature, MCD43A4 reflectance, MCD15A3H photosynthetically active radiation absorption ratio, GLASS vegetation cover, GLASS leaf area index, GLASS albedo, and GMTED2010 elevation products, etc. Meteorological reanalysis products include: wind speed, relative humidity, saturated vapor pressure deficit, air temperature, atmospheric pressure, soil moisture content, net radiation, incident shortwave radiation, and incident longwave radiation. Meteorological reanalysis products can include ERA5-Land reanalysis product, ERA5 reanalysis product, ERA-Interim reanalysis product, MERRA-2 reanalysis product, or JRA55 reanalysis product, etc.

[0018] Step 2: Preprocess the carbon-water key element co-inversion database to obtain the carbon-water key element co-inversion driving dataset; In this embodiment, the specific process of preprocessing the carbon-water key element co-inversion database to obtain the carbon-water key element co-inversion driving dataset is as follows: Based on the total primary productivity and evapotranspiration observed by global flux network stations, the ecosystem water use efficiency calculated from the station observations is obtained. The expression for the ecosystem water use efficiency calculated from the station observations is shown in the following equation (3):

[0019] in, This represents the ecosystem water use efficiency calculated from site observations. This represents the total primary productivity observed at the site. This represents the evapotranspiration observed at the station; In this embodiment, the MODIS Normalized Difference Vegetation Index is calculated from the MCD43A4 reflectance product. Based on the latitude and longitude of the Global Flux Network stations, the data values ​​of the meteorological reanalysis grid and remote sensing pixels corresponding to each Global Flux Network station are extracted. The obtained data values ​​of the corresponding meteorological reanalysis grid and remote sensing pixels are combined with the observation data of the Global Flux Network stations to obtain the carbon-water key element collaborative inversion driving dataset.

[0020] Total primary productivity (TPP) is obtained directly from observational data of Global Flux Network (GFLN) stations or by separating and solving from net ecosystem carbon exchange (NAC) data from GFLN station observational data. In this embodiment, if some GFLN stations do not directly provide TTP but provide NAC data, TTP is obtained by separating and solving from NAC using appropriate methods. In this embodiment, the appropriate methods are daytime-based or nighttime-based. If the TTP of some stations is obtained by separating and solving from NAC, when combining it with other stations that provide TTP, it is important to note that the TTP obtained by separating and solving from the daytime-based method of stations that do not provide TTP should be combined with the TTP obtained by the daytime-based method of stations that provide TTP; or the TTP obtained by separating and solving from the nighttime-based method of stations that do not provide TTP should be combined with the TTP obtained by the nighttime-based method of stations that provide TTP.

[0021] Step 3: Construct a data-driven co-inversion model for key carbon and water elements. Use the co-inversion driving dataset as the input to the co-inversion model for key carbon and water elements, and use the evapotranspiration, total primary productivity, and ecosystem water use efficiency observed by coupled global flux network stations as the output of the data-driven co-inversion model for key carbon and water elements. In this embodiment, the data-driven co-inversion model for key carbon-water elements is a deep learning model, and the data-driven co-inversion model for key carbon-water elements is as follows: The expression for the data-driven synergistic inversion model of key carbon-water elements is as follows (1): (1) in, For ecosystem water use efficiency. For total primary productivity, For evaporation, For wind speed, This refers to the relative humidity of the air. This is the saturated water vapor pressure reverse. For air temperature, Atmospheric pressure. Net radiation, For incident shortwave radiation, For incident long-wave radiation, It is chlorophyll fluorescence. For surface temperature, Normalized Difference Vegetation Index (NDVI) The proportion of photosynthetically active radiation absorbed. For vegetation coverage, Leaf area index, Albedo, For elevation, For soil temperature, This refers to soil moisture content.

[0022] In this embodiment, the specific process of using the collaborative inversion driving dataset as the input to the carbon-water key element collaborative inversion model is as follows: All available data in the co-inversion driving dataset were randomly sampled and divided into 80% model training set and 20% model validation set according to each vegetation type. For each vegetation type, the data-driven carbon-water key element co-inversion model was trained using the 80% model training set with ten-fold cross-validation, and the ecosystem water use efficiency inversion model was validated using the 20% model validation set.

[0023] Step 4: Train the data-driven carbon-water key element co-inversion model based on the co-inversion driven dataset. During training, set the cost function and obtain the model parameters with the minimum cost function through grid search to obtain the optimal data-driven carbon-water key element co-inversion model. In this embodiment, the cost function is: The expression for the cost function is shown in equation (2) below: (2) in, Represents the cost function, This is an estimate of the ecosystem's water use efficiency. Ecosystem water use efficiency calculated from site observations. The coefficient to be calibrated for the ecosystem water use efficiency term. This is an estimate of total primary productivity. The total primary productivity is obtained by separating the total primary productivity observed at the site or the total primary productivity obtained by separating the net ecosystem carbon exchange observed at the site. The coefficient to be determined for the total primary productivity item. This is an estimated value for evapotranspiration. Evapotranspiration observed at the station The constant coefficient is the coefficient to be determined for the evaporation term.

[0024] Step 5: Using the optimal data-driven co-inversion model for key carbon-water elements, and based on relevant remote sensing products and meteorological reanalysis products, co-invert global ecosystem water use efficiency, gross primary productivity, and evapotranspiration.

[0025] Example 2 In order to execute the method corresponding to Embodiment 1 above and achieve the corresponding functions and technical effects, refer to Figure 2 This embodiment provides a data-driven collaborative inversion system for key carbon-water elements, including: The data acquisition module 101 is used to acquire observation data, remote sensing products, and meteorological reanalysis products from global flux network stations, including station latitude and longitude data, automatic weather station observation data, eddy covariance instrument observation data, and ground vegetation height measurement data from global flux network stations such as Fluxnet, Ameriflux, OzFlux, EuroFlux, HiWATER (Heihe Eco-hydrological Remote Sensing Experiment), and CARN (Lanzhou University Cold and Arid Region Scientific Observation Network); remote sensing products such as GOSIF chlorophyll fluorescence, MOD11A1 surface temperature, MCD43A4 reflectance, MCD15A3H photosynthetically active radiation absorption ratio, GLASS vegetation cover, GLASS leaf area index, GLASS albedo, and GMTED2010 elevation; and meteorological reanalysis products such as ERA5-Land reanalysis product, ERA5 reanalysis product, ERA-Interim reanalysis product, MERRA-2 reanalysis product, or JRA55 reanalysis product.

[0026] The data preprocessing module 102 is used to preprocess the collected observation data, remote sensing products, and meteorological reanalysis products from the global flux network stations. This includes: obtaining the ecosystem water use efficiency calculated from the station observations based on the total primary productivity and evapotranspiration from the global flux network stations; if the observation data from some global flux network stations does not directly provide total primary productivity but provides net ecosystem carbon exchange data, then the total primary productivity is obtained by separating and solving from the net ecosystem carbon exchange using daytime-based or nighttime-based methods; extracting the data values ​​of the meteorological reanalysis grids and remote sensing pixels corresponding to each global flux network station based on their latitude and longitude, and combining the obtained data values ​​of the corresponding meteorological reanalysis grids and remote sensing pixels with the observation data from the global flux network stations. Through data preprocessing, a comprehensive and accurate data foundation with consistent format and resolution is provided for subsequent model construction and inversion calculations.

[0027] The model building and training module 103 is used to build and train a data-driven carbon-water key element co-inversion model; based on a data-driven carbon-water key element co-inversion method in Example 1, a data-driven carbon-water key element co-inversion model is built and the model is trained accordingly to obtain the optimal data-driven carbon-water key element co-inversion model.

[0028] The inversion calculation module 104 is used to perform the synergistic inversion of ecosystem water use efficiency, total primary productivity and evapotranspiration based on the corresponding remote sensing products and meteorological reanalysis products, using the optimal data-driven synergistic inversion model of key carbon-water elements.

[0029] The result verification and calibration module 105 is used to verify and calibrate the dataset by using the collaborative inversion of carbon-water key elements. Through comparative analysis, it evaluates the accuracy and reliability of the model estimation and makes necessary adjustments and optimizations to the model.

[0030] The data-driven carbon-water key element co-inversion system in this embodiment integrates multiple functional modules, including data acquisition, data preprocessing, model building and training, inversion calculation, and result verification and correction, achieving fully automated processing from data to results. Employing advanced algorithms and models, it couples evapotranspiration and total primary productivity (TPP) through co-inversion of TPP, improving the estimation accuracy of TPP. It supports real-time data processing and result output, providing timely and effective decision support for ecological water resource management and agricultural irrigation optimization. The system architecture is flexible, allowing for expansion of functional modules and data sources according to actual needs, adapting to application requirements in different regions and scenarios. It is widely applicable in agriculture, forestry, water resource management, and environmental protection, providing technical support and decision-making basis for the efficient utilization and sustainable development of ecological water resources. For example, in agricultural irrigation, estimating crop TPP allows for optimization of irrigation schemes, improving water resource utilization efficiency; in ecological environment monitoring, it assesses the ecosystem's water use status, providing a scientific basis for ecological protection and environmental restoration.

[0031] Example 3 This embodiment provides an electronic device, including a memory and a processor. The memory is used to store computer programs, and the processor runs the computer programs to enable the electronic device to execute the data-driven carbon-water key element co-inversion method of Embodiment 1.

[0032] Alternatively, the aforementioned electronic device may be a server.

[0033] In addition, this embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the data-driven carbon-water key element co-inversion method of Embodiment 1.

[0034] It is understood that the processor in the embodiments of the present invention can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The general-purpose processor can be a microprocessor or any conventional processor.

[0035] The method steps in the embodiments of the present invention can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0036] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a storage medium or transmitted through the storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0037] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A data-driven method for the synergistic inversion of key carbon-water elements, characterized in that, Includes the following steps: Acquire the data required for data-driven co-inversion of key carbon-water elements and construct a corresponding co-inversion database of key carbon-water elements; wherein, the key carbon-water elements include ecosystem water use efficiency, gross primary productivity and evapotranspiration. The data in the carbon-water key element co-inversion database are preprocessed to obtain the co-inversion driving dataset; A data-driven co-inversion model for key carbon and water elements is constructed, with the co-inversion driving dataset serving as the input to the model. The expression of the data-driven co-inversion model for key carbon and water elements is as follows (1): （1） in, For ecosystem water use efficiency. For total primary productivity, For evaporation, For wind speed, This refers to the relative humidity of the air. This is the saturated water vapor pressure reverse. For air temperature, Atmospheric pressure. Net radiation, For incident shortwave radiation, For incident long-wave radiation, It is chlorophyll fluorescence. For surface temperature, Normalized Difference Vegetation Index (NDVI) The proportion of photosynthetically active radiation absorbed. For vegetation coverage, Leaf area index, Albedo, For elevation, For soil temperature, Soil moisture content; The data-driven co-inversion model of carbon-water key elements was trained based on the co-inversion driven dataset. During training, a cost function was set, and the model parameters with the minimum cost function were obtained by grid search to obtain the optimal data-driven co-inversion model of carbon-water key elements. The expression of the cost function is shown in the following equation (2): （2） in, Represents the cost function, This is an estimate of the ecosystem's water use efficiency. Ecosystem water use efficiency calculated from site observations. The coefficient to be calibrated for the ecosystem water use efficiency term. This is an estimate of total primary productivity. The total primary productivity is obtained by separating the total primary productivity observed at the site or the total primary productivity obtained by separating the net ecosystem carbon exchange observed at the site. The coefficient to be determined for the total primary productivity item. This is an estimated value for evapotranspiration. Evapotranspiration observed at the station The coefficient to be determined for the evaporation term; By using the optimal data-driven co-inversion model for key carbon-water elements, and based on relevant remote sensing products and meteorological reanalysis products, global ecosystem water use efficiency, total primary productivity, and evapotranspiration are co-inverted.

2. The data-driven collaborative inversion method for key carbon-water elements as described in claim 1, characterized in that, The data required for the data-driven collaborative inversion of key carbon-water elements includes: observation data from global flux network sites, remote sensing products, and meteorological reanalysis products.

3. The data-driven collaborative inversion method for key carbon-water elements as described in claim 2, characterized in that, The specific process of preprocessing the carbon-water key element collaborative inversion database to obtain the collaborative inversion driving dataset is as follows: Based on the total primary productivity and evapotranspiration observed at global flux network stations, the ecosystem water use efficiency observed at the stations is obtained, and the expression for the ecosystem water use efficiency observed at the stations is shown in the following equation (3): in, This represents the ecosystem water use efficiency calculated from site observations. This represents the total primary productivity observed at the site. This represents the evapotranspiration observed at the station; Based on the latitude and longitude of the Global Flux Network stations, the data values ​​of the meteorological reanalysis grid and remote sensing pixels corresponding to each Global Flux Network station are extracted. The obtained data values ​​of the corresponding meteorological reanalysis grid and remote sensing pixels are combined with the observation data of the Global Flux Network stations to obtain the collaborative inversion driving dataset.

4. The data-driven synergistic inversion method for key carbon-water elements as described in claim 3, characterized in that, The total primary productivity is obtained directly from observation data from Global Flux Network sites or by solving for net ecosystem carbon exchange data from Global Flux Network sites.

5. The data-driven collaborative inversion method for key carbon-water elements as described in claim 1, characterized in that, The specific process of using the collaborative inversion driving dataset as input to the data-driven collaborative inversion model of key carbon-water elements is as follows: All available data in the co-inversion driving dataset were randomly sampled and divided into an 80% model training set and a 20% model validation set for each vegetation type. For each vegetation type, the data-driven co-inversion model of carbon-water key elements was trained using the 80% model training set with ten-fold cross-validation, and the data-driven co-inversion model of carbon-water key elements was validated using the 20% model validation set.

6. A data-driven collaborative inversion system for key carbon-water elements, characterized in that, include: The data acquisition module is used to acquire global flux station observation data, remote sensing products, and reanalysis products; The data preprocessing module is used to preprocess the collected global flux network site observation data, remote sensing products, and meteorological reanalysis products. The model building and training module is used to establish and train a data-driven co-inversion model of key carbon and water elements; the expression of the data-driven co-inversion model of key carbon and water elements is as follows (1): （1） in, For ecosystem water use efficiency. For total primary productivity, For evaporation, For wind speed, This refers to the relative humidity of the air. This is the saturated water vapor pressure reverse. For air temperature, Atmospheric pressure. Net radiation, For incident shortwave radiation, For incident long-wave radiation, It is chlorophyll fluorescence. For surface temperature, Normalized Difference Vegetation Index (NDVI) The proportion of photosynthetically active radiation absorbed. For vegetation coverage, Leaf area index, Albedo, For elevation, For soil temperature, Soil moisture content; The inversion calculation module is used to perform the co-inversion of ecosystem water use efficiency, total primary productivity and evapotranspiration based on the corresponding global flux station observation data, remote sensing products and reanalysis products, using the optimal data-driven co-inversion model of carbon-water key elements; a cost function is set during training, and the expression of the cost function is shown in the following equation (2): （2） in, Represents the cost function, This is an estimate of the ecosystem's water use efficiency. Ecosystem water use efficiency calculated from site observations. The coefficient to be calibrated for the ecosystem water use efficiency term. This is an estimate of total primary productivity. The total primary productivity is obtained by separating the total primary productivity observed at the site or the total primary productivity obtained by separating the net ecosystem carbon exchange observed at the site. The coefficient to be determined for the total primary productivity item. This is an estimated value for evapotranspiration. Evapotranspiration observed at the station The coefficient to be determined for the evaporation term; The results verification and correction module is used to drive the verification and correction of the dataset by using the collaborative inversion of key carbon-water elements.

7. An electronic device, characterized in that, The device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the data-driven co-inversion method for key carbon-water elements as described in any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the data-driven co-inversion method for key carbon-water elements according to any one of claims 1-6.

Images