Method for evaluating Budyko key parameter change influence factors
By constructing the dataset of the Budyko model and using the multi-dimensional Copula model and C-Vine tree structure, the influencing factors of the key parameter n value in the Budyko model are evaluated, and the problems of variability and multi-factor influence of parameter n value in space are solved, and the problem of the influence of parameter n value in space is realized. In-depth study on the distribution mode and influencing factors of the parameter n-value.
Patent Information
- Application Number
- CN202411066141.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-19
AI Technical Summary
The n-value of the model parameter in the Budyko model is highly variable in space and is affected by many factors such as climate and vegetation, resulting in huge challenges in the estimation and distribution mode research of parameter values, and the interaction mechanism between variables and model parameters is unclear.
By constructing the input dataset of the Choudhury-Yang equation in the Budyko model and the variable dataset that affects the model parameter value, using the multi-dimensional Copula model and the C-Vine series tree structure, the distribution mode and correlation of the n-value of the key parameter are quantified, and the maximum impact is selected variables and a multivariate linear regression model is constructed to determine the independent effect of key variables on parameter n-values.
A detailed study of the distribution pattern of key parameters of the Budyko model on the spatial scale, determine the main influencing factors that cause the numerical changes of the variables, reveal the interconnection between each variable and the relationship with the parameter n value, and provide an accurate evaluation of the change of the parameter n value.
Smart Images

Figure CN118982285A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of research on variables related to surface evapotranspiration, and in particular to a method for evaluating factors affecting changes in Budyko key parameters. Background Art
[0002] The Budyko model is a theoretical model for studying the response of surface water distribution and water balance to climate change. Later, Choudhury considered the impact of spatial changes in basin precipitation and net radiation on actual evapotranspiration, and improved the original Budyko equation. The proposed equation has the same form as the equation obtained by Yang et al. through mathematical reasoning, and summarized the Choudhury-Yang equation under the framework of the Budyko model, which contains a model parameter n that reflects the characteristics of the basin, indicating the proportional relationship between evapotranspiration and precipitation in the basin. Since then, the Choudhury-Yang equation has been widely used in the Budyko model.
[0003] However, since the model parameter values are highly variable in space, and this feature is more obvious in small watersheds, it poses a huge challenge to effectively estimate the parameter n value of the Budyko model. In addition, the model parameter n value is affected by multiple factors such as climate and vegetation, and the mechanism of interaction between variables and model parameters is still unclear. Therefore, developing a method to evaluate the influencing factors of the changes in the key parameters of the Budyko model is of great significance for in-depth research on the distribution pattern of the key parameters of the Budyko model on spatial scales, the change law of the values under different environmental scenarios, and the main influencing factors that cause the changes in the values of the variables. Summary of the invention
[0004] In order to overcome the deficiencies of the prior art, the object of the present invention is to provide a method for evaluating the influencing factors of Budyko key parameter changes.
[0005] To achieve the above object, the present invention provides the following solutions:
[0006] A method for evaluating factors affecting changes in Budyko key parameters, comprising:
[0007] Construct the input data set of Choudhury-Yang equation in Budyko model and the variable data set that affects the model parameter values;
[0008] The key parameter n of the model is estimated based on the Choudhury-Yang equation in the Budyko model, and the distribution pattern of the key parameter n value in the basin is quantified;
[0009] Quantify the joint influence of i variables in the variable dataset on the change of the key parameter n using a multi-dimensional Copula model, and select j (j < i) key variables that have the greatest influence on the change of the value of the key parameter n.
[0010] Establish a C-Vine series tree structure for the key variables, use the value of the key parameter n as the central node of the T1 tree, and establish the correlation relationship between the value of the key parameter n and each key variable using the Copula function.
[0011] Based on the correlation relationship between the value of the key parameter n and each key variable, construct Vine trees from the T’2 to T”j layers to obtain the correlation relationship between the key parameter n and each variable under different conditions; among them, the T’2 tree and the T”j tree are not the same group of trees as the original tree. To ensure that each group of trees can represent the correlation between the key variables and the value of the key parameter n in different situations, each layer of tree in the above method has a different root node.
[0012] Based on the correlation relationship between the key parameter n and each key variable under different conditions, introduce a weight factor, construct a multiple linear regression model, and determine the independent influence and degree of influence of the key variables on the value of the key parameter n according to the multiple linear regression model.
[0013] Preferably, the input dataset includes the input data used for estimating the Budyko model parameters and the environmental variables that affect the change of the model parameter values, where the input data used for estimating the Budyko model parameters includes precipitation pre, actual evapotranspiration ET, and potential evapotranspiration PET.
[0014] Preferably, the formula of the Choudhury-Yang equation is: where n is the key parameter of the model representing the basin characteristics.
[0015] Preferably, using a multi-dimensional Copula model to quantify the joint influence of i variables in the variable dataset on the change of the key parameter n, and selecting j (j < i) key variables that have the greatest influence on the change of the value of the key parameter n, including:
[0016] Select the marginal distribution function of each variable;
[0017] Perform optimal parameter estimation on the selected marginal distribution function so that the cumulative distribution function CDF of any variable marginal distribution is uniformly distributed within the range of [0, 1];
[0018] Determine the Copula function; the formula of the Copula function is: C(u) = Φ(Φ -1 (u1), Φ -1 (u2),..., Φ -1 (u d)); where Φ represents the standard normal distribution function, Φ -1 represents the inverse function of the standard normal distribution function, u d represents the probability distribution function of the d-th random variable, C(u) is the Copula function;
[0019] Input a multivariate Copula function parameter set, calculate and output the joint CDF of the parameters under the above Copula function;
[0020] The environmental variables in the combination with the largest joint CDF are identified as the key variables that have the greatest impact on the numerical changes of the key parameter n.
[0021] Preferably, a Copula function is used to establish a correlation between the key parameter n value and each key variable, including:
[0022] After determining the C-Vine series tree structure, estimate the binary Copula functions of all edges in the T1 tree, calculate the AIC information criterion for all selected Copula functions, and select the Copula function with the smallest AIC result;
[0023] According to the order of the C-Vine series tree structure, the maximum likelihood method is used to estimate the parameters of the Copula function with the smallest AIC result; the correlation between the key parameter n value and each key variable is represented by each edge on the Vine tree in the C-Vine series tree structure.
[0024] Preferably, the binary Copula function includes Frank Copula, Clayton Copula and Gumbel Copula.
[0025] Preferably, the calculation formula of the AIC result is: AIC=2k+2ln(L); wherein k is the number of parameters and L is the likelihood function.
[0026] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:
[0027] The present invention provides a method for evaluating the influencing factors of changes in Budyko key parameters, including: constructing an input data set for the Choudhury-Yang equation in the Budyko model and a variable data set affecting the model parameter values; estimating the key parameter n of the model based on the Choudhury-Yang equation in the Budyko model, and quantifying the distribution pattern of the key parameter n value within the basin; using a multi-dimensional Copula model to quantify the joint influence of i variables in the variable data set on the change of the key parameter n, and selecting j (j < i) key variables that have the greatest influence on the change of the key parameter n value; establishing a C-Vine series tree structure for the key variables, taking the key parameter n value as the central node of the T1 tree, and using the Copula function to establish the correlation relationship between the key parameter n value and each key variable; based on the correlation relationship between the key parameter n value and each key variable, constructing Vine trees at the T’2 to T”j levels to obtain the correlation relationship between the key parameter n and each variable under different conditions; based on the correlation relationship between the key parameter n and each key variable under different conditions, introducing a weight factor, constructing a multiple linear regression model, and determining the independent influence and influence degree of the key variables on the key parameter n value according to the multiple linear regression model. The present invention is of great significance for in-depth study of the distribution pattern of the key parameters of the Budyko model at the spatial scale, determining the main influencing factors causing the change of the variable values, exploring how the variables are interconnected with each other, and the mutual relationship between the variables and the parameter n value. Description of the Drawings
[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.
[0029] Figure 1 It is a flowchart of the method provided by the embodiment of the present invention;
[0030] Figure 2 It is a step flowchart provided by the embodiment of the present invention;
[0031] Figure 3 It is a schematic diagram of the Vine series tree provided by the embodiment of the present invention;
[0032] Figure 4 It is a schematic diagram of the mutual relationship provided by the embodiment of the present invention. Detailed Embodiments
[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.
[0034] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0035] Figure 1 A flow chart of a method provided by an embodiment of the present invention, such as Figure 1 As shown, the present invention provides a method for evaluating the influencing factors of Budyko key parameter changes, comprising:
[0036] Step 100: construct an input data set of the Choudhury-Yang equation in the Budyko model and a variable data set that affects the model parameter values;
[0037] Step 200: estimating a key parameter n of the model based on the Choudhury-Yang equation in the Budyko model, and quantifying the distribution pattern of the key parameter n value in the watershed;
[0038] Step 300: Use a multidimensional Copula model to quantify the joint impact of i variables in the variable data set on the change of key parameter n, and select j key variables that have the greatest impact on the change of the value of key parameter n; where j <i。
[0039] Step 400: Establish a C-Vine series tree structure for the key variables, use the key parameter n value as the central node of the T1 tree, and use the Copula function to establish the correlation between the key parameter n value and each key variable;
[0040] Step 500: Based on the correlation between the key parameter n value and the key variables, construct a Vine tree from T'2 to T"j layers to obtain the correlation between the key parameter n and each variable under different conditions;
[0041] Step 600: Based on the correlation between the key parameter n and each key variable under different conditions, a weight factor is introduced to construct a multiple linear regression model, and the independent influence of the key variable on the key parameter n value and the influence degree are determined according to the multiple linear regression model.
[0042] like Figure 2 As shown, the specific steps of this embodiment are as follows:
[0043] Step 1: Initially construct the input dataset for the Choudhury - Yang equation in the Budyko model and the variable dataset that affects the model parameter values;
[0044] Among them, the input dataset includes the input data used for estimating the Budyko model parameters and the environmental variables that affect the variation of the model parameter values. The input data used for estimating the Budyko model parameters includes precipitation pre (precipitation), actual evapotranspiration ET (evapotranspiration), and potential evapotranspiration PET (potential evapotranspiration). The environmental variables that affect the variation of the model parameter values are X1, X2, X3, …, X i , and the above environmental variables are reference example data and can be replaced according to the actual situation and requirements.
[0045] Step 2: Estimate the key parameter n of the model based on the Choudhury - Yang equation in the Budyko model and quantify the distribution pattern of the parameter n value within the basin;
[0046] Among them, the Choudhury - Yang equation in the Budyko model used is Equation 1:
[0047]
[0048] Among them, ET is the actual evapotranspiration (mm), pre is the total precipitation (mm), PET is the potential evapotranspiration (mm), and n is the model parameter representing the basin characteristics (dimensionless). It can not only be used to describe the relationship between precipitation and evaporation to reflect the changes in basin characteristics and the hydrological cycle, but also reflects the proportion of actual evapotranspiration in precipitation within the basin, and at the same time controls the shape of the Budyko curve.
[0049] Step 3: Use the multi - dimensional Copula model to quantify the joint influence of multiple (i) variables in Step 1 on the variation of the model parameter n value, that is, the joint distribution function, and select j (j < i) key variables that have the greatest influence on the variation of the Budyko model parameter n value;
[0050] Among them, use the multi - dimensional Copula model to quantify the joint influence of multiple variables in Step 1 on the variation of the model parameter n value, that is, the joint distribution function. Given the parameter n value, obtain j (j < i) key variables that have the greatest influence on the variation of the Budyko model parameter n value. The core part of the Copula model is to calculate and select the marginal distributions of the variables, select the Copula function, and parameter estimation. The calculation process for the multi - dimensional Copula model is as follows:
[0051] ① Select the marginal distribution function of each variable;
[0052] ② Optimal parameter estimation for the selected marginal distribution function: make the cumulative distribution function CDF of any variable marginal distribution uniformly distributed in the range [0,1];
[0053] ③Select Copula function: The commonly used multidimensional Copula function can use Gaussian Copula, formula 2:
[0054] C(u)=Φ(Φ -1 (u1),Φ -1 (u2),...,Φ -1 (u d ))
[0055] Among them, Φ represents the standard normal distribution function, Φ -1 represents the inverse function of the standard normal distribution function, u1,u2,…,u d Represents the probability distribution function of d random variables.
[0056] ④ Calculate the CDF of the joint distribution of multiple variables: input the Copula function parameter set of multiple variables, calculate and output the joint CDF of the parameters under Gaussian Copula. The combination with the largest joint CDF can be considered as the environmental variable that has the greatest impact on the numerical change of the model parameter n.
[0057] Step 4: Establish a C-Vine series tree structure for the j key variables obtained in step 3, use the parameter n value as the central node of the T1 tree, and use the Copula function to establish the central node (parameter n value) and the key variables x1, x2, ..., x j The correlation between them (represented by the edges on the Vine tree);
[0058] Establish a C-Vine series tree structure for the key variables obtained in step 3, and use the parameter n value as the central node of the T1 tree. The purpose of selecting the C-Vine tree structure is to directly obtain the correlation between each variable and the parameter n value in step 3. After determining the Vine tree structure, it is necessary to estimate the binary Copula function of all edges in the T1 tree, calculate the AIC information criterion (Akaike information criterion) for all selected Copula functions, and select the Copula with the smallest AIC result. Commonly used two-dimensional Copula functions in hydrology include Frank Copula, Clayton Copula, and Gumbel Copula. The form of the Frank Copula function is formula three:
[0059]
[0060] The form of the Gumbel Copula function is formula 4:
[0061]
[0062] The form of the Clayton Copula function is Formula 5:
[0063] C(U,V)=(U -θ +V -θ -1) -1 / θ ,θ≥0
[0064] Among them, θ is the parameter of the binary Copula function.
[0065] The AIC information criterion is a standard for measuring the goodness of fit of a statistical model. This method is based on the concept of entropy and can weigh the complexity of the estimated model and the goodness of fit of the model to the data. The calculation method of the model is formula 6:
[0066] AIC=2k+2ln(L)
[0067] Among them, k is the number of parameters, L is the likelihood function, and the smaller the AIC value, the higher the fitting degree of the Copula joint distribution function. Finally, the maximum likelihood method is used to estimate the parameters of the Copula function in step 4. Since the Vine series tree contains j layers of tree structure, it is more complicated to directly perform maximum likelihood estimation. Therefore, hierarchical estimation can be used, that is, first estimate the Copula parameters of each edge on the T1 layer tree, then estimate the Copula parameters of each edge on the T2 layer tree, and so on.
[0068] Step 5: Based on the above step 4, further construct the T'2~T"j layer (not belonging to the same series of trees) Vine tree to obtain the correlation between the parameter n and each variable under different conditions, revealing the degree of dependence between variables under given conditions;
[0069] Using the Vine Copula model established in step 4 above, the Copula values of all edges contained in the Vine series tree are calculated, that is, the composite correlation (T1 layer) or conditional correlation between different variables and parameters, revealing the degree of dependence between parameter n and each key variable under given conditions. The established Vine series tree structure is as follows: Figure 3 As shown, the number 0 is the parameter n value, and 1, 2, 3 and 4 are the key variables that affect the change of the parameter n value.
[0070] Figure 3Each node in the T1 layer tree represents the marginal distribution of the corresponding variable. Starting from the T2 layer, the nodes of each layer of the tree are the edges of the previous layer of the tree and represent the copula of two variables. Starting from the T3 layer, the nodes of each layer of the tree are only valid if and only if the two nodes share a node in the previous layer of the tree. Take 3,4; 0,1,2 in the figure as an example, it means the copula between 3 and 4 under the conditions of 0,1,2, that is, it represents the correlation between variables 3 and 4.
[0071] Step 6: Based on the conclusion of step 5, introduce weight factors and construct a multiple linear regression model. The results will reflect the independent impact of the variables on the parameter n value and the degree of impact.
[0072] Among them, it is assumed that 1, 2, 3 and 4 are four variables that affect the value of parameter n. The above variables are not independent of each other. The possible relationships among them are as follows: Figure 4 As shown, if we get the independent relationship (C') between variable 3 and parameter n value (number 0), we have formula 7:
[0073] C'(3,0)=C(3,0)-C((3,4),0)-C((1,3),0)-C((1,3,4),0)-C((1,2,3),0)-C((1,2,3,4),0)
[0074] -C((2,3),0)-C((2,3,4),0)-C((3,4),0)
[0075] Among them, C(3,0) is the correlation between variable 3 and the key parameter n value (number 0), C((3,4),0) is the correlation between variables 3 and 4 after mixing and the key parameter n value (number 0), and so on. The inner brackets represent the mixing of multiple (>2) variables.
[0076] For the relationship between multiple variables and parameter n values, it is necessary to discuss the combination conditions of different variables. For example, the correlation C((3,4),0) between variables 3 and 4 and the key parameter n value (number 0) includes C(3,0|4) and C(4,0|3), and the weights of C(3,0|4) and C(4,0|3) are unknown. Let them be a and b, then C((3,4),0)=a*C(3,0|4)+b*C(4,0|3); a+b=1. Therefore, formula 7 after introducing the weight factor can further derive formula 8:
[0077] C'(3,0)=C(3,0)-a×C(3,0|4)-b×C(4,0|3)-d×C(1,0|3)-e×C(3,0|1)-f×C(1,0|3, 4)-g×C(3,0|1,4)-h×C(4,0|1,3)-i×C(1,0|2,3)-j×C(2,0|1,3)-k×C(3,0|1,2)-m× C(1,0|2,3,4)-n×C(2,0|1,3,4)-p×C(3,0|1,2,4)-q×C(4,0|1,2,3)-r×C(2,0|3)-s ×C(3,0|2)-t×C(2,0|3,4)-u×C(3,0|2,4)-v×C(4,0|2,3)-x×C(3,0|4)-y×C(4,0|3)
[0078] Among them, the relationship between each weight factor is as shown in Formula 9:
[0079] a+b=1; d+e=1; f+g+h=1; i+j+k=1;
[0080] m+n+p+q=1; r+s=1; t+u+v=1; x+y=1
[0081] In practical applications, a raster dataset with a large enough data volume is brought into the model for calculation to obtain the weight factor of each conditional Copula function and the independent correlation between the variable and the parameter n, that is, the true impact of the studied variable on the parameter n value after excluding the interference of other environmental variables.
[0082] The method for evaluating the influencing factors of Budyko key parameter changes proposed by the present invention has the following characteristics: 1) Selecting the statistical model Copula to explore the relationship between multiple variables and parameter n values in different scenarios, that is, the joint influence of multiple variables on the value size and change of parameter n value; 2) Using the joint probability size between multiple variables and parameter n value, select the key variables that have the greatest influence on the numerical change of model parameters; 3) Constructing the Vine series tree structure, stripping a single variable from the mutual relationship between multiple variables, and obtaining the independent influence and influence degree of the variable on the parameter n value. The method of the present invention can analyze how multiple variables affect the numerical size of parameter n, and can more accurately give the key factors affecting the parameter n value.
[0083] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0084] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of the present invention. At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.
Claims
1. A method for evaluating the influencing factors of Budyko key parameter changes, characterized in that: include: Construct the input data set of Choudhury-Yang equation in Budyko model and the variable data set that affects the model parameter values; The key parameter n of the model is estimated based on the Choudhury-Yang equation in the Budyko model, and the distribution pattern of the key parameter n value in the basin is quantified; Using a multidimensional Copula model to quantify the joint impact of i variables in the variable data set on the change of key parameter n, and selecting j key variables that have the greatest impact on the change of the value of key parameter n; A C-Vine series tree structure is established for the key variables, the key parameter n value is used as the central node of the T1 tree, and the Copula function is used to establish the correlation between the key parameter n value and each key variable; Based on the correlation between the key parameter n value and each key variable, a Vine tree from T'2 to T"j layers is constructed to obtain the correlation between the key parameter n and each variable under different conditions; Based on the correlation between the key parameter n and each key variable under different conditions, a weight factor is introduced to construct a multiple linear regression model, and the independent influence and influence degree of the key variable on the key parameter n value are determined according to the multiple linear regression model.
2. The method for evaluating the influencing factors of Budyko key parameter changes according to claim 1, characterized in that: The input data set includes input data used for parameter estimation of the Budyko model and environmental variables that affect the numerical changes of model parameters, wherein the input data used for parameter estimation of the Budyko model includes precipitation pre, actual evapotranspiration ET and potential evapotranspiration PET.
3. The method for evaluating the influencing factors of Budyko key parameter changes according to claim 2, characterized in that: The formula of the Choudhury-Yang equation is: Among them, n is the key parameter of the model representing the characteristics of the watershed.
4. The method for evaluating the influencing factors of Budyko key parameter changes according to claim 1, characterized in that: A multidimensional Copula model is used to quantify the joint impact of i variables in the variable data set on the change of key parameter n, and j key variables with the greatest impact on the change of the value of key parameter n are selected, including: Select the marginal distribution function for each variable; Optimal parameter estimation is performed for the selected marginal distribution function so that the cumulative distribution function (CDF) of any variable marginal distribution is uniformly distributed in the range of [0, 1]; Determine the Copula function; the formula of the Copula function is: C(u)=Φ(Φ -1 (u1),Φ -1 (u2),...,Φ -1 (u d )); where Φ represents the standard normal distribution function, Φ -1 Represents the inverse function of the standard normal distribution function, u d represents the probability distribution function of the d-th random variable, C(u) is the Copula function; Input a multivariate Copula function parameter set, calculate and output the joint CDF of the parameters under the Copula function; The environmental variables in the combination with the largest joint CDF are identified as the key variables that have the greatest impact on the numerical changes of the key parameter n.
5. The method for evaluating the influencing factors of Budyko key parameter changes according to claim 1, characterized in that: The Copula function is used to establish the correlation between the key parameter n value and each key variable, including: After determining the C-Vine series tree structure, estimate the binary Copula functions of all edges in the T1 tree, calculate the AIC information criterion for all selected Copula functions, and select the Copula function with the smallest AIC result; According to the order of the C-Vine series tree structure, the maximum likelihood method is used to estimate the parameters of the Copula function with the smallest AIC result; the correlation between the key parameter n value and each key variable is represented by each edge on the Vine tree in the C-Vine series tree structure.
6. The method for evaluating the influencing factors of Budyko key parameter changes according to claim 5, characterized in that: The binary Copula functions include Frank Copula, Clayton Copula and Gumbel Copula.
7. The method for evaluating the influencing factors of Budyko key parameter changes according to claim 5, characterized in that: The calculation formula of the AIC result is: AIC=2k+2ln(L); wherein k is the number of parameters and L is the likelihood function.
Citation Information
Patent Citations
Regional evapotranspiration change attribution analysis method considering multi-element spatial dependency
CN114462518A
Method for identifying and controlling associated risk of river and lake water system communication
CN115115269A
Runoff attribution method based on watershed hydrothermal coupling Buddyko framework
CN115809561A
