Sunlight-induced chlorophyll fluorescence reconstruction method and system based on satellite observation data

Abstract

The invention provides a sunlight-induced chlorophyll fluorescence reconstruction method and system based on satellite observation data. Comprising the following steps: collecting and processing ground site observation data, air temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir correction reflectivity data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, and OCO-2 and TROPOMI satellite observation SIF data; performing data reconstruction by using the OCO-2 SIF and the TROPOMI SIF as original data respectively by using a machine learning method; then, an empirical formula between the two kinds of reconstructed SIF data and the site observation SIF is constructed, and estimation parameters on the site are obtained through optimization; then, calculating estimation parameters of the empirical formula on each land coverage type; and finally, substituting the two types of reconstructed SIF data into an empirical formula to obtain a reconstructed SIF with site SIF information.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing data reconstruction, and in particular to a method and system for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data. Background Art

[0002] Solar-induced Chlorophyll Fluorescence (SIF) is one of the by-products of plant photosynthesis. It is directly related to the physiological state of plants and can be observed through remote sensing. It has been widely used in the field of ecology in recent years.

[0003] At present, sunlight-induced chlorophyll fluorescence data mainly comes from satellite remote sensing observations, but the existing satellite remote sensing observation sunlight-induced chlorophyll fluorescence data has the problem of low spatial coverage. Therefore, it is necessary to develop a spatial reconstruction technology for remote sensing data, which aims to use the information provided by other remote sensing and meteorological data related to the target data, establish the relationship between them and the target data, and regenerate the target data with higher spatial coverage.

[0004] In recent years, there have been a variety of methods for spatial reconstruction based on satellite remote sensing observations of sunlight-induced chlorophyll fluorescence data. The research methods used for SIF reconstruction can be roughly divided into two categories, one is based on the physical model of light energy utilization, and the other is a data-driven statistical method, mostly achieved by building a machine learning model. With the rapid development of machine learning algorithms, the existing SIF reconstruction methods based on machine learning can achieve higher accuracy and better consistency with the original satellite remote sensing observation data, and are widely used in the field of SIF data reconstruction.

[0005] Observation of SIF at the site is a recent trend. While global-scale measurements from satellites using high-resolution spectroscopy in the Fraunhofer band are becoming increasingly available, observations using these techniques at the mesocanopy scale are rare. Datasets with a large number of SIF ground-based observation sites include ChinaSpec and FLUXNET2015.

[0006] At present, there is still controversy in the effectiveness of reconstructed SIF data. Existing studies generally have the following problems: (1) SIF reconstructed data and ground station observations are poorly matched; (2) SIF remote sensing observation data from a single source are often used in the reconstruction, and the amount of information provided by the original data is slightly insufficient. In general, existing studies fail to fully reflect the actual SIF information on the surface. Summary of the invention

[0007] The present invention provides a method and system for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data, which is used to solve the defect that the reconstruction technology for SIF in the prior art fails to fully reflect the real surface conditions.

[0008] In a first aspect, the present invention provides a method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data, comprising: Collect and preprocess ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectance data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data; Acquire chlorophyll fluorescence SIF observation data of ground stations, and perform time averaging and correction on the chlorophyll fluorescence SIF observation data of the ground stations; Using the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data as original data, a machine learning algorithm is used to reconstruct the data to obtain reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data; Constructing an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data, calibrating the parameters of the empirical formula, and obtaining an estimated parameter set of the empirical formula at each ground station by using a differential evolution algorithm for each ground station; Classifying the ground stations based on global land cover classification data to obtain updated station data, and updating the empirical formula using the station data and the estimated parameter set; The reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data are substituted into the updated empirical formula to obtain the cross-platform fused observation SIF data.

[0009] According to a method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention, ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectance data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data are collected and preprocessed, including: Extract the temperature variables with a preset time resolution of days from the ERA5-Land meteorological reanalysis data at the hourly scale, and unify the spatial resolution of the temperature variables to the preset degree through spatial interpolation; Filtering the cloud-free data and the data with a cloud percentage less than a preset percentage in the OCO-2 satellite observation SIF data; Processing the leaf area index data, the MODIS bidirectional reflectance distribution function nadir-corrected reflectivity data, the enhanced vegetation index data, the normalized difference moisture index data, the OCO-2 satellite observation SIF data, and the TROPOMI satellite observation SIF data according to the preset day time resolution and the preset degree temporal and spatial resolution; The instantaneous values ​​of the TROPOMI satellite observed SIF data are averaged to obtain the daily average value, so that the TROPOMI satellite observed SIF data is consistent with the OCO-2 satellite observed SIF data: ; Where SIF stands for transient chlorophyll fluorescence, represents the daily average chlorophyll fluorescence, is the measurement time of instantaneous SIF, For measurement The solar zenith angle at is the time step, here we take 10 minutes for numerical calculation, H is the heaviside step function, For hours.

[0010] According to a method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention, chlorophyll fluorescence SIF observation data of ground stations are obtained, and time averaging and correction are performed on the chlorophyll fluorescence SIF observation data of the ground stations, including: Performing time averaging on the chlorophyll fluorescence SIF observation data of the ground station at the preset time resolution of days; According to the approximately linear variation trend of the SIF spectral characteristic curve within a specified range, the ground station chlorophyll fluorescence SIF observation data are uniformly corrected to a wavelength consistent with the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data.

[0011] According to a method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention, the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data are used as original data, and a machine learning algorithm is used to reconstruct the data to obtain reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data, including: The XGBoost machine learning method is adopted to optimize the model hyperparameters by random search, and the ERA5-Land meteorological reanalysis temperature data, the leaf area index data, the red, blue, green and near-infrared band data in the MODIS bidirectional reflectance distribution function nadir-corrected reflectivity data, the enhanced vegetation index data and the normalized difference moisture index data are used as independent variables, and the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data are used as target variables for training, verification and testing, respectively, to generate machine learning models corresponding to the respective target variables; The independent variables are respectively input into the machine learning models corresponding to the respective target variables to obtain the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data.

[0012] According to a method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention, constructing an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data, and the reconstructed TROPOMI satellite observation SIF data includes: Constructing cross-platform fusion observation SIF data to be solved Reconstructed SIF data from OCO-2 satellite observations , the reconstructed TROPOMI satellite observation SIF data The empirical formula between:

[0013] in, Parameters to be determined.

[0014] According to a sunlight-induced chlorophyll fluorescence reconstruction method based on satellite observation data provided by the present invention, the empirical formula is calibrated, and a set of estimated parameters of the empirical formula at each ground station is obtained by using a differential evolution algorithm for each ground station, including: Adjust the empirical formula to a linear form:

[0015] Differential evolution algorithm is used to analyze the and , Perform parameter calibration and obtain the parameters A collection of estimated parameters at each ground station.

[0016] According to a method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention, the above-mentioned ground stations are classified based on global land cover classification data to obtain updated station data, including: Classifying the ground stations according to the global land cover classification data; The updated site data are obtained by performing arithmetic averaging on the site parameters belonging to the same land cover type.

[0017] According to a method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention, the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data are substituted into an updated empirical formula to obtain cross-platform fused observation SIF data, including: The reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data Substitute into the empirical formula Perform calculations to obtain cross-platform fusion observation SIF data .

[0018] In a second aspect, the present invention further provides a sunlight-induced chlorophyll fluorescence reconstruction system based on satellite observation data, comprising: The acquisition module is used to collect and preprocess ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectance data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data; An acquisition module is used to acquire chlorophyll fluorescence SIF observation data of ground stations, and perform time averaging and correction on the chlorophyll fluorescence SIF observation data of the ground stations; A reconstruction module is used to use the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data as original data and adopt a machine learning algorithm to reconstruct the data to obtain reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data; A calibration module is used to construct an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data, calibrate the parameters of the empirical formula, and obtain an estimated parameter set of the empirical formula at each ground station by using a differential evolution algorithm for each ground station; A classification module, used for classifying the ground stations based on global land cover classification data to obtain updated station data, and updating the empirical formula using the station data and the estimated parameter set; The processing module is used to substitute the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data into the updated empirical formula to obtain the cross-platform fusion observation SIF data.

[0019] In a third aspect, the present invention also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a method for reconstructing daylight-induced chlorophyll fluorescence based on satellite observation data as described above is implemented.

[0020] The method and system for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention can better reflect the actual SIF information of the surface by making full use of ground site observation data that is more accurate than satellite remote sensing observations, and calibrating the obtained reconstructed SIF data. The SIF information of two different satellite remote sensing observation platforms, OCO-2 and TROPOMI, is taken into account, and is integrated with the site observation SIF as the target, so as to generate cross-platform SIF reconstruction data with richer information, which is scientific, reasonable and close to engineering practice. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present invention or 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 some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 This is one of the flow charts of the method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention; Figure 2 This is the second flow chart of the method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by the present invention; Figure 3 is a schematic diagram of the probability density function of the global reconstructed SIF mean provided by the present invention; Figure 4 It is a structural schematic diagram of a sunlight-induced chlorophyll fluorescence reconstruction system based on satellite observation data provided by the present invention; Figure 5 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] In view of the defects of the prior art, the present invention proposes a global sunlight-induced chlorophyll fluorescence reconstruction method based on OCO-2 and TROPOMI satellite data, taking into account ground-observed SIF signals.

[0025] Figure 1 FIG. 1 is a flow chart of a method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by an embodiment of the present invention. Figure 1 As shown, including: Step 100: Collect and preprocess ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectance data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data; Step 200: Acquire chlorophyll fluorescence SIF observation data of ground stations, and perform time averaging and correction on the chlorophyll fluorescence SIF observation data of the ground stations; Step 300: using the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data as original data, and using a machine learning algorithm to reconstruct the data to obtain reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data; Step 400: construct an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data, calibrate the parameters of the empirical formula, and obtain an estimated parameter set of the empirical formula at each ground station by using a differential evolution algorithm for each ground station; Step 500: Classifying the ground stations based on global land cover classification data to obtain updated station data, and updating the empirical formula using the station data and the estimated parameter set; Step 600: Substitute the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data into the updated empirical formula to obtain cross-platform fused observation SIF data.

[0026] Specifically, Figure 2 As shown in the figure, firstly, pre-processed ground station data, OCO-2 SIF and TROPOMI SIF data, meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectivity data, enhanced vegetation index data, normalized difference moisture index data, and global land cover classification data are collected; then, machine learning methods are used to reconstruct data with OCO-2 SIF and TROPOMI SIF as target variables, and the reconstructed data are obtained respectively. and ; Then, build and With site observation information The empirical formula between is used to calibrate the parameters at each ground station using the differential evolution algorithm to obtain the estimated parameter set at each station. Then, the stations are classified according to the global land cover classification data, and the parameters of the stations with the same coverage type are averaged. Finally, and Substitute it into the model to obtain the reconstructed SIF data with the site observation SIF information .

[0027] The present invention makes full use of ground site observation data, which is more accurate than satellite remote sensing observations, and calibrates the obtained reconstructed SIF data, which can better reflect the actual SIF information of the surface. The SIF information of two different satellite remote sensing observation platforms, OCO-2 and TROPOMI, is taken into account, and the site observation SIF is integrated to generate cross-platform SIF reconstruction data with richer information. The data is scientific, reasonable and close to engineering practice.

[0028] Based on the above embodiment, step 100 includes: Collect and process ERA5-Land meteorological reanalysis temperature (T air ) data, leaf area index (LAI) data, MODIS bidirectional reflectance distribution function nadir corrected reflectance (NBAR) data, enhanced vegetation index (EVI) data, normalized difference moisture index (NDMI) data, global land cover (IGBP) classification data, OCO-2 satellite observation SIF data (OCO-2 SIF data) and TROPOMI satellite observation SIF data (TROPOMI SIF data).

[0029] Specifically, based on the ERA5-Land hourly meteorological reanalysis data, the temperature with a time resolution of 4 days (T air) variables, and the spatial resolution of the above data was unified to 0.05° through spatial interpolation. The TROPOMI SIF data were averaged, and the instantaneous observations were averaged to the daily average to be consistent with the OCO-2 SIF data used, specifically: ; Where SIF represents the measured instantaneous chlorophyll fluorescence, represents the daily average chlorophyll fluorescence, is the measurement time of instantaneous SIF, For measurement The solar zenith angle at is the time step, here we take 10 minutes for numerical calculation, H is the heaviside step function, For hours.

[0030] The quality control of satellite observation SIF data was performed, and the cloud-free OCO-2 SIF data and the TROPOMI SIF data with cloud percentage less than 10% were selected.

[0031] The temporal and spatial resolutions of LAI data, NBAR data, EVI data, NDMI data, OCO-2 SIF and TROPOMI SIF data were processed to 4 days and 0.05°.

[0032] Based on the above embodiment, step 200 includes: Performing time averaging on the chlorophyll fluorescence SIF observation data of the ground station at the preset time resolution of days; According to the approximately linear variation trend of the SIF spectral characteristic curve within a specified range, the ground station chlorophyll fluorescence SIF observation data are uniformly corrected to a wavelength consistent with the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data.

[0033] Specifically, the time resolution of ground station observations is relatively high, usually ranging from several hours to several minutes. To keep consistent with other variables, the station observation data are averaged here at a time resolution of 4 days. The selected ground station information is shown in Table 1. The wavelength center commonly used for ground station observations of SIF is between 740nm and 760nm (see Table 1). The observation data needs to be uniformly corrected to a wavelength (740nm) consistent with the OCO-2 and TROPOMI used. Here, the station observation data is uniformly corrected to 740nm according to the SIF spectral characteristic curve. The relationship between SIF signals of different wavelengths can be approximated as: ; ; in, , , They represent the SIF observation values ​​with 740nm, 750nm, and 760nm as the wavelength center. According to the variation trend of the SIF spectrum curve, it is approximately linear between 740nm and 760nm, so the relationship between the SIF observation values ​​can be obtained by linear interpolation.

[0034] Table 1 Ground station information

[0035] Based on the above embodiment, step 300 includes: The XGBoost machine learning method is adopted to optimize the model hyperparameters by random search, and the ERA5-Land meteorological reanalysis temperature data, the leaf area index data, the red, blue, green and near-infrared band data in the MODIS bidirectional reflectance distribution function nadir-corrected reflectivity data, the enhanced vegetation index data and the normalized difference moisture index data are used as independent variables, and the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data are used as target variables for training, verification and testing, respectively, to generate machine learning models corresponding to the respective target variables; The independent variables are respectively input into the machine learning models corresponding to the respective target variables to obtain the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data.

[0036] Specifically, the embodiment of the present invention adopts the XGBoost machine learning method, uses a random search method to optimize the model hyperparameters, and uses T air Data, LAI data, NBAR red (RED), blue (BLUE), green (GREEN), near infrared (NIR) band data, EVI data, NDMI data are used as independent variables, OCO-2 SIF and TROPOMI SIF are used as target variables for training, verification and testing, and XGBoost machine learning models are generated respectively. The present invention uses a random search method to optimize the hyperparameters of the machine learning model. Random search is an optimization method based on random numbers, which is mainly used to solve the optimal solution or approximate optimal solution of a function. The basic idea is to randomly sample a set of parameters in a given parameter space, and calculate the objective function values ​​corresponding to these parameter combinations, and gradually approach the optimal solution through continuous random sampling and evaluation. For the XGBoost machine learning model, the types and value ranges of the selected hyperparameters are as follows: "learning_rate": [0.001, 0.05]; "max_depth": [6, 12]; "n_estimators": [100, 400]; "num_leaves": [20, 40]; "learning_rate" indicates the learning rate of the model, also known as the step size; "max_depth" indicates the maximum depth of the decision tree, that is, the maximum path length that the decision tree can grow; "n_estimators" indicates the number of iterations, that is, the number of trees to be fitted; "num_leaves" indicates the number of leaf nodes, which should be consistent with the maximum depth of the decision tree. However, when the "max_depth" value is set to a large value, the number of leaf nodes can hardly reach the critical maximum value. At this time, in order to speed up the model training, the upper limit tends to be set smaller.

[0037] Afterwards, the XGBoost machine learning model was trained using the respective variable data with OCO-2 SIF and TROPOMI SIF as the target variables. The basic operation process of XGBoost machine learning is as follows: (1) Establishing the objective function: For a data set with n samples and m features, the tree ensemble model uses K additive functions to predict the output: ; in, is the regression tree space, each Corresponding to an independent tree structure and leaf weights ,use Represents the weight on the i-th leaf. Set the objective function for: ; in, is the loss function, , is the model complexity penalty function, is the complexity penalty coefficient of the tree, is the regularization parameter of the weight, is the number of leaf nodes in the tree, For the prediction, For the prediction The corresponding quantity to be predicted.

[0038] (2) Gradient tree boosting: After establishing the objective function, the model needs to continuously iterate the prediction amount To minimize the objective function. Since the objective function contains function parameters, the traditional optimization method cannot be used. Here we use the following optimization method: ; In the formula, t represents the number of iterations, and the second-order Taylor expansion approximation is used, then: ; in, , , since at the tth iteration, The term is already a constant, so the quantity to be optimized can be further written as: ; set up is the instance set of blade j, the above formula can be derived as follows: ; (3) Tree node splitting: For tree structure , the optimal weight of leaf j is: ; The corresponding optimal value is: ; Assume that after a single leaf splits, the instance sets of the left and right nodes are and ,make , then the loss reduction after splitting is: ; This formula is usually used to calculate candidate nodes for splitting. The exact greedy algorithm, approximate greedy algorithm and other algorithms can be used to find the optimal split point, realize the splitting of tree nodes, and then realize the training of XGBoost model.

[0039] Subsequently, the respective variable data were input into the model and the obtained XGBoost machine learning model was used for prediction to obtain the reconstructed SIF data based on OCO-2 SIF and TROPOMI SIF, respectively. and .

[0040] Based on the above embodiment, step 400 includes: Build and With site observation information The empirical formula between: ; In this formula, For reconstructed SIF data with site observation SIF information, the input includes: reconstructed SIF data based on OCO-2 SIF and TROPOMI SIF and ,parameter In this embodiment, only a linear empirical formula is used for illustration. The specific form of the empirical formula can be adjusted according to actual data. The linear empirical formula is: ; Differential evolution algorithm is used in each ground station and The relationship between the SIF data observed at the site is used to calibrate the parameters and obtain the parameters in the empirical formula. The estimated parameter set at each ground station. The differential evolution algorithm in the embodiment of the present invention is a conventional optimization technology. The basic operation process of the differential evolution algorithm is as follows: (1) Initialize the population: Create an initial population for the parameters to be optimized in the empirical formula, and assign a value to each dimension of each individual in the population. Each individual is represented as follows: ; in, represents the number of individuals in the population, represents the evolutionary algebra, represents the population size. In differential evolution algorithms, it is generally assumed that all randomly initialized populations conform to uniform distribution. Let the bounds of the parameter variables be ,but ,in Indicated in Produces uniform real numbers between; (2) Individual fitness evaluation: Calculate the fitness of each individual in the population. Here, the inverse of the model simulation error is used as the fitness function. (3) Mutation operation: A new parameter vector is generated by adding the weighted difference vector between two members of the population to the third member. This operation is called "mutation". After the initial population is generated, the mutation operation is performed. For each target , the mutation vector of the basic differential evolution algorithm is generated as follows: ; in, are three randomly selected individual numbers, which must be different from each other and have the same sequence as the target vector. Population size , mutation operator is a real constant factor that controls the scaling of the deviation variable. If a solution outside the feasible domain is compiled during the mutation process, that is, or ,So: ; (4) Crossover operation: The parameters of the mutation vector are mixed with the parameters of another predetermined target vector according to certain rules to generate a test vector. In order to increase the diversity of the interference parameter vector, the crossover operation is introduced, and the test vector becomes: ; ; in, Indicates the generation The jth estimate of the random number generator between represents a randomly selected sequence (this sequence must be in between, so there must be one that is equal to j), use it to ensure At least from Gets a parameter. represents the crossover operator, and its value range is ; (5) Selection operation: Compare the objective function value of the test vector with the cost function value of the target vector. If the objective function value of the test vector is lower than the cost function value of the target vector, then use the test vector to replace the target vector. (6) Termination: Determine whether the current generation is the maximum evolution generation. If so, terminate the operation and output the best individual in the current population as the optimal solution. The parameter calibration results of the differential evolution algorithm at each ground station are shown in Table 2.

[0041] Table 2 Parameter calibration results

[0042] Based on the above embodiment, step 500 includes: According to the global land cover classification data, the selected stations are classified, and the parameters of the stations belonging to the same land cover type are arithmetic averaged to obtain the parameter values ​​under each land cover type in the world. The parameter value results under each land cover type in the world are shown in Table 3.

[0043] Table 3 Parameter value results for each land cover type in the world

[0044] Based on the above embodiment, step 600 includes: The reconstructed data obtained and Substitute into the empirical formula The reconstructed SIF data with the site observation SIF information is obtained by calculating .like Figure 3 Shown is a schematic diagram of the probability density function of the global reconstructed SIF mean.

[0045] The following describes the sunlight-induced chlorophyll fluorescence reconstruction system based on satellite observation data provided by the present invention. The sunlight-induced chlorophyll fluorescence reconstruction system based on satellite observation data described below and the sunlight-induced chlorophyll fluorescence reconstruction method based on satellite observation data described above can be referenced to each other.

[0046] Figure 4 FIG. 1 is a schematic diagram of a system for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data provided by an embodiment of the present invention. Figure 4 As shown, it includes: a collection module 41, an acquisition module 42, a reconstruction module 43, a calibration module 44, a classification module 45 and a processing module 46, wherein: The acquisition module 41 is used to collect and preprocess ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectivity data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data; the acquisition module 42 is used to acquire ground station chlorophyll fluorescence SIF observation data, and perform time averaging and correction on the ground station chlorophyll fluorescence SIF observation data; the reconstruction module 43 is used to use the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data as original data, and use a machine learning algorithm to reconstruct the data to obtain the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data. SIF data; the calibration module 44 is used to construct an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data, calibrate the parameters of the empirical formula, and obtain the estimated parameter set of the empirical formula in each ground station by using the differential evolution algorithm for each ground station; the classification module 45 is used to classify the each ground station based on the global land cover classification data to obtain updated station data, and update the empirical formula using the station data and the estimated parameter set; the processing module 46 is used to substitute the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data into the updated empirical formula to obtain the cross-platform fusion observation SIF data.

[0047] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5As shown, the electronic device may include: a processor (processor) 510, a communication interface (Communications Interface) 520, a memory (memory) 530 and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 can call the logic instructions in the memory 530 to execute the daylight-induced chlorophyll fluorescence reconstruction method based on satellite observation data, the method comprising: collecting and preprocessing ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectance data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data; obtaining ground station chlorophyll fluorescence SIF observation data, time averaging and correcting the ground station chlorophyll fluorescence SIF observation data; using the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data as raw data, using a machine learning algorithm to reconstruct the data, and obtaining the reconstructed OC O-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data; construct an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data, calibrate the parameters of the empirical formula, and obtain the estimated parameter set of the empirical formula in each ground station by using the differential evolution algorithm for each ground station; classify the each ground station based on the global land cover classification data to obtain updated station data, and update the empirical formula using the station data and the estimated parameter set; substitute the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data into the updated empirical formula to obtain cross-platform fusion observation SIF data.

[0048] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0049] The device embodiments described above are merely illustrative, 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 this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0050] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data, characterized in that: include: Collect and preprocess ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectance data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data; Acquire chlorophyll fluorescence SIF observation data of ground stations, and perform time averaging and correction on the chlorophyll fluorescence SIF observation data of the ground stations; Using the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data as original data, a machine learning algorithm is used to reconstruct the data to obtain reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data; Constructing an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data, calibrating the parameters of the empirical formula, and obtaining an estimated parameter set of the empirical formula at each ground station by using a differential evolution algorithm for each ground station; Classifying the ground stations based on global land cover classification data to obtain updated station data, and updating the empirical formula using the station data and the estimated parameter set; The reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data are substituted into the updated empirical formula to obtain the cross-platform fused observation SIF data.

2. The method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data according to claim 1, characterized in that: Collect and preprocess ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectance data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data, including: Extract the temperature variables with a preset time resolution of days from the ERA5-Land meteorological reanalysis data at the hourly scale, and unify the spatial resolution of the temperature variables to the preset degree through spatial interpolation; Filtering the cloud-free data and the data with a cloud percentage less than a preset percentage in the OCO-2 satellite observation SIF data; Processing the leaf area index data, the MODIS bidirectional reflectance distribution function nadir-corrected reflectivity data, the enhanced vegetation index data, the normalized difference moisture index data, the OCO-2 satellite observation SIF data, and the TROPOMI satellite observation SIF data according to the preset day time resolution and the preset degree temporal and spatial resolution; The instantaneous values ​​of the TROPOMI satellite observed SIF data are averaged to obtain the daily average value, so that the TROPOMI satellite observed SIF data is consistent with the OCO-2 satellite observed SIF data: ； Where SIF stands for transient chlorophyll fluorescence, represents the daily average chlorophyll fluorescence, is the measurement time of instantaneous SIF, For measurement The solar zenith angle at is the time step, here we take 10 minutes for numerical calculation, H is the heaviside step function, For hours.

3. The method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data according to claim 2, characterized in that: Acquiring ground station chlorophyll fluorescence SIF observation data, and performing time averaging and correction on the ground station chlorophyll fluorescence SIF observation data, including: Performing time averaging on the chlorophyll fluorescence SIF observation data of the ground station at the preset time resolution of days; According to the approximately linear variation trend of the SIF spectral characteristic curve within a specified range, the ground station chlorophyll fluorescence SIF observation data are uniformly corrected to a wavelength consistent with the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data.

4. The method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data according to claim 1, characterized in that: The OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data are used as original data, and a machine learning algorithm is used to reconstruct the data to obtain reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data, including: The XGBoost machine learning method is adopted to optimize the model hyperparameters by random search, and the ERA5-Land meteorological reanalysis temperature data, the leaf area index data, the red, blue, green and near-infrared band data in the MODIS bidirectional reflectance distribution function nadir-corrected reflectivity data, the enhanced vegetation index data and the normalized difference moisture index data are used as independent variables, and the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data are used as target variables for training, verification and testing, respectively, to generate machine learning models corresponding to the respective target variables; The independent variables are respectively input into the machine learning models corresponding to the respective target variables to obtain the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data.

5. The method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data according to claim 1, characterized in that: Constructing an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data, and the reconstructed TROPOMI satellite observation SIF data includes: Constructing cross-platform fusion observation SIF data to be solved Reconstructed SIF data from OCO-2 satellite observations , the reconstructed TROPOMI satellite observation SIF data The empirical formula between: in, Parameters to be determined.

6. The method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data according to claim 5, characterized in that: The empirical formula is calibrated by using a differential evolution algorithm for each ground station to obtain a set of estimated parameters of the empirical formula at each ground station, including: Adjust the empirical formula to a linear form: Differential evolution algorithm is used to analyze the and , Perform parameter calibration and obtain the parameters A collection of estimated parameters at each ground station.

7. The method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data according to claim 1, characterized in that: The above-mentioned ground stations are classified based on the global land cover classification data to obtain updated station data, including: Classifying the ground stations according to the global land cover classification data; The updated site data are obtained by performing arithmetic averaging on the site parameters belonging to the same land cover type.

8. The method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data according to claim 1, characterized in that: Substituting the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data into the updated empirical formula to obtain cross-platform fusion observation SIF data, including: The reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data Substitute into the empirical formula Perform calculations to obtain cross-platform fusion observation SIF data .

9. A sunlight-induced chlorophyll fluorescence reconstruction system based on satellite observation data, characterized in that: include: The acquisition module is used to collect and preprocess ERA5-Land meteorological reanalysis temperature data, leaf area index data, MODIS bidirectional reflectance distribution function nadir-corrected reflectance data, enhanced vegetation index data, normalized difference moisture index data, global land cover classification data, OCO-2 satellite observation SIF data and TROPOMI satellite observation SIF data; An acquisition module is used to acquire chlorophyll fluorescence SIF observation data of ground stations, and perform time averaging and correction on the chlorophyll fluorescence SIF observation data of the ground stations; A reconstruction module is used to use the OCO-2 satellite observation SIF data and the TROPOMI satellite observation SIF data as original data and adopt a machine learning algorithm to reconstruct the data to obtain reconstructed OCO-2 satellite observation SIF data and reconstructed TROPOMI satellite observation SIF data; A calibration module is used to construct an empirical formula between the cross-platform fusion observation SIF data to be solved, the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data, calibrate the parameters of the empirical formula, and obtain an estimated parameter set of the empirical formula at each ground station by using a differential evolution algorithm for each ground station; A classification module, used for classifying the ground stations based on global land cover classification data to obtain updated station data, and updating the empirical formula using the station data and the estimated parameter set; The processing module is used to substitute the reconstructed OCO-2 satellite observation SIF data and the reconstructed TROPOMI satellite observation SIF data into the updated empirical formula to obtain the cross-platform fusion observation SIF data.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for reconstructing sunlight-induced chlorophyll fluorescence based on satellite observation data as described in any one of claims 1 to 8 is implemented.

Images