Knowledge-guided and process-driven forest soil layered carbon sink simulation system

Abstract

The invention discloses a knowledge-guided and process-driven forest soil layered carbon sink simulation system, which comprises the following steps of: screening a continuous and complete soil profile according to forest soil organic carbon layered observation data; optimizing process model parameters of the terrestrial ecosystem at each measuring point by adopting a Bayesian data assimilation technology; processing the current vegetation terrain soil attribute data to obtain multi-modal standardized data as an input variable of a random forest model, performing heterogeneity learning and scale expansion on optimization parameters by using the random forest model, and simulating the organic carbon reserve of the current layered soil based on steady-state hypothesis and external driving data of a lattice point scale; and based on an instantaneous hypothesis, generating a future forest soil layered carbon sink simulation result by taking scale extension parameters and future external driving data as driving and combining current layered soil organic carbon reserves, so as to realize accurate, digital and intelligent monitoring and evaluation of the forest soil carbon sink capacity.

Description

Technical Field

[0001] This invention relates to the field of ecological and environmental protection technology, and in particular to a knowledge-guided and process-driven forest soil stratification carbon sequestration simulation system. Background Technology

[0002] In recent years, forest soil carbon dynamics have been affected by climate change and intensified human activities. Given the important role of forest soil carbon sinks in nature-based solutions, accurate and efficient assessment of the current status and changes in forest soil carbon storage is of great significance for maintaining ecological security.

[0003] Currently, the estimation of soil carbon balance in forest ecosystems mainly adopts a bottom-up approach, which extends results from the sampling point scale to the regional scale. This approach primarily includes the inventory method, the eddy covariance method, and the ecosystem process simulation method. The inventory method directly calculates the carbon sink intensity by comparing soil organic carbon data of forest ecosystems at different periods. However, the inventory method has limitations such as a long inventory period and significant uncertainty. The eddy covariance method, based on micrometeorological principles, directly measures the net CO2 exchange between the forest ecosystem and the atmosphere within a given coverage area, and then estimates the net ecosystem productivity of the entire region through scale extension. The eddy covariance method can achieve long-term time-series observations of carbon flux, but still has considerable uncertainty. The ecosystem process simulation method estimates soil carbon dynamics within a given region by simulating the carbon cycle process of forest ecosystems. This method is adopted by the Global Carbon Programme for regional and global soil carbon sink assessments. Compared to the other two methods, the ecosystem process simulation method can not only predict the current soil organic carbon dynamics of forests but also predict future changes in forest soil carbon sinks and can quantitatively distinguish the contributions of different factors to carbon sink changes. However, due to limitations in model structure, parameterization scheme, and external driving data, this method also has certain uncertainties.

[0004] The explosive growth of ecological data and the emergence of advanced machine learning algorithms have provided an important opportunity for mining forest soil organic carbon information. The advantage of machine learning models lies in their high explanatory power in predicting soil organic carbon. However, machine learning models are essentially still statistical models, which means that the simulation results lack constraints on key processes of the carbon cycle and are relatively lacking in mechanistic explanations.

[0005] Therefore, there is an urgent need to leverage machine learning algorithms to provide a large-scale terrestrial ecosystem process model-big data fusion technology to address the significant uncertainties and lack of mechanistic explanations between soil carbon cycle process model simulation results and observational data. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a knowledge-guided and process-driven forest soil stratified carbon sequestration simulation system.

[0007] In a first aspect, the present invention provides a knowledge-guided and process-driven forest soil stratified carbon sequestration simulation system, the technical solution of which is as follows: It includes: a screening module, an optimization module, a processing module, an extension module, a simulation module, an extraction module, and a prediction module; The screening module is used to: select soil profiles with continuous and complete observations based on the stratified observation data of forest soil organic carbon in the target area; The optimization module is used to: determine multiple soil organic carbon observation points based on the soil profile, and optimize the parameters of the terrestrial ecosystem process model at each soil organic carbon observation point using Bayesian data assimilation technology based on external driving data, so as to obtain the optimized parameters of each soil organic carbon observation point. The processing module is used to: perform quality control, resampling and extraction processing on the current vegetation, topography and soil attribute related data of the target area to obtain multimodal normalized data; The extension module is used to: use the multimodal normalized data as input variables of the random forest model, and use the random forest model to perform heterogeneity learning on the optimization parameters of each soil organic carbon observation point and extend them to the overall forest ecosystem scale of the target area to obtain the scale extension parameters of the terrestrial ecosystem process model; The simulation module is used to: simulate the current stratified soil organic carbon storage in the target area using the scale extension parameters and based on the steady-state hypothesis and external driving data at the grid scale; The extraction module is used to: extract future vegetation, topography, and soil attribute data of the target area based on the simulation results of the CESM2 model of CMIP6; The prediction module is used to: generate knowledge-guided and process-driven simulation results of forest soil stratification carbon sink in the target area over a preset time period based on the transient hypothesis, driven by the scale expansion parameters and the relevant data of future vegetation, topography and soil properties, and combined with the current stratified soil organic carbon storage.

[0008] The beneficial effects of the knowledge-guided and process-driven forest soil stratified carbon sequestration simulation system of the present invention are as follows: The system of this invention can effectively solve the problem of large uncertainty in the current regional scale assessment of soil carbon dynamics, and is more conducive to understanding the carbon cycle mechanism of forest soil, thereby realizing accurate, digital and intelligent monitoring and assessment of the carbon sink capacity of forest soil.

[0009] Based on the above scheme, the knowledge-guided and process-driven forest soil stratified carbon sequestration simulation system of the present invention can be further improved as follows.

[0010] In one alternative approach, the filtering module is specifically used for: The stratified observation data of forest soil organic carbon in the target area were collected, and the quality control of the stratified observation data was carried out to select soil profiles with continuous and complete observations.

[0011] In one alternative approach, the filtering module is specifically used for: Five continuous and fully observed soil profiles at depths of 0-10 cm, 10-20 cm, 20-30 cm, 30-50 cm, and 50-100 cm were selected from the stratified observation data, and the soil profiles were evenly distributed in different forest ecosystems in the target area.

[0012] In one alternative approach, the optimization module is specifically used for: Based on the Markov chain Monte Carlo method, multiple independent parameter chains are constructed for each soil organic carbon observation point; each parameter chain contains a test chain with a first number of iterations and a formal chain with a second number of iterations. A target loss function is constructed to calculate the difference between simulated soil organic carbon storage and measured soil organic carbon storage in the terrestrial ecosystem process model, and the acceptance rate of the target loss function is set to a predetermined range value. Perform the data assimilation process for each soil organic carbon observation point, calculate the GR value for each soil organic carbon observation point, randomly select a parameter chain to generate the posterior probability distribution of each parameter on the corresponding soil organic carbon observation point, and delete soil organic carbon observation points with GR values ​​less than a predetermined threshold. Multiple random samplings were performed on the posterior probability distribution of the parameters, and the results of each sampling were applied to the calculation of the terrestrial ecosystem process model to obtain the average value of organic carbon storage in each soil layer as the optimization result of the corresponding soil organic carbon observation point. The consistency between the optimization results and the observation data of each soil organic carbon observation point was evaluated using the coefficient of determination and root mean square error evaluation index, and effective soil organic carbon observation points were screened. Simulation results of multiple model parameters were derived from all valid soil organic carbon observation points and used as optimization parameters.

[0013] In one alternative approach, the extension module is specifically used for: The environmental variables in the multimodal normalized data are normalized to obtain normalized environmental variable data, and the normalized environmental variable data is resampled to generate a spatially continuous normalized environmental variable map. The optimized parameters of each soil organic carbon observation point are used as the training target variable of the random forest model, and the normalized environmental variable data are used as input features to train the random forest model to learn the mapping relationship between environmental variables and model parameters. The random forest model is constructed and optimized using a cross-validation method that is repeated multiple times. The normalized environmental variable map is input into the trained random forest model to predict the values ​​of multiple parameters of the terrestrial ecosystem process model at each spatial grid point in the target area, so as to obtain the scale expansion parameters.

[0014] In one alternative embodiment, the simulation module is specifically used for: The scale expansion parameters and the external driving data at the grid scale are input into the matrix equation of the terrestrial ecosystem process model, and the matrix equation is as follows: ;in, This indicates the current organic carbon storage in the stratified soil, including the organic carbon storage in each current layer. This represents the vector of allocation coefficients from plant carbon input to each soil carbon pool. This represents plant carbon input, in units of: ; This represents the carbon transfer coefficient matrix between the litter carbon pool and the mineral soil carbon pool. This represents the matrix of background degradation rates for different carbon pools. A scalar function representing the impact of environmental variables on the degradation of different carbon pools. Represents the vertical transport rate matrix of soil organic carbon; When the soil system reaches a steady state The soil organic carbon storage of each layer in the target area is calculated based on the matrix equation.

[0015] In one alternative approach, the prediction module is specifically used for: The current stratified soil organic carbon storage is taken as the initial state, the scale expansion parameter is taken as the heterogeneity parameter of the terrestrial ecosystem process model at each grid point, and the future vegetation, topography, and soil attribute related data are taken as the external driving data of the terrestrial ecosystem process model. Based on the transient state hypothesis SOC i+1 =ΔSOC i +SOC iThe matrix equation is used to calculate the change in soil organic carbon storage ΔSOC for each year within the preset future time period in chronological order. i This change was then compared with the previous year's soil organic carbon (SOC) storage. i Add them together to get the soil organic carbon storage (SOC) for the next year. i+1 The predicted results of the stratified soil organic carbon storage for each year within the preset future time period are obtained. The arithmetic mean of the carbon sequestration intensity for each year within the predetermined future time period is calculated as the average carbon sequestration intensity, and the difference between the soil organic carbon storage in the last year of the predetermined future time period and the current soil organic carbon storage is calculated as the cumulative carbon sequestration amount; wherein, the soil organic carbon storage is the sum of all stratified soil organic carbon storage, and the average carbon sequestration intensity and the cumulative carbon sequestration amount together constitute the knowledge-guided and process-driven forest soil stratified carbon sequestration simulation results for the target area.

[0016] Secondly, this invention provides a knowledge-guided and process-driven method for simulating forest soil stratified carbon sequestration, employing a knowledge-guided and process-driven forest soil stratified carbon sequestration simulation system as provided in this application. The technical solution of this method is as follows: Based on the stratified observation data of forest soil organic carbon in the target area, soil profiles with continuous and complete observations were selected. Based on the soil profile, multiple soil organic carbon observation points were identified. Based on external driving data, Bayesian data assimilation technology was used to optimize the parameters of the terrestrial ecosystem process model at each soil organic carbon observation point, resulting in optimized parameters for each soil organic carbon observation point. The current vegetation, topography, and soil attribute data of the target area are subjected to quality control, resampling, and extraction processing to obtain multimodal normalized data; The multimodal normalized data is used as the input variable of the random forest model. The random forest model is then used to learn the heterogeneity of the optimization parameters for each soil organic carbon observation point and extend them to the overall forest ecosystem scale of the target area to obtain the scale extension parameters of the terrestrial ecosystem process model. Using the scale extension parameters and based on the steady-state hypothesis and external driving data at the grid scale, the current stratified soil organic carbon storage in the target area is simulated. Based on the simulation results of the CESM2 model of CMIP6, data related to future vegetation, topography, and soil properties of the target area are extracted. Based on the transient hypothesis, and driven by the scale expansion parameters and the relevant data of future vegetation, topography, and soil properties, and combined with the current stratified soil organic carbon storage, the simulation results of forest soil stratification carbon sink in the target area over a future preset time period are generated.

[0017] The beneficial effects of the knowledge-guided and process-driven forest soil stratified carbon sequestration simulation method of the present invention are as follows: The method of this invention can effectively solve the problem of large uncertainty in the current regional scale soil carbon dynamic assessment, and is more conducive to understanding the forest soil carbon cycle mechanism, thereby realizing accurate, digital and intelligent monitoring and assessment of forest soil carbon sink capacity.

[0018] Thirdly, the technical solution of an electronic device according to the present invention is as follows: It includes a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps of the knowledge-guided and process-driven forest soil stratification carbon sink simulation method of the present invention.

[0019] Fourthly, the technical solution of a computer-readable storage medium provided by the present invention is as follows: The computer-readable storage medium stores instructions that, when read, cause the computer-readable storage medium to perform the steps of the knowledge-guided and process-driven forest soil stratification carbon sequestration simulation method of the present invention.

[0020] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0021] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a schematic diagram of an embodiment of a knowledge-guided and process-driven forest soil stratified carbon sequestration simulation system according to the present invention. Figure 2 A framework diagram for simulating forest soil carbon storage; Figure 3 A schematic diagram of the soil carbon cycle structure of CLM5; Figure 4 A framework diagram for simulating forest soil carbon dynamics; Figure 5This is a flowchart illustrating an embodiment of a knowledge-guided and process-driven method for simulating forest soil stratified carbon sequestration according to the present invention. Figure 6 This is a schematic diagram of an embodiment of an electronic device according to the present invention. Detailed Implementation

[0022] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0023] Figure 1 This diagram illustrates a structural schematic of an embodiment of a knowledge-guided and process-driven forest soil stratification carbon sequestration simulation system 100 provided by the present invention. Figure 1 As shown, the system 100 includes: a screening module 110, an optimization module 120, a processing module 130, an expansion module 140, a simulation module 150, an extraction module 160, and a prediction module 170.

[0024] The screening module 110 is used to: screen soil profiles with continuous and complete observations based on the stratified observation data of forest soil organic carbon in the target area.

[0025] The target area refers to the specific geographical space in which the method of this invention is applied, which can be a country, region, city, etc., without limitation. For example, the temperate coniferous and broad-leaved mixed forest distribution area of ​​region A. Forest soil organic carbon refers to the organic carbon components present in the forest soil of the target area, such as the organic carbon formed by litter, root exudates, and microbial residues in the forest soil of region A. Stratified observation data refers to soil organic carbon density data collected and measured at different depths, such as soil organic carbon density data measured at five layers in region A: 0-10 cm, 10-20 cm, 20-30 cm, 30-50 cm, and 50-100 cm. Soil profile refers to a section excavated vertically downward from the surface to a certain depth for observing and sampling the soil stratification structure, such as a soil profile excavated at a depth of 100 cm at a sampling point in region A.

[0026] The optimization module 120 is used to: determine multiple soil organic carbon observation points based on the soil profile, and optimize the parameters of the terrestrial ecosystem process model at each soil organic carbon observation point using Bayesian data assimilation technology based on external driving data, so as to obtain the optimized parameters of each soil organic carbon observation point.

[0027] In this context, "soil organic carbon observation point" refers to the geographical location where soil profile sampling and organic carbon measurement are conducted, such as the latitude and longitude coordinates of a specific point in region A. "External driving data" refers to the input data, such as climate, vegetation, and soil data, required to drive the operation of the terrestrial ecosystem process model, such as daily temperature, precipitation, solar radiation, and leaf area index data for region A. "Bayesian data assimilation technology" refers to the method of combining model simulation with observational data based on Bayesian statistical theory to optimize model parameters, such as optimizing terrestrial ecosystem process model parameters using soil organic carbon observation data at a specific observation point in region A. "Terrestrial ecosystem process model" refers to a mechanism model constructed based on the physiological and ecological characteristics of an ecosystem, used to simulate the cycling of elements such as carbon, nitrogen, and water, and the flow of energy within the ecosystem. This embodiment uses the CLM5 (Community Land Model version 5) model, also known as the terrestrial ecosystem process model, to simulate the carbon, water, and energy cycling processes on the land surface. Specifically, this embodiment uses the soil carbon cycle module of the terrestrial ecosystem process model, for example, a physical mechanism model used to simulate the dynamics of forest soil carbon in region A. Optimized parameters refer to the parameters of terrestrial ecosystem process models calibrated using Bayesian data assimilation techniques, such as the carbon decomposition rate parameters obtained after optimization at a certain observation point in region A.

[0028] It should be noted that the parameters used for data assimilation optimization in CLM5 are shown in Table 1.

[0029] Table 1: The processing module 130 is used to: perform quality control, resampling and extraction processing on the current vegetation, topography and soil attribute related data of the target area to obtain multimodal normalized data.

[0030] Among them, the current vegetation, topography, and soil attribute data refer to data such as vegetation type, elevation, slope, soil texture, and pH value of the target area at this stage, such as the current vegetation type distribution map, digital elevation model, and soil pH spatial data of region A. Multimodal normalized data refers to the collection of multi-source spatial data after quality control, resampling, and extraction processing, such as the multimodal dataset composed of normalized vegetation, topography, and soil attribute data of region A.

[0031] The extension module 140 is used to: use the multimodal normalized data as input variables of the random forest model, and use the random forest model to perform heterogeneity learning on the optimization parameters of each soil organic carbon observation point and extend them to the overall forest ecosystem scale of the target area, so as to obtain the scale extension parameters of the terrestrial ecosystem process model.

[0032] Random forest models are machine learning algorithms based on decision tree ensembles, used to establish nonlinear mappings between environmental variables and model parameters. For example, a random forest model can be used to predict the spatial distribution of parameters for terrestrial ecosystem processes in region A. Heterogeneity learning refers to the machine learning model learning the non-uniform distribution characteristics of parameters as environmental variables change from samples. For example, a random forest model can learn the differences in model parameters under different terrain and vegetation conditions in region A. The overall forest ecosystem scale refers to a macroscopic scale covering the entire spatial extent of the target area, such as the spatial extent of the entire forest ecosystem in region A. Scale-extended parameters refer to the spatialized parameters obtained by extrapolating the site optimization parameters to the entire target area using a random forest model, such as the spatial distribution map of terrestrial ecosystem process model parameters on a 1km × 1km grid in region A.

[0033] The simulation module 150 is used to: simulate the current stratified soil organic carbon storage in the target area using the scale extension parameters and based on the steady-state hypothesis and external driving data at the grid scale.

[0034] The steady-state hypothesis refers to the theory that assumes soil carbon input and output are in equilibrium, such as assuming that the forest soil carbon pool in region A is in dynamic equilibrium over a certain period of time. Grid-scale external driving data refers to driving data whose spatial resolution matches the model grid, such as temperature and precipitation data at a 1km resolution in region A. Current stratified soil organic carbon storage refers to the organic carbon storage at each soil layer in the target area at the current stage, such as the spatial distribution of current organic carbon storage in the 0-100cm soil layers in region A.

[0035] The extraction module 160 is used to extract data related to future vegetation, topography, and soil properties of the target area based on the simulation results of the CESM2 model of CMIP6.

[0036] CMIP6 refers to the Sixth Coupled Model Intercomparison Project, which provides global climate model simulation results, such as CMIP6 data used to provide future climate scenarios for region A. CESM2 refers to Community Earth System Model version 2, used to simulate global climate system changes, such as CESM2 model simulation results used to generate future climate scenarios for region A. Simulation results refer to the data output after running a climate model or process model, such as future temperature and precipitation data for region A simulated by the CESM2 model. Future vegetation, topography, and soil property related data refer to future vegetation, topography, and soil property data estimated based on climate simulation results, such as predicted vegetation type and soil moisture for region A in 2060.

[0037] The prediction module 170 is used to: generate, based on the transient hypothesis and driven by the scale expansion parameters and the relevant data of future vegetation, topography, and soil properties, and combined with the current stratified soil organic carbon storage, a knowledge-guided and process-driven simulation result of forest soil stratification carbon sink in the target area over a preset time period.

[0038] The transient hypothesis refers to theories that consider the dynamic changes of the soil carbon pool over time, such as simulating the annual dynamic process of forest soil carbon pool changes in region A. The future pre-defined time period refers to the future time range for which the prediction is implemented, such as the period from 2025 to 2060. The knowledge-guided and process-driven simulation results of forest soil stratified carbon sequestration refer to the predicted results of forest soil carbon absorption capacity in future periods, such as the predicted average carbon sequestration intensity and cumulative carbon sequestration of region A from 2025 to 2060.

[0039] The technical solution of this embodiment effectively solves the problem of large uncertainty in the current regional scale soil carbon dynamic assessment, and is more conducive to understanding the forest soil carbon cycle mechanism, thereby realizing accurate, digital and intelligent monitoring and assessment of forest soil carbon sink capacity.

[0040] In an alternative embodiment, the filtering module 110 is specifically used for: The stratified observation data of forest soil organic carbon in the target area were collected, and the quality control of the stratified observation data was carried out to select soil profiles with continuous and complete observations.

[0041] In the above-mentioned optional methods, the quality of the basic data for subsequent model optimization and prediction is further ensured by collecting and quality-controlling stratified observation data, thus providing a reliable data guarantee for improving prediction accuracy.

[0042] In an alternative embodiment, the filtering module 110 is specifically used for: Five continuous and fully observed soil profiles at depths of 0-10 cm, 10-20 cm, 20-30 cm, 30-50 cm, and 50-100 cm were selected from the stratified observation data, and the soil profiles were evenly distributed in different forest ecosystems in the target area.

[0043] Among the above-mentioned alternative methods, further selecting a continuous and complete soil profile at a specific depth and distributing it evenly in the target area helps to comprehensively reflect the soil characteristics of different forest ecosystems and enhances the representativeness and credibility of the model prediction results.

[0044] In an alternative embodiment, the optimization module 120 is specifically used for: Based on the Markov chain Monte Carlo method, multiple independent parameter chains are constructed for each soil organic carbon observation point; each parameter chain contains a test chain with a first number of iterations and a formal chain with a second number of iterations.

[0045] The Markov chain Monte Carlo method is a random sampling algorithm based on Markov chains, used for parameter estimation and uncertainty quantification. For example, it's used to sample the posterior distribution of parameters in a terrestrial ecosystem process model in region A. The parameter chain refers to the sequence used for parameter sampling in the Markov chain Monte Carlo method, such as a test chain with 10,000 iterations and a formal chain with 40,000 iterations. The first number of iterations refers to the number of iterations in the test chain, for example, 10,000 iterations. The test chain is the initial chain used to debug the sampled parameters in the Markov chain Monte Carlo method, such as the 10,000-iteration test chain used in parameter optimization at an observation point in region A. The second number of iterations refers to the number of iterations in the formal chain, for example, 40,000 iterations. The formal chain is the chain used for formally sampling parameters in the Markov chain Monte Carlo method, such as the 40,000-iteration formal chain used in parameter optimization at an observation point in region A.

[0046] A target loss function is constructed to calculate the difference between simulated soil organic carbon storage and measured soil organic carbon storage in the terrestrial ecosystem process model, and the acceptance rate of the target loss function is set to a predetermined range value.

[0047] Simulated soil organic carbon storage refers to the soil organic carbon storage value output by the terrestrial ecosystem process model, such as the 0-100cm soil organic carbon storage simulated by the terrestrial ecosystem process model at an observation point in region A. Measured soil organic carbon storage refers to the soil organic carbon storage value actually measured in the field, such as the 0-100cm soil organic carbon storage actually measured at an observation point in region A. The target loss function refers to the function used to measure the difference between simulated and observed values, such as the mean squared error function. The acceptance rate refers to the probability that a new parameter is accepted in the Markov chain Monte Carlo method, such as an acceptance rate of 0.4. The predetermined range value refers to the pre-set range of parameter values, such as an acceptance rate range of 0.3 to 0.5.

[0048] Specifically, the mean squared error function or the negative log-likelihood function is used as the basic form of the target loss function to quantify the difference between the simulated and measured values, and the acceptance rate of the target loss function in the Markov chain Monte Carlo sampling process is set to a predetermined range of 0.3 to 0.5.

[0049] Perform a data assimilation process for each soil organic carbon observation point, calculate the GR value for each soil organic carbon observation point, randomly select a parameter chain to generate the posterior probability distribution of each parameter for the corresponding soil organic carbon observation point, and delete soil organic carbon observation points with GR values ​​less than a predetermined threshold.

[0050] Here, GR value refers to the Gelman-Rubin statistic, used to evaluate the convergence of a Markov chain, such as the GR value of a parameter chain at an observation point in region A. Posterior probability distribution refers to the parameter probability distribution obtained based on Bayes' theorem, such as the posterior distribution of the carbon decomposition rate parameter at an observation point in region A. Preset threshold refers to a pre-defined critical value, such as a threshold where the GR value is less than 0.95.

[0051] Specifically, after completing the Markov chain Monte Carlo sampling, the results of all parameter chain warm-up phases are discarded, and the Gelman-Rubin statistic (GR) value of each parameter is calculated based on the remaining samples. A converged parameter chain is randomly selected to generate the posterior probability distribution of each model parameter at the soil organic carbon observation point, and soil organic carbon observation points with GR values ​​less than 0.95 and their corresponding parameter distributions are deleted from subsequent analyses.

[0052] Multiple random samplings were performed on the posterior probability distribution of the parameters, and the results of each sampling were applied to the calculation of the terrestrial ecosystem process model to obtain the average value of organic carbon storage in each soil layer as the optimization result of the corresponding soil organic carbon observation point.

[0053] The optimization result refers to the final result obtained after parameter optimization, such as the average soil organic carbon storage of each layer at a certain observation point in region A after optimization.

[0054] Specifically, a predetermined number of parameter samples are independently and randomly drawn from the posterior probability distribution of parameters at each effective soil organic carbon observation point, and each set of parameter samples is substituted into the terrestrial ecosystem process model to obtain the simulated value of organic carbon storage for each soil layer; the arithmetic mean of the organic carbon storage results of each layer obtained from all simulations is calculated, and the average value is used as the optimization result for each soil layer at the observation point.

[0055] The consistency between the optimization results and the observation data of each soil organic carbon observation point was evaluated using the coefficient of determination and root mean square error evaluation index, and effective soil organic carbon observation points were selected.

[0056] Among them, the coefficient of determination and root mean square error evaluation metrics refer to statistical indicators used to evaluate model performance, such as R². 2 And RMSE. Observational data refers to: data actually measured, such as the measured soil organic carbon density data at an observation point in region A.

[0057] Specifically, the optimization results of each soil layer at each soil organic carbon observation point are calculated, namely the coefficient of determination and root mean square error between the simulated organic carbon storage value and the measured organic carbon storage value at the corresponding depth. Soil organic carbon observation points with a coefficient of determination higher than a predetermined threshold and a root mean square error lower than a predetermined threshold are identified as effective soil organic carbon observation points with high consistency between simulation and observation, thus completing the screening.

[0058] Simulation results of multiple model parameters were derived from all valid soil organic carbon observation points and used as optimization parameters.

[0059] The simulation results refer to the data output by the model, such as the soil organic carbon density in region A simulated by the terrestrial ecosystem process model.

[0060] Specifically, for each effective soil organic carbon observation point, the central tendency feature value of the posterior probability distribution of each model parameter obtained through the Bayesian data assimilation process is extracted, and the central tendency feature value is used as the optimization parameter of the effective soil organic carbon observation point, thereby forming a set of multiple sets of model parameter optimization results corresponding to all effective observation points.

[0061] Among the above-mentioned optional methods, the Markov chain Monte Carlo method is further used to construct the parameter chain. Combined with the objective loss function and GR value screening, the optimization effect of the model parameters is ensured, the accuracy of the model in simulating soil organic carbon storage is improved, and a high-precision model foundation is provided for subsequent predictions.

[0062] In one alternative approach, the extension module is specifically used for: The environmental variables in the multimodal normalized data are normalized to obtain normalized environmental variable data, and the normalized environmental variable data is resampled to generate a spatially continuous normalized environmental variable map.

[0063] Environmental variables refer to environmental factors that affect soil carbon cycling, such as temperature, precipitation, and soil pH. Specific sources of environmental variables include: the 1km spatial resolution climate dataset for the global land surface released by WorldClim, the Soilgrid 250m soil stratification attribute dataset, and global station meteorological datasets and vegetation-related datasets for a preset time period after steady-state simulations from terrestrial ecosystem process models provided by NOAA. The specific data types of environmental variables are shown in Table 2.

[0064] Table 2: Normalized environmental variable data refers to environmental variable data that has undergone normalization processing, such as temperature data normalized to the [0,1] interval. Normalized environmental variable maps refer to spatialized normalized environmental variable data, such as normalized temperature spatial distribution maps of region A at 1km, 10km, or other resolutions.

[0065] Specifically, based on the maximum and minimum values ​​of each environmental variable within the target area, a linear transformation method is used to normalize the original data of each environmental variable to the range of 0 to 1, thereby obtaining normalized environmental variable data. Normalized environmental variable data from different sources and resolutions are uniformly resampled to a preset spatial resolution through spatial interpolation or aggregation methods, and a spatially continuous normalized environmental variable map covering the entire target area is generated.

[0066] The optimized parameters of each soil organic carbon observation point are used as the training target variable of the random forest model, and the normalized environmental variable data are used as input features to train the random forest model to learn the mapping relationship between environmental variables and model parameters. The random forest model is constructed and optimized using a cross-validation method that is repeated multiple times.

[0067] Here, the training target variable refers to the variable to be predicted in the machine learning model, such as the parameters of a terrestrial ecosystem process model to be predicted in a random forest model. The mapping relationship refers to the correspondence between independent and dependent variables, such as the nonlinear relationship between environmental variables and terrestrial ecosystem process model parameters. The cross-validation method refers to a statistical method used to evaluate the model's generalization ability, such as 10-fold cross-validation.

[0068] Specifically, the spatial location of each effective soil organic carbon observation point is matched with the normalized environmental variable data grid to construct a training sample set. Each training sample contains the normalized environmental variable data of the effective soil organic carbon observation point as input features and the corresponding optimization parameters as training target variables. The random forest algorithm is used to establish a nonlinear mapping relationship between the input features and the training target variables. The hyperparameters of the random forest model are adjusted through repeated K-fold cross-validation. The model performance is evaluated to select the random forest model with the highest prediction accuracy as the final training result.

[0069] The normalized environmental variable map is input into the trained random forest model to predict the values ​​of multiple parameters of the terrestrial ecosystem process model at each spatial grid point in the target area, so as to obtain the scale expansion parameters.

[0070] Specifically, the normalized environmental variable data (normalized environmental variable map) corresponding to each spatial grid point in the target area is input into the trained random forest model. The random forest model calculates the predicted values ​​of each parameter of the terrestrial ecosystem process model corresponding to each spatial grid point. Finally, the predicted values ​​of parameters of all spatial grid points are combined to form the scale-extended parameter spatial distribution map covering the entire target area.

[0071] Among the above-mentioned optional methods, further normalization of environmental variables and cross-validation optimization of the model have enabled the extension of parameters from observation points to regional scales. This makes the parameters of the terrestrial ecosystem process model more closely match the actual regional characteristics, improves the model's predictive ability at the regional scale, and provides strong support for the accurate assessment of forest soil carbon sinks.

[0072] In one alternative approach, such as Figure 2 As shown, the simulation module is specifically used for: The scale expansion parameters and the external driving data at the grid scale are input into the matrix equation of the terrestrial ecosystem process model, and the matrix equation is as follows: ;in, This indicates the current organic carbon storage in the stratified soil, including the organic carbon storage in each current layer. This represents the vector of allocation coefficients from plant carbon input to each soil carbon pool. This represents plant carbon input, in units of: ; This represents the carbon transfer coefficient matrix between the litter carbon pool and the mineral soil carbon pool. This represents the matrix of background degradation rates for different carbon pools. A scalar function representing the impact of environmental variables on the degradation of different carbon pools. Represents the vertical transport rate matrix of soil organic carbon; When the soil system reaches a steady state The soil organic carbon storage of each layer in the target area is calculated based on the matrix equation.

[0073] In matrix equations, such as Figure 3 As shown, it should be noted that: 1) A 35×1 vector was used to describe the distribution of plant carbon inputs to each soil carbon pool (unit: ): ;in, Both 0 and 0 are 5×1 vectors; - These represent four litter carbon pools in each soil layer: coarse woody debris, metabolizable litter, cellulose litter, and lignin litter; the three 5×1 zero matrices represent three mineral soil organic carbon pools: readily degradable organic carbon, slowly degradable organic carbon, and recalcitrant organic carbon.

[0074] 2) The dimensionless carbon transfer coefficients represent the carbon transfer coefficients between litter carbon pools and mineral soil carbon pools, including the transfer coefficients from four litter pools to three mineral soil carbon pools, as well as the transfer coefficients between mineral soil carbon pools within the same soil layer. ;in It is a 5×5 matrix, representing the donor carbon pool ( ) to receptor carbon pool ( The transfer coefficient of ).

[0075] 3) This represents the dimensionless influence of environmental variables on the degradation of different carbon pools, including temperature. Soil water Oxygen content Soil nitrogen and depth Scalar.

[0076] 4) 𝑲 represents the background degradation rate of different carbon pools (unit: ), 𝑽 represents the vertical transport of soil organic carbon between different soil layers due to biological disturbance and freeze-thaw action (unit: ).

[0077] In the above-mentioned optional methods, the scale expansion parameters and external driving data are further input into the matrix equation of the terrestrial ecosystem process model to calculate the soil organic carbon storage of each layer, thereby achieving an accurate assessment of the current soil carbon storage status and laying the foundation for further prediction of future trends.

[0078] In one alternative approach, such as Figure 4 As shown, the prediction module is specifically used for: The current stratified soil organic carbon storage is used as the initial state, the scale expansion parameter is used as the heterogeneity parameter of the terrestrial ecosystem process model at each grid point, and the future vegetation, topography, and soil attribute related data are used as the external driving data of the terrestrial ecosystem process model.

[0079] The initial state refers to the system state at the start of the model simulation, such as the soil organic carbon storage in region A in 2025 as the initial prediction value. Fixed parameters refer to parameters that remain unchanged during the model simulation, such as terrestrial ecosystem process model parameters that remain unchanged in future predictions.

[0080] Based on the transient state hypothesis SOC i+1 =ΔSOC i +SOC i The matrix equation is used to calculate the change in soil organic carbon storage ΔSOC for each year within the preset future time period in chronological order. i This change was then compared with the previous year's soil organic carbon (SOC) storage. i Add them together to get the soil organic carbon storage (SOC) for the next year. i+1 The predicted results of the stratified soil organic carbon storage for each year within the preset future time period are obtained.

[0081] Among them, SOC i SOC represents the soil organic carbon storage in year i. i+1 ΔSOC represents the soil organic carbon storage in year i+1. i Change in soil organic carbon storage in year i.

[0082] The arithmetic mean of the carbon sequestration intensity for each year within the predetermined future time period is calculated as the average carbon sequestration intensity, and the difference between the soil organic carbon storage in the last year of the predetermined future time period and the current soil organic carbon storage is calculated as the cumulative carbon sequestration amount; wherein, the soil organic carbon storage is the sum of all stratified soil organic carbon storage, and the average carbon sequestration intensity and the cumulative carbon sequestration amount together constitute the knowledge-guided and process-driven forest soil stratified carbon sequestration simulation results for the target area.

[0083] Average carbon sequestration intensity refers to the amount of soil carbon sequestration per unit area per unit time, such as the average annual carbon absorption per hectare of soil in region A from 2025 to 2060. Cumulative carbon sequestration refers to the total amount of soil carbon sequestration over a certain period of time, such as the total amount of soil carbon absorbed in region A from 2025 to 2060.

[0084] Among the above-mentioned optional methods, by calculating the changes in soil organic carbon storage year by year, we can obtain the prediction of stratified soil organic carbon storage in the future time period, as well as the average carbon sink intensity and cumulative carbon sink, which provides quantitative indicators for estimating the future carbon absorption capacity of forest soils and helps to formulate strategies for addressing climate change and ecological protection in advance.

[0085] Figure 5 This diagram illustrates an embodiment of a knowledge-guided and process-driven forest soil stratification carbon sequestration simulation method provided by the present invention. This method employs a knowledge-guided and process-driven forest soil stratification carbon sequestration simulation system as provided by the present invention. Figure 5 As shown, it includes the following steps: S1. Based on the stratified observation data of forest soil organic carbon in the target area, select soil profiles with continuous and complete observations. S2. Based on the soil profile, multiple soil organic carbon observation points are determined, and based on external driving data, Bayesian data assimilation technology is used to optimize the parameters of the terrestrial ecosystem process model at each soil organic carbon observation point to obtain the optimized parameters for each soil organic carbon observation point. S3. Perform quality control, resampling, and extraction processing on the current vegetation, topography, and soil attribute data of the target area to obtain multimodal normalized data; S4. The multimodal normalized data is used as the input variable of the random forest model. The random forest model is used to perform heterogeneity learning on the optimization parameters of each soil organic carbon observation point and extend them to the overall forest ecosystem scale of the target area to obtain the scale extension parameters of the terrestrial ecosystem process model. S5. Using the scale expansion parameters and based on the steady-state hypothesis and external driving data at the grid scale, simulate the current stratified soil organic carbon storage in the target area. S6. Based on the simulation results of the CESM2 model of CMIP6, extract the relevant data of future vegetation, topography and soil properties of the target area; S7. Based on the transient hypothesis, and driven by the scale expansion parameters and the relevant data of future vegetation, topography, and soil properties, and combined with the current stratified soil organic carbon storage, generate the knowledge-guided and process-driven forest soil stratified carbon sink simulation results for the target area in the future within a preset time period.

[0086] It should be noted that the beneficial effects of the knowledge-guided and process-driven forest soil stratified carbon sequestration simulation method provided in the above embodiments are the same as those of the knowledge-guided and process-driven forest soil stratified carbon sequestration simulation system 100 described above, and will not be repeated here. Furthermore, the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, and will not be repeated here.

[0087] The knowledge-guided and process-driven forest soil stratification carbon sink simulation system 100 of the present invention can be a computer program (including program code) running on a computer device. For example, the knowledge-guided and process-driven forest soil stratification carbon sink simulation system 100 of the present invention is an application software that can be used to execute the corresponding steps in the knowledge-guided and process-driven forest soil stratification carbon sink simulation method of the present invention.

[0088] In some embodiments, the knowledge-guided and process-driven forest soil stratification carbon sink simulation system 100 of the present invention can be implemented in a combination of hardware and software. As an example, the knowledge-guided and process-driven forest soil stratification carbon sink simulation system 100 of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the knowledge-guided and process-driven forest soil stratification carbon sink simulation method of the present invention. For example, the processor in the form of a hardware decoding processor can adopt one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0089] The modules described in the embodiments of this invention can be implemented in software or hardware. The names of the modules are not, in some cases, limiting the scope of the module itself.

[0090] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned knowledge-guided and process-driven forest soil stratification carbon sequestration simulation methods. That is, an electronic device according to an embodiment of the present invention may include, but is not limited to: a processor and a memory; the memory is used to store the computer program; the processor is used to execute the knowledge-guided and process-driven forest soil stratification carbon sequestration simulation method shown in any embodiment of the present invention by calling the computer program.

[0091] In one alternative embodiment, an electronic device is provided, such as Figure 6 As shown, Figure 6 The illustrated electronic device 4000 includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0092] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 4001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0093] Bus 4002 may include a path for transmitting information between the aforementioned components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 4002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6The bus 4002 is represented by only one thick line, but this does not mean that there is only one bus or one type of bus.

[0094] The memory 4003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0095] The memory 4003 stores application code (computer program) for executing the present invention, and its execution is controlled by the processor 4001. The processor 4001 executes the application code stored in the memory 4003 to implement the content shown in the foregoing method embodiments.

[0096] Among them, electronic devices can also be terminal devices. A terminal device can be any terminal device that can install applications and access web pages through applications, including at least one of smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, smart TVs, and smart in-vehicle devices.

[0097] It should be noted that, Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0098] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the above-mentioned knowledge-guided and process-driven forest soil stratified carbon sequestration simulation methods.

[0099] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0100] In an exemplary embodiment, a computer program product or computer program is also provided, comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the aforementioned knowledge-guided and process-driven forest soil stratification carbon sequestration simulation method.

[0101] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0102] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0103] The computer-readable storage medium provided in this invention can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a 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, apparatus, or device.

[0104] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0105] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0106] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0107] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0108] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A knowledge-based guidance and process-driven forest soil stratification carbon sink simulation system, characterized in that, include: The module includes a screening module, an optimization module, a processing module, an extension module, a simulation module, an extraction module, and a prediction module. The screening module is used to: select soil profiles with continuous and complete observations based on the stratified observation data of forest soil organic carbon in the target area; The optimization module is used to: determine multiple soil organic carbon observation points based on the soil profile, and optimize the parameters of the terrestrial ecosystem process model at each soil organic carbon observation point using Bayesian data assimilation technology based on external driving data, so as to obtain the optimized parameters of each soil organic carbon observation point. The processing module is used to: perform quality control, resampling and extraction processing on the current vegetation, topography and soil attribute related data of the target area to obtain multimodal normalized data; The extension module is used to: use the multimodal normalized data as input variables of the random forest model, and use the random forest model to perform heterogeneity learning on the optimization parameters of each soil organic carbon observation point and extend them to the overall forest ecosystem scale of the target area to obtain the scale extension parameters of the terrestrial ecosystem process model; The simulation module is used to: simulate the current stratified soil organic carbon storage in the target area using the scale extension parameters and based on the steady-state hypothesis and external driving data at the grid scale; The extraction module is used to: extract future vegetation, topography, and soil attribute data of the target area based on the simulation results of the CESM2 model of CMIP6; The prediction module is used to: generate knowledge-guided and process-driven simulation results of forest soil stratification carbon sink in the target area over a preset time period based on the transient hypothesis, driven by the scale expansion parameters and the relevant data of future vegetation, topography and soil properties, and combined with the current stratified soil organic carbon storage.

2. The knowledge-guided and process-driven forest soil stratified carbon sink modeling system of claim 1, wherein, The filtering module is specifically used for: The stratified observation data of forest soil organic carbon in the target area were collected, and the quality control of the stratified observation data was carried out to select soil profiles with continuous and complete observations.

3. The knowledge-guided and process-driven forest soil stratified carbon sink modeling system of claim 2, wherein, The filtering module is specifically used for: Five continuous and fully observed soil profiles at depths of 0-10 cm, 10-20 cm, 20-30 cm, 30-50 cm, and 50-100 cm were selected from the stratified observation data, and the soil profiles were evenly distributed in different forest ecosystems in the target area.

4. The knowledge-guided and process-driven forest soil stratified carbon sink modeling system of claim 1, wherein, The optimization module is specifically used for: Based on the Markov chain Monte Carlo method, multiple independent parameter chains are constructed for each soil organic carbon observation point; each parameter chain contains a test chain with a first number of iterations and a formal chain with a second number of iterations. A target loss function is constructed to calculate the difference between simulated soil organic carbon storage and measured soil organic carbon storage in the terrestrial ecosystem process model, and the acceptance rate of the target loss function is set to a predetermined range value. performing data assimilation process for each soil organic carbon observation point, calculating G-R value of each soil organic carbon observation point, and randomly selecting a parameter chain to generate posterior probability distribution of each parameter of corresponding soil organic carbon observation point, and deleting soil organic carbon observation point with G-R value less than a predetermined threshold value; performing multiple random sampling on the parameter posterior probability distribution and applying each sampling result to the calculation of the land ecosystem process model to obtain the average value of each soil layer organic carbon storage as the optimization result of the corresponding soil organic carbon observation point; evaluating the consistency of the optimization result of each soil organic carbon observation point with the observation data by using the coefficient of determination and the root mean square error evaluation index, and screening to obtain effective soil organic carbon observation points; deriving the simulation results of multiple model parameters from all effective soil organic carbon observation points as optimized parameters.

5. The knowledge-guided and process-driven forest soil stratified carbon sink modeling system of claim 1, wherein, The extension module is specifically used for: normalizing the environmental variables in the multi-modal normalized data to obtain normalized environmental variable data, and resampling the normalized environmental variable data to generate a spatially continuous normalized environmental variable map; using the optimized parameters of each soil organic carbon observation point as the training target variable of the random forest model, and using the normalized environmental variable data as input features to train the random forest model to learn the mapping relationship between the environmental variables and the model parameters, and using repeated cross-validation method to build and optimize the random forest model; inputting the normalized environmental variable map into the trained random forest model to predict the values of the multiple parameters of the land ecosystem process model at each spatial grid point in the target area to obtain the scale expansion parameters.

6. The knowledge-guided and process-driven forest soil stratified carbon sink modeling system of claim 1, wherein, The simulation module is specifically used for: inputting the scale expansion parameter with the external driving data of the grid scale into a matrix equation of the terrestrial ecosystem process model, the matrix equation being: ; wherein, represents a current stratified soil organic carbon storage, including a soil organic carbon storage of each layer at present, represents a distribution coefficient vector of plant carbon input to each soil carbon pool, represents plant carbon input, with a unit of: ; represents a carbon transfer coefficient matrix between a litter carbon pool and a mineral soil carbon pool, represents a background degradation rate matrix of different carbon pools, represents a scalar function of an influence of an environmental variable on degradation of different carbon pools, represents a soil organic carbon vertical transport rate matrix; When the soil system reaches steady state, the current soil organic carbon stock of each layer of the target area is calculated according to the matrix equation.

7. The knowledge-guided and process-driven forest soil stratified carbon sink modeling system of claim 6, wherein, The prediction module is specifically used for: using the current layered soil organic carbon storage as the initial state, using the scale expansion parameters as the heterogeneity parameters of the land ecosystem process model at each grid point, and using the future vegetation, terrain, and soil property related data as the external driving data of the land ecosystem process model; SOC based on the transient state hypothesis i+1 = ΔSOC i + SOC i , and the matrix equation is used to calculate the soil organic carbon storage change amount ΔSOC of each year in the future preset time period in chronological order i , and the change amount is added to the soil organic carbon storage SOC of the previous year i to obtain the soil organic carbon storage SOC of the next year i+1 , and the prediction result of the stratified soil organic carbon storage of each year in the future preset time period is obtained. calculating the arithmetic mean of the carbon sink intensity of each year in the future preset time period as the average carbon sink intensity, and calculating the difference between the soil organic carbon storage of the last year in the future preset time period and the current soil organic carbon storage as the cumulative carbon sink amount; wherein the soil organic carbon storage is the sum of all layered soil organic carbon storages, and the average carbon sink intensity and the cumulative carbon sink amount together constitute the knowledge-guided and process-driven forest soil layered carbon sink simulation result of the target area.

8. A knowledge-guided and process-driven forest soil-hierarchy carbon sink simulation method, characterized in that, The knowledge-guided and process-driven forest soil layered carbon sink simulation system according to any one of claims 1 to 7, the method comprising: screening a soil profile that is continuous in profile and complete in observation according to the layered observation data of forest soil organic carbon of a target area; determining multiple soil organic carbon observation points based on the soil profile, and optimizing the parameters of the land ecosystem process model at each soil organic carbon observation point based on external driving data by using Bayesian data assimilation technology to obtain the optimized parameters of each soil organic carbon observation point; Quality control, resampling and extraction are performed on current vegetation, terrain, soil property related data of the target region to obtain multi-modal normalized data; The multi-modal normalized data are taken as input variables of a random forest model, and the random forest model is used to perform heterogeneity learning on the optimization parameters of each soil organic carbon observation point and extend to the whole forest ecosystem scale of the target region to obtain scale expansion parameters of the land ecosystem process model; The scale expansion parameters are used to simulate the current stratified soil organic carbon storage of the target region based on the steady state hypothesis and external driving data of the grid scale; Based on the simulation results of the CESM2 model of CMIP6, future vegetation, terrain, soil property related data of the target region are extracted; Based on the transient state hypothesis, the scale expansion parameters and the future vegetation, terrain, soil property related data are taken as driving data, and the current stratified soil organic carbon storage is combined to generate knowledge guided and process driven forest soil stratified carbon sink simulation results of the target region in a future preset time period.

9. An electronic device, comprising: The electronic device includes a processor coupled with a memory, and the memory stores at least one computer program, which is loaded and executed by the processor to enable the electronic device to implement the knowledge guided and process driven forest soil stratified carbon sink simulation method of claim 8.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores at least one computer program, which is loaded and executed by the processor to enable the computer readable storage medium to implement the knowledge guided and process driven forest soil stratified carbon sink simulation method of claim 8.

Images