Inversion optimization algorithm based on vegetation leaf ecological correlation
By introducing ecological rules into vegetation parameter inversion, the ecological correlation of leaf biochemical parameters is analyzed and the confidence interval is constructed, the pathological problems in physical model inversion are solved, and the credibility and accuracy of the inversion results are improved.
Patent Information
- Application Number
- CN202411537994.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-31
- Publication Date
- 2025-02-18
AI Technical Summary
The vegetation parameter inversion method based on physical models is prone to pathological problems, resulting in instability and inaccuracy of the inversion results. The existing methods that introduce prior knowledge have not effectively solved this problem.
By introducing ecological rules, the ecological correlation between the biochemical parameters of vegetation leaves is analyzed, and the probability sample confidence interval of the relevant biochemical parameters is constructed, as a constraint on the inversion parameters, and the probability of incorrect parameter combination is reduced.
It effectively alleviates the problem of pathological inversion, improves the credibility and accuracy of the inversion results, and enhances the universality of the method.
Smart Images
Figure CN119474646A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of remote sensing inversion of vegetation parameters, and in particular to an inversion optimization algorithm based on ecological correlation of vegetation leaves. Background Art
[0002] The spectral characteristics of vegetation are closely related to its biochemical information, and the remote sensing monitoring of vegetation growth status can be achieved through the change of spectral characteristics. With the development and maturity of hyperspectral remote sensing technology, a large number of scholars have carried out quantitative analysis of vegetation biochemical content based on hyperspectral data. The researchers initially used empirical statistical methods for inversion, and established a regression relationship between remote sensing parameters and the studied biochemical parameters by conducting statistical correlation analysis on a series of observation data (usually the reflectance of the corresponding vegetation or its mathematical variation form, spectral index, etc.) and the corresponding biochemical content. This method can have a certain application effect on its specific observation area, but its robustness and portability are low. Therefore, starting from the radiation transmission process, based on the interaction mechanism between solar radiation and vegetation, the researchers proposed a physical model with certain universal significance, and each parameter in the model has a clear physical meaning. Based on the physical model, by inputting the biochemical parameters of the leaves, the vegetation spectrum can be simulated. This process is called the forward process, also known as forward modeling. If the vegetation spectrum is known, the biochemical parameters of the vegetation leaves corresponding to the spectrum can be obtained through the backward process (i.e., inversion). Compared with statistical methods, the physical model inversion method has clear physical meaning and can be applied regardless of time, location, or vegetation type. The physical model is not targeted at a certain vegetation type, growth period, or region, so it has good scalability. However, the vegetation parameter inversion based on the physical model method is often pathological, because the information contained in the remote sensing data is limited, the model and measurement have certain uncertainties, and different parameter combinations may correspond to approximately the same canopy reflectance spectrum. If there is no corresponding adjustment strategy for this type of problem, the final inversion result will be unstable and inaccurate.
[0003] The ill-conditioned inversion problem is an important problem that is widely present in the remote sensing inversion method of vegetation parameters based on physical models. Many scholars at home and abroad have proposed many methods or measures to alleviate this problem, but none of them have achieved good results. Introducing prior knowledge is one of the methods used to alleviate the uncertainty of inversion results caused by ill-conditioned characteristics. For example: limiting the range of values of the free parameters of the model to regularize the inversion results.
[0004] However, the defect of the prior art is that: this method can avoid the generation of outliers to a certain extent, but within the range of values, the free combination of parameters will still appear contrary to the actual situation, causing the problem of ill-conditioned inversion, resulting in uncertainty in the inversion results, or even failure. This happens because previous studies believe that the free parameters of the model are independent of each other, which is not in line with the actual situation of natural ecosystems. In fact, the biochemical parameters of leaves in nature are not independent, but interdependent. Ignoring this interdependence between the biochemical parameters of leaves in nature will inevitably introduce the situation of incorrect combination of free parameters in the inversion process, aggravating the ill-conditioned problem, resulting in abnormal inversion results or even failure.
[0005] Therefore, the present invention proposes a new idea of utilizing prior knowledge, that is, to deal with the problem of ill-conditioned inversion by introducing ecological rules, fully considering the ecological correlation between the biochemical components of vegetation leaves, constructing the probability sample confidence interval of relevant biochemical parameters, reducing the probability of erroneous parameter combinations, and alleviating the problem of ill-conditioned inversion. Summary of the invention
[0006] The object of the present invention is to provide an inversion optimization algorithm based on the ecological correlation of vegetation leaves to alleviate the pathological problem of parameter inversion of vegetation parameter physical model.
[0007] In order to solve the above technical problems, the present invention provides a technical solution: an inversion optimization algorithm based on the ecological correlation of vegetation leaves, comprising:
[0008] S1. Generate a simulated data set through a leaf radiation transfer model, analyze the correlation between different leaf biochemical parameters according to the simulated data set, select a group of leaf biochemical parameters whose correlation coefficient is greater than a set threshold, and generate a linear regression equation for the group of biochemical parameters based on the simulated data set;
[0009] S2. According to the linear regression equation and the set confidence level, a probability sample confidence interval of the group of leaf biochemical parameters is constructed, and a confidence interval determination condition is set according to the probability sample confidence interval;
[0010] S3, judging whether a group of initial inversion values of biochemical parameters obtained by inversion of the leaf radiation transfer model meets the confidence interval judgment condition, if so, taking the group of initial inversion values as the final inversion values of the parameters, otherwise, executing the subsequent steps;
[0011] S4, according to the coordinate points formed by the inversion initial values of the group of biochemical parameters, obtain the vertical points from the coordinate points to the fitting line formed by the linear regression equation, and calculate the adjustment variables corresponding to the inversion initial values of the group of biochemical parameters according to the coordinates of the coordinate points and the vertical points respectively;
[0012] S5. Call each adjustment weight in the set adjustment weight sequence in turn, and cyclically calculate the adjusted parameter value based on the adjustment variable in turn. When the adjusted parameter value meets the confidence interval judgment condition, the adjusted parameter value is used as the final inversion value of the parameter.
[0013] According to the above scheme, the blade radiation transfer model is the PROSPECT blade radiation transfer model.
[0014] According to the above scheme, a set of leaf biochemical parameters selected in step S1 is leaf chlorophyll content-leaf carotenoid content.
[0015] According to the above scheme, the confidence interval judgment condition is: for a certain set of biochemical parameter inversion initial values / adjusted parameter values x, y, if y is within the confidence interval of the set confidence level, then it is considered that the set of biochemical parameter inversion initial values / adjusted parameter values x, y meets the confidence interval judgment condition; the confidence interval is obtained based on the biochemical parameter inversion initial values / adjusted parameter values, the linear regression equation, and the set confidence level.
[0016] According to the above scheme, the initial values of biochemical parameter inversion are expressed as CabM and CarM, and the method for determining the adjustment variables is: draw a perpendicular line from the coordinate point (CabM, CarM) of the initial value of the biochemical parameter inversion to the fitting straight line, and the coordinates of the vertical point are (CabR, CarR), and then calculate DCab=CabR-CabM, DCar=CarR-CarM, and DCab and DCar are used as adjustment variables.
[0017] According to the above scheme, the method for cyclically calculating the adjusted parameter value based on the adjustment variable is: first take the first adjustment weight in the adjustment weight sequence, multiply the adjustment weight with the two parameters of the adjustment variable respectively, and then add it with the corresponding biochemical parameter inversion initial value to obtain the adjusted parameter value; if the adjusted parameter value still does not meet the confidence interval judgment condition, take the subsequent adjustment weights of the first adjustment weight in turn, and calculate the adjusted parameter value with the adjustment variable until the adjusted parameter value meets the confidence interval condition.
[0018] According to the above scheme, the value range of the adjustment weight is (0, 1].
[0019] According to the above scheme, the adjustment weight sequence is expressed as Weight = [0.2, 0.3, 0.5, 0.7, 0.8, 1].
[0020] The present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned inversion optimization algorithm based on the ecological correlation of vegetation leaves when executing the computer program.
[0021] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the inversion optimization algorithm based on the ecological correlation of vegetation leaves described above are implemented.
[0022] The beneficial effects of the present invention are as follows: the present invention analyzes the correlation between the leaf biochemical parameters of vegetation leaves, selects a pair of leaf biochemical parameters with a higher correlation to construct a probability sample confidence interval of the leaf biochemical parameters, and uses the obtained confidence interval as a constraint condition for the inversion parameter, thereby reducing the uncertainty of the experimental results caused by the wrong combination of the inversion parameters, alleviating the pathological problem, and ultimately improving the inversion accuracy. Compared with the existing inversion result optimization method, this method does not rely too much on measured data and has better universality. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 is a schematic diagram of the linear relationship between leaf chlorophyll and leaf carotenoid obtained based on a simulated data set according to an embodiment of the present invention;
[0024] Figure 2 1 is a schematic diagram of a confidence interval at a confidence level of 95% obtained based on the Cab and Car linear regression equations according to an embodiment of the present invention;
[0025] Figure 3 is a schematic diagram of obtaining an intermediate parameter pair according to an embodiment of the present invention;
[0026] Figure 4 This is a flow chart of an inversion optimization algorithm that considers the ecological correlation between biochemical components of vegetation leaves according to an embodiment of the present invention. DETAILED DESCRIPTION
[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings of the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.
[0028] In order to solve the ill-conditioned inversion problem during the inversion of physical models, this embodiment provides an inversion optimization algorithm that considers the ecological correlation between biochemical components of vegetation leaves. This method is different from the existing idea of using prior knowledge. It innovatively introduces ecological rules to deal with the ill-conditioned inversion problem, fully considers the ecological correlation between biochemical components of vegetation leaves, constructs the probability sample confidence interval of relevant biochemical parameters, reduces the probability of erroneous parameter combinations, alleviates the ill-conditioned inversion problem, and improves the credibility of the inversion results. Figure 4, the method comprises the following steps:
[0029] S1. Generate a simulated data set through a leaf radiation transfer model, analyze the correlation between different leaf biochemical parameters according to the simulated data set, select a group of leaf biochemical parameters whose correlation coefficient is greater than a set threshold, and generate a linear regression equation for the group of biochemical parameters based on the simulated data set;
[0030] In this embodiment, the blade radiation transfer model adopts the PROSPECT radiation transfer model.
[0031] S2. According to the linear regression equation and the set confidence level, a probability sample confidence interval of the group of leaf biochemical parameters is constructed, and a confidence interval determination condition is set according to the probability sample confidence interval;
[0032] Through correlation analysis, the grouped leaf biochemical parameters that meet the set threshold requirements include leaf chlorophyll content (Cab)-leaf carotenoid content (Car); it should be understood that there are several groups of leaf biochemical parameters that meet the set threshold requirements of correlation, among which the correlation coefficient between leaf chlorophyll content and leaf carotenoid content is the highest, which is 0.85. The linear relationship between leaf chlorophyll content and leaf carotenoid content in the simulated data set of this embodiment is as follows Figure 1 As shown, this group of leaf biochemical parameters is selected in this embodiment to calculate subsequent constraints.
[0033] According to the principle of confidence interval, the confidence interval of y at the significance level α is:
[0034]
[0035] In the above formula, x represents Cab, x i represents the value of the i-th Cab in the sample, represents the mean of x in the sample, represents the estimated value of Car obtained by the linear regression equation, y represents the actual value of Car, and t α / 2 represents the critical value of t distribution under the significance level α, n represents the number of samples, S represents the standard error of regression, and nm-1 represents the degree of freedom;
[0036] For a leaf biochemical parameter set of leaf chlorophyll content-leaf carotenoid content, if the leaf carotenoid content is within the set confidence interval, the parameter pair is considered to meet the confidence interval judgment condition, wherein the confidence interval is obtained by the leaf chlorophyll content in the parameter pair, the linear regression equation (the linear regression equation is used to obtain the predicted value of the leaf carotenoid content corresponding to the leaf chlorophyll content in the set of leaf biochemical parameters), and the set confidence level. In this embodiment, the confidence interval is: a confidence interval with leaf chlorophyll content as the independent variable, leaf carotenoid content as the dependent variable, and a confidence level of 95%. See Figure 2 , LCar and UCar represent the lower and upper limits of the confidence level of the dependent variable for a certain leaf chlorophyll content, respectively. As the independent variable gradually increases from 0, the corresponding lower and upper limits of the dependent variable also gradually change, forming a Figure 2 The two dotted lines above and below are shown in .
[0037] S3, judging whether a group of initial inversion values of biochemical parameters obtained by inversion of the leaf radiation transfer model meets the confidence interval judgment condition, if so, taking the regrouped initial inversion value as the final inversion value of the parameter, otherwise executing the subsequent steps;
[0038] For example, the initial value of the biochemical parameter inversion of a set of leaf chlorophyll content-leaf carotenoid content (i.e., a set of inversion parameter pairs) obtained by model inversion is expressed as (CabM, CarM); based on CabM and the linear regression equation, the leaf carotenoid content CarM' is calculated, and then the confidence interval [LCar, UCar] corresponding to CarM' is calculated; if CarM is in [LCar, UCar], then (CabM, CarM) meets the confidence interval judgment condition, Figure 2 In the figure, the point represented by (CabM, CarM) is located between the dotted lines represented by LCar and UCar.
[0039] S4, according to the coordinate points formed by the inversion initial values of the group of biochemical parameters, obtain the vertical points from the coordinate points to the fitting line formed by the linear regression equation, and calculate the adjustment variables corresponding to the inversion initial values of the group of biochemical parameters according to the coordinates of the coordinate points and the vertical points respectively;
[0040] Specifically, see Figure 3 First, draw a vertical line from the coordinate point of the initial value of the biochemical parameter inversion to the fitting line. The coordinates of the vertical point are marked as (CabR, CarR). Then calculate DCab = CabR-CabM, DCar = CarR-CarM. DCab and DCar are the adjustment variables.
[0041] S5. Call each adjustment weight in the set adjustment weight sequence in turn, and cyclically calculate the adjusted parameter value based on the adjustment variable in turn. When the adjusted parameter value meets the confidence interval judgment condition, the adjusted parameter value pair is used as the final inversion value of the parameter.
[0042] The adjustment weights in the adjustment weight sequence are arranged in order from small to large, and the value range of the adjustment weight is (0, 1]. In this embodiment, the adjustment weight sequence is set to Weight = [0.2, 0.3, 0.5, 0.7, 0.8, 1]. For example: for the intermediate parameter pair DCab and DCar, the adjusted parameter inversion value pair is expressed as CabA and CarA, CabA = CabM + Weight*DCab, CarA = CarM + Weight*DCar; first take the first adjustment weight in the adjustment weight sequence, that is, Weight = 0.2, and after obtaining the adjusted parameter values CabA and CarA, recalculate the new confidence interval based on the adjusted parameter inversion value. When CarA is not within its corresponding confidence interval, take the next adjustment weight, that is, Weight = 0.3, and so on, until CabA and CarA meet the confidence interval judgment condition.
[0043] Repeat steps S4 to S5 until the initial values of all biochemical parameters are screened.
[0044] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. An inversion optimization algorithm based on vegetation leaf ecological correlation, characterized in that: include: S1. Generate a simulated data set through a leaf radiation transfer model, analyze the correlation between different leaf biochemical parameters according to the simulated data set, select a group of leaf biochemical parameters whose correlation coefficient is greater than a set threshold, and generate a linear regression equation for the group of biochemical parameters based on the simulated data set; S2. According to the linear regression equation and the set confidence level, a probability sample confidence interval of the group of leaf biochemical parameters is constructed, and a confidence interval determination condition is set according to the probability sample confidence interval; S3, judging whether a group of initial inversion values of biochemical parameters obtained by inversion of the leaf radiation transfer model meets the confidence interval judgment condition, if so, taking the group of initial inversion values as the final inversion values of the parameters, otherwise, executing the subsequent steps; S4, according to the coordinate points formed by the inversion initial values of the group of biochemical parameters, obtain the vertical points from the coordinate points to the fitting line formed by the linear regression equation, and calculate the adjustment variables corresponding to the inversion initial values of the group of biochemical parameters according to the coordinates of the coordinate points and the vertical points respectively; S5. Call each adjustment weight in the set adjustment weight sequence in turn, and cyclically calculate the adjusted parameter value based on the adjustment variable in turn. When the adjusted parameter value meets the confidence interval judgment condition, the adjusted parameter value is used as the final inversion value of the parameter.
2. The inverse optimization algorithm based on vegetation leaf ecological correlation according to claim 1 is characterized in that: The blade radiation transfer model is the PROSPECT blade radiation transfer model.
3. The inverse optimization algorithm based on vegetation leaf ecological correlation according to claim 1 is characterized in that: The set of leaf biochemical parameters selected in step S1 is leaf chlorophyll content-leaf carotenoid content.
4. The inverse optimization algorithm based on vegetation leaf ecological correlation according to claim 1 is characterized in that: The confidence interval determination condition is: for a certain set of biochemical parameter inversion initial values / adjusted parameter values x, y, if y is within the confidence interval of the set confidence level, then the set of biochemical parameter inversion initial values / adjusted parameter values x, y is considered to meet the confidence interval determination condition; The confidence interval is obtained based on the initial value / adjusted parameter value of the biochemical parameter inversion, the linear regression equation, and the set confidence level.
5. The inverse optimization algorithm based on vegetation leaf ecological correlation according to claim 1 is characterized in that: The initial values of biochemical parameter inversion are expressed as CabM and CarM. The method for determining the adjustment variables is: draw a perpendicular line from the coordinate point (CabM, CarM) of the initial value of the biochemical parameter inversion to the fitting straight line, and the coordinates of the vertical point are (CabR, CarR), and then calculate DCab=CabR-CabM, DCar=CarR-CarM, and DCab and DCar are used as adjustment variables.
6. The inverse optimization algorithm based on vegetation leaf ecological correlation according to claim 1 is characterized in that: The method for cyclically calculating the adjusted parameter value based on the adjustment variable is as follows: first, take the first adjustment weight in the adjustment weight sequence, multiply the adjustment weight by the two parameters of the adjustment variable respectively, and then add the adjustment weight to the corresponding biochemical parameter inversion initial value to obtain the adjusted parameter value; if the adjusted parameter value still does not meet the confidence interval judgment condition, take the subsequent adjustment weights of the first adjustment weight in turn, and calculate the adjusted parameter value with the adjustment variable until the adjusted parameter value meets the confidence interval condition.
7. The inverse optimization algorithm based on vegetation leaf ecological correlation according to claim 1 is characterized in that: The value range of the adjustment weight is (0, 1].
8. The inverse optimization algorithm based on vegetation leaf ecological correlation according to claim 1 is characterized in that: The adjustment weight sequence is expressed as Weight = [0.2, 0.3, 0.5, 0.7, 0.8, 1].
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the inversion optimization algorithm based on the ecological correlation of vegetation leaves as described in any one of claims 1-8 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the inversion optimization algorithm based on the ecological correlation of vegetation leaves as described in any one of claims 1 to 8 are implemented.