Vegetation index time series data prediction method based on VAR model and climate change driving

Abstract

The invention discloses a vegetation index time series data prediction method based on a VAR model and climate change driving, and belongs to the technical field of geographic information, and the method comprises the steps: determining a research region, vegetation index data, meteorological data and landform data; taking each element data as an influence component, processing the component influencing the vegetation index by adopting geographic detector factor detection, screening main influence factors, and determining prediction model parameters; and predicting the vegetation index through the VAR model, and outputting a final vegetation index prediction value. According to the method, a vegetation index is used as a dependent variable, meteorological data and landform data are used as independent variables, a geographic detector is used for screening important influence factors, and time sequence characteristics and spatial heterogeneity of spatial data are considered at the same time. A vegetation index prediction model is established according to different data characteristics, and the defect that first-order linear prediction has poor fitting performance on the vegetation index data prediction condition is overcome by using a VAR model, so that the prediction result is more in line with the actual condition of the vegetation index data.

Description

Technical Field

[0001] The present invention belongs to the field of geographic information technology, and in particular relates to a vegetation index time series data prediction method based on a VAR model and climate change driven. Background Art

[0002] Vegetation index is an effective numerical indicator to measure the status of surface vegetation. It is obtained through continuous WeChat remote sensing data time series and shows good correlation with meteorological elements and topographic data. For example, NDVI, as one of the current mainstream vegetation coverage calculation indices, can effectively evaluate the vegetation growth and coverage in the study area.

[0003] Traditional vegetation index data time series expansion methods mostly use regression analysis methods, which are suitable for processing data prediction within a fixed time period with first-order linear correlation. However, the correlation between vegetation index and meteorological elements at different time window scales will be different. The first-order linear correlation calculation may affect the correlation between the two types of elements to a certain extent, thereby affecting the accuracy of the predicted vegetation index data. At the same time, the regression analysis method is difficult to clearly reveal the non-first-order linear relationship between related elements. Therefore, it is particularly important to objectively select fitting indicators and establish a multivariate high-order correlation model to restore the true future prediction value. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the vegetation index time series data prediction method based on the VAR model and climate change driven provided by the present invention solves the problem that the existing related methods do not pay attention to the correlation between the vegetation index and meteorological elements at different time window scales, thereby affecting the accuracy of the prediction results.

[0005] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a vegetation index time series data prediction method based on a VAR model and climate change driven, comprising the following steps:

[0006] S1. Obtain the normalized vegetation index data, meteorological element data and topographic data of the study area and pre-process them;

[0007] S2, taking the pre-processed normalized vegetation index data as the dependent variable, the meteorological element data and the topographic data as the independent variables, and using the geographic detector factor detection to reduce the dimension of the influencing factors of the pre-processed vegetation index data, and then screen the main influencing factors of the vegetation index;

[0008] S3. The main influencing factors screened were used as model parameters to construct a vegetation index prediction model based on the VAR model and climate change driven;

[0009] S4. Predict the vegetation index of a specified year using the constructed vegetation index prediction model.

[0010] Furthermore, the preprocessing in step S1 includes:

[0011] The obtained NDVI data are synthesized into annual data using the maximum value synthesis method, and then the average NDVI value in the required time period is calculated using the arithmetic mean method;

[0012] The meteorological element data obtained are calculated using the arithmetic mean method to obtain the multi-year average value within the required time period;

[0013] For the acquired topographic data, the elevation data is used to calculate the slope aspect, surface undulation and slope data.

[0014] Furthermore, in step S2, the factor detector of the geographic detector is represented as:

[0015]

[0016] SST=Nσ 2

[0017] In the formula, q represents the explanatory power. The greater the explanatory power, the more significant the spatial heterogeneity. h and N represent the number of cells in layer h and the study area, respectively. and σ 2 denote the variance of the normalized difference vegetation index values ​​of layer h and the study area, respectively; L denotes the total number of layers; SSW and SST denote the intra-layer equation and the total variance of the whole area, respectively.

[0018] Furthermore, in step S2, the explanatory power q value and its significance calculated by the factor detector of the geographic detector are used to determine the explanatory power of the independent variable and the dependent variable, thereby screening the main influencing factors;

[0019] The calculation formula for the significance p of the explanatory power q value is:

[0020]

[0021] In the formula, F(·) indicates that the explanatory power q value satisfies the non-central distribution, λ indicates the non-central parameter, represents the mean of layer h.

[0022] Furthermore, the step S3 includes the following sub-steps:

[0023] S31. Standardize the main influencing factors screened;

[0024] S32, determining the data stability of the main influencing factors after the standardization processing, and processing the non-stationary main influencing factor data to obtain stable main factor impact data;

[0025] S33. Determine the optimal lag order of the VAR model and test its stability;

[0026] S34. Based on the stable main influencing factor data and the optimal lag order, a VAR model is constructed and used as a vegetation index prediction model.

[0027] Furthermore, in step S32, the unit root test method is used to determine the data stability of the main influencing factors, and for the non-stationary main influencing factor data, the Johansen statistical method is used to perform an overall cointegration test on the data sequence to obtain the stable main influencing factor data.

[0028] Furthermore, in step S33, the optimal lag order of the VAR model is determined by using the Akaike Information Criterion statistical test index, and its stability is tested by using the unit circle test method;

[0029] Among them, Akaike information criterion is expressed as:

[0030] AIC=2z-2ln(L)

[0031] Where z represents the number of parameters in the model and L represents the maximum likelihood function of the model.

[0032] Furthermore, in step S34, the constructed vegetation index prediction model is expressed as:

[0033]

[0034] Where V t represents the predicted value of the normalized vegetation index at time t, c represents the constant term, A i and B i Respectively represent the hysteresis coefficients of vegetation index and meteorological data / topographic data, V t-i and G t-i They represent the vegetation index and meteorological data / topographic data of lag i period respectively, ∈ t represents the error term.

[0035] Furthermore, in step S4, the vegetation index of a specified year is predicted by the constructed vegetation index prediction model to obtain standardized prediction data, which is then restored to obtain final vegetation index prediction data.

[0036] The beneficial effects of the present invention are:

[0037] (1) The present invention uses vegetation index as the dependent variable and meteorological data as the independent variable, uses geographic detectors to screen important influencing factors, and considers the temporal characteristics and spatial heterogeneity of spatial data. A vegetation index prediction model is established for different data characteristics to achieve future prediction of vegetation index;

[0038] (2) The VAR model used in the present invention overcomes the shortcoming of poor fitting of the first-order linear prediction to the vegetation index data prediction, making the prediction result more consistent with the actual situation of the vegetation index data.

[0039] (3) Based on full consideration of the spatial heterogeneity of spatial data, the present invention objectively selects influencing factors and then constructs a model to reveal the response of future vegetation cover changes to climate change. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a flow chart of the vegetation index time series data prediction method based on the VAR model and climate change driven provided by the present invention. DETAILED DESCRIPTION

[0041] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0042] The embodiment of the present invention provides a method for predicting vegetation index time series data based on a VAR model and climate change drive, such as Figure 1 As shown, the following steps are included:

[0043] S1. Obtain the normalized vegetation index data, meteorological element data and topographic data of the study area and pre-process them;

[0044] S2, taking the pre-processed normalized vegetation index data as the dependent variable, the meteorological element data and the topographic data as the independent variables, and using the geographic detector factor detection to reduce the dimension of the influencing factors of the pre-processed vegetation index data, and then screen the main influencing factors of the vegetation index;

[0045] S3. The main influencing factors screened were used as model parameters to construct a vegetation index prediction model based on the VAR model and climate change driven;

[0046] S4. Predict the vegetation index of a specified year using the constructed vegetation index prediction model.

[0047] The preprocessing in step S1 of the embodiment of the present invention includes:

[0048] The obtained Normalized Difference Vegetation Index (NDVI) data were synthesized into annual data using the maximum value synthesis method, and then the average Normalized Difference Vegetation Index value in the required time period was calculated using the arithmetic mean method;

[0049] The meteorological element data obtained are calculated using the arithmetic mean method to obtain the multi-year average value within the required time period; specifically, the meteorological element data include temperature, precipitation, and net long-wave radiation, etc.;

[0050] For the acquired topographic data, the elevation data is used to calculate the slope direction, surface undulation and slope data; specifically, the topographic data includes data such as elevation, slope direction and surface undulation.

[0051] In step S2 of the embodiment of the present invention, the study area is based on the clear and specific vector range of the study area, the normalized difference vegetation index (NDVI) is used as the dependent variable, the global land surface data assimilation dataset is used as the meteorological data input element and as the independent variable, the data includes 33 parameters such as net longwave radiation, snowfall rate, precipitation, etc., the above data are integrated in units of years and have a time correspondence with the dependent variable; DEM digital elevation model data is used as topographic input data and as the independent variable.

[0052] In step S2 of the embodiment of the present invention, the factor detector of the geographic detector is expressed as:

[0053]

[0054] SST=Nσ 2

[0055] In the formula, q represents the explanatory power. The greater the explanatory power, the more significant the spatial heterogeneity. h and N represent the number of cells in layer h and the study area, respectively. and σ 2 denote the variance of the normalized difference vegetation index values ​​of layer h and the study area, respectively; L denotes the total number of layers; SSW and SST denote the intra-layer equation and the total variance of the whole area, respectively.

[0056] In this embodiment, the explanatory power q value and its significance calculated by the factor detector of the geographic detector are used to determine the explanatory power of the independent variable and the dependent variable, and then the main influencing factors are screened;

[0057] The calculation formula for the significance p of the explanatory power q value is:

[0058]

[0059] In the formula, F(·) indicates that the explanatory power q value satisfies the non-central distribution, λ indicates the non-central parameter, represents the mean of layer h.

[0060] In the above process, the value range of q is [0,1]. The larger the q value, the stronger the explanatory power of the independent variable for the Normalized Difference Vegetation Index (NDVI), and vice versa.

[0061] In this embodiment, independent variables with explanatory power q>0.5 and significance p<0.05 are selected as the main influencing factors (M1, M2, ...) of the vegetation index.

[0062] Step S3 of the embodiment of the present invention includes the following sub-steps:

[0063] S31. Standardize the main influencing factors screened;

[0064] S32, determining the data stability of the main influencing factors after the standardization processing, and processing the non-stationary main influencing factor data to obtain stable main factor impact data;

[0065] S33. Determine the optimal lag order of the VAR model and test its stability;

[0066] S34. Based on the stable main influencing factor data and the optimal lag order, a VAR model is constructed and used as a vegetation index prediction model.

[0067] In step S31 of this embodiment, the Z-score standardization method is used to remove the inter-annual variation trend of the data, and the main influencing factors (M1, M2, ...) are standardized to G1, G2, ...; wherein, the calculation of the Z-score standardization method is as follows:

[0068]

[0069] Among them, G represents the standardized data; M represents the original data; μ represents the mean of the original data; σ represents the standard deviation of the original data.

[0070] In step S32 of this embodiment, the unit root test method is used to determine the data stability of the main influencing factors. For the non-stationary main influencing factor data, the Johansen statistical method is used to perform an overall cointegration test on the data sequence to obtain the stable main influencing factor data.

[0071] Specifically, the unit root test uses the ADF test method to test whether there is a unit root in the sequence. The radius of the unit circle is 1. If the reciprocals of all root modules are less than 1, the variable is stable; if the reciprocals of the root modules are greater than 1, that is, there are scattered points outside the unit circle, then the variable is unstable.

[0072] Corresponding to the non-stationary main influencing factor data, the Johansen statistical method is used to perform an overall cointegration test on the data series, that is, for k time series V t =(V 1t ,V 2t ,V 3t ,……,V t )′, t=1,2,…,T, the components of the k-dimensional vector time series are called d- and b-order cointegration.

[0073] In step S33 of this embodiment, the optimal lag order of the VAR model is determined by using the Akaike Information Criterion statistical test index, and its stability is tested by using the unit circle test method;

[0074] Among them, Akaike information criterion is expressed as:

[0075] AIC=2z-2ln(L)

[0076] Where z represents the number of parameters in the model and L represents the maximum likelihood function of the model.

[0077] The -2ln(L) in the above formula measures how well the model fits the data. That is, the larger the likelihood function value, the better the model fit. The second half, 2z, is the penalty term for the complexity of the model. The more parameters there are, the larger the penalty term.

[0078] When using the unit circle test method to test stability, it is tested whether the characteristic roots of all endogenous variables in the model fall within the unit circle, so as to prove whether the modulus of the characteristic roots of the system of equations is less than 1.

[0079] In step S34 of this embodiment, the constructed vegetation index prediction model is expressed as:

[0080]

[0081] Where V t represents the predicted value of the normalized vegetation index at time t, c represents the constant term, A i and B i Respectively represent the hysteresis coefficients of vegetation index and meteorological data / topographic data, V t-i and G t-i They represent the vegetation index and meteorological data / topographic data of lag i period respectively, ∈ t represents the error term.

[0082] In step S4 of the embodiment of the present invention, the vegetation index of the specified year is predicted by the constructed vegetation index prediction model to obtain standardized prediction data, which is restored to obtain the final vegetation index prediction data, which is expressed as:

[0083] Vt '=Gσ+μ

[0084] In the formula, G represents the standardized data, V t ' represents the predicted data after restoration, μ represents the mean of the original data, and σ represents the standard deviation of the original data.

[0085] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. 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 a limitation on the present invention.

[0086] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A vegetation index time series data prediction method based on VAR model and climate change driven, characterized by: The following steps are involved: S1. Obtain the normalized vegetation index data, meteorological element data and topographic data of the study area and pre-process them; S2, taking the pre-processed normalized vegetation index data as the dependent variable, the meteorological element data and the topographic data as the independent variables, and using the geographic detector factor detection to reduce the dimension of the influencing factors of the pre-processed vegetation index data, and then screen the main influencing factors of the vegetation index; S3. Taking the main influencing factors screened as model parameters, a vegetation index prediction model based on VAR model and climate change driven was constructed; S4. Predict the vegetation index of a specified year using the constructed vegetation index prediction model.

2. The vegetation index time series data prediction method based on VAR model and climate change driving according to claim 1 is characterized in that: The preprocessing in step S1 includes: The obtained NDVI data are synthesized into annual data using the maximum value synthesis method, and then the average NDVI value in the required time period is calculated using the arithmetic mean method; The meteorological element data obtained are calculated using the arithmetic mean method to obtain the multi-year average value within the required time period; For the acquired topographic data, the elevation data is used to calculate the slope aspect, surface undulation and slope data.

3. The vegetation index time series data prediction method based on VAR model and climate change driving according to claim 1 is characterized in that: In step S2, the factor detector of the geographic detector is represented as: SST=Nσ 2 In the formula, q represents the explanatory power. The greater the explanatory power, the more significant the spatial heterogeneity. h and N represent the number of cells in layer h and the study area, respectively. and σ 2 denote the variance of the normalized difference vegetation index values ​​of layer h and the study area, respectively; L denotes the total number of layers; SSW and SST denote the intra-layer equation and the total variance of the whole area, respectively.

4. The method for predicting vegetation index time series data based on VAR model and climate change drive according to claim 3 is characterized in that: In step S2, the explanatory power q value and its significance calculated by the factor detector of the geographic detector are used to determine the explanatory power of the independent variable and the dependent variable, and then the main influencing factors are screened; The calculation formula for the significance p of the explanatory power q value is: In the formula, F(·) indicates that the explanatory power q value satisfies the non-central distribution, λ indicates the non-central parameter, represents the mean of layer h.

5. The vegetation index time series data prediction method based on VAR model and climate change driving according to claim 1 is characterized in that: The step S3 comprises the following sub-steps: S31. Standardize the main influencing factors screened; S32, determining the data stability of the main influencing factors after the standardization processing, and processing the non-stationary main influencing factor data to obtain stable main factor impact data; S33. Determine the optimal lag order of the VAR model and test its stability; S34. Based on the stable main influencing factor data and the optimal lag order, a VAR model is constructed and used as a vegetation index prediction model.

6. The method for predicting vegetation index time series data based on VAR model and climate change drive according to claim 5 is characterized in that: In step S32, the unit root test method is used to determine the data stability of the main influencing factors. For the non-stationary main influencing factor data, the Johansen statistical method is used to perform an overall cointegration test on the data sequence to obtain the stable main influencing factor data.

7. The method for predicting vegetation index time series data based on VAR model and climate change drive according to claim 5 is characterized in that: In step S33, the optimal lag order of the VAR model is determined by using the Akaike Information Criterion statistical test index, and its stability is tested by using the unit circle test method; Among them, Akaike information criterion is expressed as: AIC=2z-2ln(L) Where z represents the number of parameters in the model and L represents the maximum likelihood function of the model.

8. The method for predicting vegetation index time series data based on VAR model and climate change drive according to claim 5 is characterized in that: In step S34, the constructed vegetation index prediction model is expressed as: Where V t represents the predicted value of the normalized vegetation index at time t, c represents the constant term, A i and B i Respectively represent the hysteresis coefficients of vegetation index and meteorological data / topographic data, V t-i and G t-i They represent the vegetation index and meteorological data / topographic data of lag i period respectively, ∈ t represents the error term.

9. The method for predicting vegetation index time series data based on VAR model and climate change drive according to claim 1, characterized in that: In step S4, the vegetation index of a specified year is predicted by the constructed vegetation index prediction model to obtain standardized prediction data, which is then restored to obtain final vegetation index prediction data.

Images