Moso bamboo forest carbon cycle simulation method and device based on data assimilation

Abstract

The invention discloses a moso bamboo forest carbon cycle simulation method and device based on data assimilation. The moso bamboo forest carbon cycle simulation method comprises the following steps: firstly, acquiring elevation data, soil data, moso bamboo forest abundance data, moso bamboo forest physiological and ecological parameters and meteorological data; then qualitatively screening and quantitatively analyzing sensitive parameters of the BIOME-BGC model; coupling flux observation data and the BIOME-BGC model by using a particle filter algorithm, and constructing an assimilation model to obtain an optimal parameter set for moso bamboo forest carbon flux simulation; finally, the obtained data is used for driving the assimilated BIOME-BGC model, and the regional moso bamboo forest carbon sink space-time pattern is optimized and simulated. According to the method, the moso bamboo forest carbon flux simulation precision is remarkably improved, and the uncertainty of model simulation is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of carbon cycle simulation, and in particular to a method and device for simulating the carbon cycle of a bamboo forest based on data assimilation. Background Art

[0002] Net ecosystem productivity (NEP), as an important indicator to measure the carbon source and sink function of terrestrial ecosystems, is the difference between the net primary productivity of vegetation and heterotrophic respiration, and can quantitatively describe the net carbon exchange between terrestrial ecosystems and the atmosphere. Bamboo is known as the "second largest forest in the world", and its carbon storage accounts for about 0.94% of the global forest ecosystem carbon storage; China is known as the "World Bamboo Kingdom" and has abundant bamboo resources. Among them, moso bamboo accounts for about 70% of my country's bamboo forest resource area. It has the characteristics of rapid growth and great carbon fixation potential, and plays an important role in actively responding to global climate change and regional ecological health. Therefore, accurate monitoring of the carbon sink capacity of moso bamboo forests will help to carry out more comprehensive and detailed research on climate change and carbon emission management.

[0003] As a typical ecological process model, the BIOME-BGC model can simulate the photosynthesis, respiration, decomposition and other physiological and ecological processes of the ecosystem at different scales. It has strong mechanism, high comprehensiveness and good extension. It has been applied to forest carbon sink estimation at different scales and has achieved good estimation results. However, the model adopts the phenological state of constant growth and fall for evergreen forests, and does not consider the effects of bamboo forest alternate years and human management interference on the carbon cycle. It is not suitable for the spatiotemporal simulation of the carbon cycle of bamboo forest ecosystems. Therefore, we need to use the BIOME-BGC model, which is specially improved for the unique physiological and ecological structure of bamboo forests, to simulate the carbon cycle of bamboo forests. At the same time, the reasonable estimation of model parameters is the key to ensuring the accuracy of model simulation and an important source of model uncertainty.

[0004] The flux site observation data has good temporal resolution and observation accuracy, but this method has high requirements on the accuracy and stability of the instrument, and is complex to operate and maintain. Moreover, the method is still a small-scale observation method, and the results obtained at one site are difficult to extrapolate to other sites. The data assimilation method provides a new idea for integrating multi-source spatial information. This method fully considers the spatial and temporal distribution characteristics of the data, model uncertainty, and observation errors. During the model operation, the method continuously integrates observation information, optimizes model parameters, and adjusts the model operation trajectory, thereby reducing model errors and observation errors, and significantly improving the simulation accuracy of the model.

[0005] Model data fusion can evaluate and reduce the uncertainty of the model and its parameters, obtain the best estimate of the parameters, and significantly enhance the simulation accuracy of the carbon and water flux of the model. In particular, with the continuous strengthening and improvement of regional and global ecological flux observation networks, the acquisition of long-term series and multi-observation index flux data is becoming increasingly convenient, which will undoubtedly promote the continuous deepening and expansion of related research fields and provide an effective solution to effectively deal with the uncertainty problems that are common in the simulation and prediction process of ecological models.

[0006] At present, there is only one most similar prior art invention, which is "A method for inverting surface carbon flux based on satellite CO2 column concentration observation" (application number: 202010554099.5, application publication number: CN111723482A). This invention builds a carbon flux inversion system based on the ensemble Kalman filter (EnKF) data assimilation method and couples it to the MOZART-4 global atmospheric chemical transport model to correct carbon flux errors and simulation accuracy.

[0007] The EnKF algorithm is an upgrade of the Kalman filter algorithm. It is a data assimilation algorithm based on the Gaussian distribution assumption, but it only considers the first-order and second-order moments of the probability density distribution. This approximation will lead to information loss. In addition, the Kalman filter series of algorithms can only obtain the optimal solution when dealing with linear and Gaussian problems, while most ecological process models are relatively complex and nonlinear models. Particle filtering is a sequential data assimilation method developed based on the importance sampling filter idea of ​​Bayesian sampling estimation. Its core idea is to use a group of weighted random sample particles in the state space to approximate the posterior probability distribution of the state. Compared with the Kalman filter series of algorithms, it is not constrained by the model state variables and the Gaussian distribution assumption of the error, and is applicable to any nonlinear non-Gaussian dynamic system. At the same time, it uses the Monte Carlo sampling method to approximate the entire posterior probability density distribution of the state variables, which can better represent the change information of the nonlinear system, which is very important for complex land surface process models.

[0008] The shortcomings of the existing technology are: (1) When using the BIOME-BGC model to simulate the carbon cycle, the bamboo forest as a special forest is not treated separately, resulting in great uncertainty in the estimation results of the carbon flux of the bamboo forest; (2) It is impossible to quickly identify sensitive parameters related to the bamboo forest; (3) Due to model defects, external multi-source observation information is not incorporated to reduce the uncertainty of the bamboo forest carbon cycle simulation. Summary of the invention

[0009] The present invention aims to address the deficiencies of the prior art and to propose a method and device for simulating the carbon cycle of a bamboo forest based on data assimilation.

[0010] The objective of the present invention is achieved through the following technical scheme: a method for simulating the carbon cycle of a bamboo forest based on data assimilation, the method comprising the following steps:

[0011] (1) Obtain elevation data, soil data, bamboo forest abundance data, bamboo forest physiological and ecological parameters, meteorological data, and eddy covariance carbon flux observation data;

[0012] (2) Qualitative screening and quantitative analysis of sensitive parameters of the BIOME-BGC model;

[0013] (3) Use the particle filter algorithm to couple flux observation data and the BIOME-BGC model to build an assimilation model;

[0014] (4) Use the data obtained in step (1) to drive the assimilated BIOME-BGC model to optimize and simulate the spatiotemporal pattern of carbon sequestration in regional bamboo forests.

[0015] Furthermore, the bamboo forest abundance data is firstly obtained by using the maximum likelihood supervised classification to extract forest information, and then the fully constrained least squares mixed pixel decomposition is used to solve the bamboo forest abundance.

[0016] Furthermore, the Morris method was used to qualitatively screen the sensitive parameters of the BIOME-BGC model, and the EFAST method was used to quantitatively analyze the sensitive parameters of the BIOME-BGC model.

[0017] Furthermore, the weighted particles are introduced into the BIOME-BGC model, and according to the model state time evolution equation, the state variable set at the next moment is obtained as the initial state of the next cycle. At the same time, the particle weights are updated based on the predicted optimal parameter set of the NEP simulation.

[0018] Furthermore, the model synchronization process is as follows:

[0019] Step 1. Assume that the state variable is x, k is time, and initially k = 0. Sample in the distribution function of the state variable, obtain N weighted particles, and initialize the particle weights.

[0020] Step 2. Bring in the BIOME-BGC model and obtain the state variable set at the next moment according to the model state time evolution equation, and also obtain the predicted NEP set;

[0021] Step 3. Determine whether there is an observed NEP at the next moment, set it as Z k+1 , if it exists, calculate the likelihood function Update particle weights and normalize them, is the state variable of the ith particle at time k+1; otherwise, the set of state variables obtained at that time is used as the initial state of the next cycle;

[0022] Step 4. Determine whether the particles are degraded. When the number of valid particles is less than the set threshold, it indicates that the particles are seriously degraded, and residual resampling is performed to obtain new N weighted particles. Otherwise, go directly to step 5.

[0023] Step 5. Calculate the state estimate of the state variable;

[0024] 6. k = k + 1, bring the best model state variables back into step 2 to start the calculation until no more external observation data is added, then the assimilation is completed and the assimilation result can be obtained.

[0025] Furthermore, if the number of valid particles is less than the set threshold during the cycle, the residual resampling method is used for resampling.

[0026] In a second aspect, the present invention also provides a bamboo forest carbon cycle simulation device based on data assimilation, comprising a memory and one or more processors, wherein the memory stores executable code, and when the processor executes the executable code, the method for simulating the carbon cycle of a bamboo forest based on data assimilation is implemented.

[0027] In a third aspect, the present invention further provides a computer-readable storage medium having a program stored thereon, and when the program is executed by a processor, the method for simulating the carbon cycle of a bamboo forest based on data assimilation is implemented.

[0028] In a fourth aspect, the present invention further provides a computer program product, including a computer program, which, when executed by a processor, implements the method for simulating the carbon cycle of a bamboo forest based on data assimilation.

[0029] Beneficial effects of the present invention:

[0030] (1) The present invention uses Morris method to qualitatively screen sensitive parameters, uses EFAST method to quantitatively analyze sensitive parameters, and combines two global sensitivity analysis methods to comprehensively screen out highly sensitive parameters for carbon flux simulation in bamboo forests;

[0031] (2) The present invention uses the residual resampling particle filter algorithm, couples the flux observation data and the BIOME-BGC model, and constructs an assimilation model to obtain the optimal parameter set for the simulation of the carbon flux of the bamboo forest, which significantly improves the simulation accuracy of the carbon flux of the bamboo forest and reduces the uncertainty of the model simulation;

[0032] (3) The present invention can select observation information from any source (such as remote sensing observations, carbon and water flux observation stations, sample plot surveys, etc.), and combine it with bamboo forest abundance data, meteorological, soil, topographic, physiological and ecological parameter data to drive the assimilated BIOME-BGC model, optimize and simulate the spatiotemporal pattern of regional bamboo forest carbon flux, thereby improving the large-scale carbon flux simulation accuracy of bamboo forests. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0034] Figure 1 It is a schematic diagram of a process flow of a bamboo forest carbon cycle simulation method based on data assimilation provided by the present invention;

[0035] Figure 2 It is a structural diagram of a bamboo forest carbon cycle simulation device based on data assimilation provided by the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with the accompanying drawings and implementation examples. It should be understood that the specific implementation examples described herein are only used to explain the present invention and are not used to limit the present invention.

[0037] like Figure 1 As shown, the present invention provides a bamboo forest carbon cycle simulation method based on data assimilation. The present invention integrates two global sensitivity analysis methods, EFAST method and Morris method, to obtain the key physiological and ecological parameters of carbon assimilation of the improved BIOME-BGC model, innovates the coupled particle filter algorithm and the improved BIOME-BGC model, and constructs a bamboo forest carbon cycle assimilation system. Based on long-term carbon flux observation data, the spatiotemporal pattern of carbon flux in the bamboo forest is optimized and simulated, and the process is fully programmed to provide scientific support for accurately evaluating the carbon sink capacity of the bamboo forest. The method of the present invention comprises the following steps:

[0038] (1) BIOME-BGC model-driven data and flux observation data acquisition:

[0039] The input data driving the BIOME-BGC model include initialization files (elevation data, soil data, bamboo forest abundance data, etc.), meteorological data, physiological and ecological parameters, etc. The elevation data (DEM) comes from The ASTER Global Digital Elevation Model (ASTER GDEM) Version 3 (V3). The soil texture data comes from the Harmonized World Soil Database (HWSD1.2). The bamboo forest abundance data first uses the maximum likelihood supervised classification method to extract forest information, and then uses the fully constrained least squares method to decompose the mixed pixels to solve the bamboo forest abundance. The baseline values ​​of the bamboo forest physiological and ecological parameters are obtained through field measurements and a large number of literature summaries.

[0040] Meteorological data are from flux observation stations. Meteorological elements include daily maximum temperature (℃), daily minimum temperature (℃), daily precipitation (mm), daily relative humidity (%) and total solar radiation (WJ·m -2 The present invention uses the inverse distance weighted interpolation method with a spatial resolution of 1000m to interpolate the data of 824 meteorological stations, wherein the temperature data is corrected for altitude using digital elevation data, assuming that the decreasing rate of temperature with increasing altitude is 6.5℃·km -1 , and use the sunshine time measurement of each station to simulate solar radiation. The monthly and annual meteorological data are obtained by taking the average and summing the interpolated daily meteorological data.

[0041] Carbon flux observation data can be obtained from large flux observation networks such as ChinaFlux and Fluxnet, or from self-built flux observation stations. However, flux observation stations may have equipment failures that lead to data loss. In order to ensure the accuracy and continuity of observation data, data from complete years must be selected. Self-built flux stations must follow the standard processing procedures of ChinaFLUX for data processing, quality control, and data interpolation to ensure the accuracy of NEP assimilation.

[0042] (2) BIOME-BGC model sensitivity analysis:

[0043] The present invention utilizes Morris method to qualitatively screen sensitive parameters, utilizes EFAST method to quantitatively analyze sensitive parameters, and combines two global sensitivity analysis methods to comprehensively screen out highly sensitive parameters for carbon cycle simulation of bamboo forests.

[0044] Morris method can effectively screen out sensitive parameters when the sample size is small. It is very applicable to models with many analysis parameters and large computational load. This method calculates the sensitivity of each parameter by differential method and is currently a widely used qualitative global sensitivity analysis method. The Morris method is briefly introduced as follows:

[0045] First, assume that the system model is Y=f(xi,xi,...,xk), where k is the dimension of the model parameters. The Morris method first maps the value range of each parameter to the interval [0,1] and divides it into p levels, that is, a k-dimensional p-level sample space. The element effect of the i-th parameter is defined as:

[0046]

[0047] Wherein: Δ is a value between 1 / p-1 and 1-1 / (p-1).

[0048] Using this method to calculate the element effect of each variable one by one requires a total of k+1 model operations, but a single calculation cannot show the overall situation of the sample space, so it is usually necessary to generate different samples to calculate S multiple times. i And find its average value μ i and standard deviation σ i , the formula is as follows:

[0049]

[0050] Where: r is the number of repetitions.

[0051] In order to eliminate the positive and negative μ caused by the non-monotonic function relationship i When the values ​​cancel each other out, we usually use For evaluation, see the following formula:

[0052]

[0053] Can reflect parameter x i The sensitivity of The larger the value is, the greater the influence of this parameter on the model output results; i Reflect parameter x i Interaction with other parameters.

[0054] The EFAST method is a quantitative global sensitivity analysis method that combines the Sobol method that can calculate the interaction between parameters and the FAST method's high-efficiency sampling improvement. It can not only analyze the impact of changes in each parameter on the model output results, but also analyze the impact of interactions between multiple parameters on the model output results. It is widely used in many nonlinear models. The EFAST method is briefly introduced as follows:

[0055] Assume the model is Y = f(x) = f(x i,j,k,... ), the input parameters are (x i,j,k,...). The sensitivity is calculated by estimating the variance contribution of each input parameter change to the output result Y. The total variance of the model output result V Y Can be decomposed into:

[0056]

[0057] Where V Y is the total variance of the model result Y; V i is the parameter x i Variance of ij ~V ij...k is the variance of the interaction of each parameter.

[0058] V i =V[E(Y / x i )] (6)

[0059] In the formula, E(Y / x i ) is Y versus x i The conditional expectation of V[E(Y / x i )] is Y versus x i The variance of the conditional expectation of .

[0060] V ij =V[E(Y / x i ,x j )]-V i -V j (7)

[0061] In the formula, E(Y / x i ,x j ) is Y versus x i 、x j The conditional expectation of V[E(Y / x i ,x j )] is Y versus x i 、x j The variance of the conditional expectation of .

[0062] The variance of the conditional expectation is called the main effect, which reflects the parameter x i The significance of the variance of the model result Y. Therefore, the sensitivity index can be obtained by calculating the ratio of the variance of each parameter and the interaction of parameters to the total variance. The first-order sensitivity index S i Can reflect the parameter x i The direct contribution rate to the total variance of the model results can be defined as:

[0063]

[0064] Overall sensitivity index S iT is the sum of the sensitivity indexes of each order, reflecting the parameter x iDirect contribution rate and indirect contribution rate to the total variance of model output through interactive coupling between parameters:

[0065] S iT =S i +S ij +······+S ij······k (9)

[0066] Parameter x i The larger the sensitivity index is, the greater the contribution of this parameter to the model result Y directly and indirectly through the interaction and coupling between parameters, and the more it can explain the variance of the model result.

[0067] (3) Constructing a carbon cycle model incorporating a particle filter algorithm

[0068] The present invention utilizes the residual resampling particle filter algorithm, couples flux observation data and the BIOME-BGC model, and constructs an assimilation model to obtain an optimal parameter set for simulating the carbon cycle of bamboo forests.

[0069] The particle filter can effectively estimate the system state without knowing the exact nature of the error distribution, so it can theoretically be applied to any system that can be described by a state space model, regardless of whether its dynamics are linear or whether the noise distribution is Gaussian. The particle filter algorithm is briefly introduced as follows:

[0070] Assuming that N weighted particles are obtained independently from the state posterior probability distribution, the posterior probability distribution of the state can be approximated by the following formula:

[0071]

[0072] Where δ is the Dirac function, p(X k |Y 1:k ) is the posterior probability distribution, k is the time, is the particle weight, Y k is the observed value. Since p(X k |Y 1:k ) is usually non-Gaussian, and it is difficult to directly sample from the posterior probability distribution function. In order to solve this problem, the proposed distribution function is introduced and the weight is calculated recursively. The formula is as follows:

[0073]

[0074] In the formula, is the proposed distribution function, is the likelihood function. From formulas (11) and (12), we get:

[0075]

[0076] After normalizing the particle weights, we get:

[0077]

[0078] The estimated value of the final state variable is the weighted average of all particle state values:

[0079]

[0080] For particle filtering, its main drawback is the particle degradation problem, that is, after several iterations of the algorithm, only a few particles have large weights, and most particles have very small weights. Currently, Neff is often used to judge the degree of particle degradation. When Neff is less than a certain threshold, it indicates that the particle degradation is more serious, as shown in the following formula:

[0081]

[0082] Where Neff is the estimated value of the effective number of particles, is the normalized weight.

[0083] The resampling algorithm is mainly used to solve the problem of particle degradation. Compared with other resampling methods, the residual resampling method has low computational overhead and small sampling variance. Therefore, the present invention uses the residual resampling method to resample severely degraded particles. The main principles are:

[0084]

[0085] In the formula, k i is the resampling coefficient, |·| is the integer part of the corresponding value, is the normalized residual weight.

[0086] The present invention sets the state variables as the sensitive parameters selected in the sensitivity analysis to participate in the assimilation experiment. The assimilation scheme based on the residual resampling particle filter algorithm is briefly introduced as follows:

[0087] 1. Assume that the state variable is x, k is time, and initially k = 0. Sample the distribution function of the state variable to obtain N (N = 100) weighted particles, that is,

[0088] 2. Bring in the BIOME-BGC model and obtain the state variable set at the next moment according to the model state time evolution equation, that is, At the same time, the predicted NEP set is obtained.

[0089] 3. Determine whether there is an observed NEP at this time, set it to Z k+1 , if it exists, calculate the likelihood function Update particle weights And normalized Otherwise, the set of state variables obtained at that moment is used as the initial state of the next cycle.

[0090] 4. Determine whether the particles are degraded. When the number of valid particles is less than 20, it indicates that the particles are seriously degraded. Then perform residual resampling to obtain new N weighted particles. Otherwise, go directly to step 5.

[0091] 5. Calculate the state estimate of the state variable

[0092] 6. k = k + 1, bring the best model state variables back into step 2 to start the calculation until no more external observation data is added, then the assimilation is completed and the assimilation result can be obtained.

[0093] (4) Estimation of carbon flux in bamboo forests:

[0094] After the assimilation model was constructed, it was combined with data on bamboo forest abundance, meteorology, soil, topography, physiological and ecological parameters, etc. to drive the assimilated BIOME-BGC model, which significantly improved the simulation accuracy of bamboo forest carbon flux and optimized and simulated the spatiotemporal pattern of regional bamboo forest carbon sinks.

[0095] Corresponding to the aforementioned embodiment of a method for simulating carbon cycle in a bamboo forest based on data assimilation, the present invention also provides an embodiment of a device for simulating carbon cycle in a bamboo forest based on data assimilation.

[0096] See also Figure 2 A bamboo forest carbon cycle simulation device based on data assimilation is provided in an embodiment of the present invention, comprising a memory and one or more processors, wherein the memory stores executable code, and when the processor executes the executable code, it is used to implement a bamboo forest carbon cycle simulation method based on data assimilation in the above embodiment.

[0097] The embodiment of the bamboo forest carbon cycle simulation device based on data assimilation provided by the present invention can be applied to any device with data processing capability, and the device with data processing capability can be a device or apparatus such as a computer. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of any device with data processing capability in which it is located reading the corresponding computer program instructions in the non-volatile memory into the internal memory for execution. From the hardware level, if Figure 2 As shown, it is a hardware structure diagram of a bamboo forest carbon cycle simulation device based on data assimilation provided by the present invention, in which any device with data processing capability is located, except Figure 2In addition to the processor, memory, network interface, and non-volatile memory shown, any device with data processing capabilities in which the apparatus in the embodiments is located may also include other hardware, generally based on the actual functions of the device with data processing capabilities, which will not be described in detail.

[0098] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0099] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of the present invention. Ordinary technicians in this field can understand and implement it without paying creative work.

[0100] An embodiment of the present invention further provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, a method for simulating the carbon cycle of a bamboo forest based on data assimilation in the above embodiment is implemented.

[0101] The computer-readable storage medium may be an internal storage unit of any device with data processing capability described in any of the aforementioned embodiments, such as a hard disk or a memory. The computer-readable storage medium may also be an external storage device of any device with data processing capability, such as a plug-in hard disk, a smart media card (SMC), an SD card, a flash card, etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit and an external storage device of any device with data processing capability. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capability, and may also be used to temporarily store data that has been output or is to be output.

[0102] The present invention also provides a computer program product, including a computer program, and when the computer program is executed by a processor, the method for simulating the carbon cycle of a bamboo forest based on data assimilation is implemented.

[0103] The above embodiments are used to illustrate the present invention rather than to limit the present invention. Any modification and change made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for simulating the carbon cycle of bamboo forests based on data assimilation, characterized in that: The method comprises the following steps: (1) Obtain elevation data, soil data, bamboo forest abundance data, bamboo forest physiological and ecological parameters, meteorological data, and eddy covariance carbon flux observation data; (2) Qualitative screening and quantitative analysis of sensitive parameters of the BIOME-BGC model; (3) Use the particle filter algorithm to couple flux observation data and the BIOME-BGC model to build an assimilation model; (4) Use the data obtained in step (1) to drive the assimilated BIOME-BGC model to optimize and simulate the spatiotemporal pattern of carbon sequestration in regional bamboo forests.

2. The method for simulating the carbon cycle of a bamboo forest based on data assimilation according to claim 1, characterized in that: The bamboo forest abundance data is first obtained by extracting forest information through maximum likelihood supervised classification, and then the bamboo forest abundance is solved by fully constrained least squares mixed pixel decomposition.

3. The method for simulating the carbon cycle of a bamboo forest based on data assimilation according to claim 1, characterized in that: Morris method was used to qualitatively screen the sensitive parameters of BIOME-BGC model, and EFAST method was used to quantitatively analyze the sensitive parameters of BIOME-BGC model.

4. The method for simulating the carbon cycle of a bamboo forest based on data assimilation according to claim 1, characterized in that: The weighted particles are introduced into the BIOME-BGC model, and according to the model state time evolution equation, the state variable set at the next moment is obtained as the initial state of the next cycle. At the same time, the particle weights are updated based on the predicted optimal parameter set of the NEP simulation.

5. The method for simulating the carbon cycle of bamboo forests based on data assimilation according to claim 4, characterized in that: The model synchronization process is as follows: Step 1. Assume that the state variable is x, k is time, and initially k = 0. Sample in the distribution function of the state variable, obtain N weighted particles, and initialize the particle weights. Step 2. Bring in the BIOME-BGC model and obtain the state variable set at the next moment according to the model state time evolution equation, and also obtain the predicted NEP set; Step 3. Determine whether there is an observed NEP at the next moment, set it as Z k+1 , if it exists, calculate the likelihood function Update particle weights and normalize them, is the state variable of the ith particle at time k+1; otherwise, the set of state variables obtained at that time is used as the initial state of the next cycle; Step 4. Determine whether the particles are degraded. When the number of valid particles is less than the set threshold, it indicates that the particles are seriously degraded, and residual resampling is performed to obtain new N weighted particles. Otherwise, go directly to step 5. Step 5. Calculate the state estimate of the state variable; 6. k = k + 1, bring the best model state variables back into step 2 to start the calculation until no more external observation data is added, then the assimilation is completed and the assimilation result can be obtained.

6. The method for simulating the carbon cycle of bamboo forests based on data assimilation according to claim 5, characterized in that: If the number of valid particles is less than the set threshold during the cycle, the residual resampling method is used for resampling.

7. A bamboo forest carbon cycle simulation device based on data assimilation, comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, a method for simulating the carbon cycle of a bamboo forest based on data assimilation according to any one of claims 1 to 6 is implemented.

8. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, a method for simulating the carbon cycle of a bamboo forest based on data assimilation as described in any one of claims 1 to 6 is implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for simulating the carbon cycle of a bamboo forest based on data assimilation as described in any one of claims 1 to 6 is implemented.

Images