A calculation method for the impact of environmental changes on the probability of drought transmission

Abstract

The present invention discloses a method for calculating the impact of environmental changes on the probability of drought transmission. By using a conditional probability model of a non-consistent meteorological drought index (NSPI) with a large-scale climate factor (CI) as a covariate and a non-consistent hydrological drought index (NSRI) with a meteorological variable as a covariate, the method calculates the probability of different NSPI values under changing environmental conditions being converted into different NSRI drought levels in the forecast period. By comparing the drought transmission probabilities calculated based on the consistent drought index, the impact of the changing environment on the drought transmission probability is analyzed. The method is applicable to hydrological drought prediction under a changing environment, further improving the calculation accuracy and reliability of the drought transmission probability.

Description

Technical Field

[0001] The present invention relates to the field of drought prediction, and in particular to a method for calculating the impact of environmental changes on drought transmission probability. Background Art

[0002] The frequent occurrence of drought disasters has profound impacts on socioeconomic and ecological systems. Meteorological drought often occurs first in a drought event, and insufficient precipitation then develops into hydrological or agricultural drought through the hydrological cycle. This evolution from one drought form to another is called drought transmission. However, under the influence of a changing environment, the transmission process of drought within a watershed may be altered, requiring traditional drought transmission analysis methods to further consider the impact of this changing environment.

[0003] In recent years, research on drought transmission has primarily focused on the correlation between different types of drought indices and the joint distribution characteristics of duration and intensity across different drought types, aiming to determine the response time and lag time between different drought types. Some researchers have used mathematical statistical models such as Markov chains and multivariate normal distribution models to estimate the transition probability from current drought levels to future drought levels. However, these methods calculate the transmission probability between drought levels and are therefore less accurate. Furthermore, the drought transmission process within a watershed may be altered by a changing environment. This impact of environmental change on drought transmission needs to be considered, and further improvements should be made to the calculation methods for drought transmission probability to make them applicable to drought transmission analysis under changing environments. Summary of the Invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a method for calculating the impact of environmental changes on the probability of drought transmission. By comparing the drought transmission probability calculated based on the consistency drought index, the impact of environmental changes on the probability of drought transmission is analyzed. This method can be applied to hydrological drought prediction under changing environments, further improving the calculation accuracy and reliability of the drought transmission probability.

[0005] Technical solution: The method for calculating the impact of environmental changes on the probability of drought transmission described in the present invention specifically includes the following steps:

[0006] S1: Use the Pettitt test method to analyze the precipitation sequence x=(x1,x2,…x n ) and runoff sequence y=(y1,y2,…y n ) to conduct inconsistency test, and combine the Mann-Kendall test method to obtain the trend change of the sample sequence;

[0007] S2: Calculate the cumulative precipitation series x within the time scale ω ω , select the cumulative precipitation sequence x within the time scale of ω in the basinω Large-scale climate factors with significant correlations CI;

[0008] S3: According to the cumulative precipitation sequence x of the basin ω , using the GAMLSS model to calculate the cumulative precipitation series x ω The distribution of is fitted and the non-uniform meteorological drought index NSPI is calculated with the large-scale climate factor CI as the covariate;

[0009] S4: Based on the meteorological variables of basin precipitation, evaporation, relative humidity and wind speed, the non-uniform hydrological drought index (NSRI) is calculated with meteorological variables as covariates;

[0010] S5: The normality test of NSPI and NSRI sequences was performed using the Kolmogorov-Smirnov KS test method;

[0011] S6: Construct a conditional probability model based on the NSPI and NSRI series to calculate the drought transmission probability of the current NSPI value converted into NSRI levels in different forecast periods considering the changing environment;

[0012] S7: Based on the consistent drought index SPI series and SRI series, the probability of converting the SPI value to the SRI level in different forecast periods without considering the changing environment is calculated;

[0013] S8: Compare the drought transmission probability of different forecast periods converted from the current NSPI value considering the changing environment in S6 to the NSRI level and the probability of different forecast periods converted from the SPI value without considering the changing environment in S7 to obtain the impact value of the changing environment on the drought transmission probability.

[0014] As a preference, the Pettitt test method is used in S1 to analyze the precipitation sequence x=(x1, x2, ... x n )The formula for inconsistency test is as follows:

[0015]

[0016]

[0017]

[0018] Among them U t,n is the test statistic, which indicates the cumulative number of values at time t that are greater than or less than the value at time i; t0 is the mutation point;

[0019] Similarly, the Pettitt test method is used to analyze the runoff sequence y=(y1,y2,…y n)The formula for inconsistency test is as follows:

[0020]

[0021]

[0022]

[0023] Preferably, the large-scale climate factor CI in S2 is selected from the sea temperature anomaly Nino3.4SST, Arctic Oscillation Index AO, Pacific Decadal Oscillation Index PDO and North Atlantic Oscillation Index NAO. Specifically, the Pearson correlation test is used to select the cumulative precipitation sequence x under the ω time scale. ω The large-scale climate factors with good correlation are used as covariates to analyze the large-scale climate factors and the cumulative precipitation series x. ω Teleconnection relationship, using Pearson correlation index λ xc Characterize the cumulative precipitation series x ω The correlation with the large-scale climate factor CI is as follows:

[0024]

[0025] Preferably, the calculation process of the non-uniform meteorological drought index NSPI using the large-scale climate factor CI as a covariate in S3 is as follows:

[0026] The cumulative precipitation series x in the time scale ω ω When the Gamma distribution is satisfied, the cumulative precipitation sequence x under this time scale is ω The cumulative probability is:

[0027]

[0028] The scale parameter α and location parameter β here are respectively treated as covariates by p and q large-scale climate factors CI. Based on the GAMLSS framework, the probability density function G of these two parameters can be calculated. x Fit the distribution of (x|α,β):

[0029] g1(α)=α 10 +α 11 *CI1+α 12 *CI2+…+α 1p *CI p

[0030] g2(β)=α 20 +α 21 *CI1+α 22 *CI2+…+α 2q *CIq

[0031] Where g1(α) is a link function, which is determined by the domain of the statistical parameter. That is, if the domain of the distribution parameter α is α∈R, then the link function is g1(α)=α. If the distribution parameter α>0, g1(α)=lnα.

[0032] α ij (i=1,2j=0,1,2,…,p or q) are parameter coefficients. RS algorithm is used to estimate the parameter coefficients and model residuals. The model residuals are analyzed to see whether they approximately obey the normal distribution. The AIC criterion is used to select the best fitting distribution:

[0033]

[0034] AIC=ln(MSE)+2m

[0035] Where n is the sample length, Pe i and P i are the empirical frequency and theoretical frequency of the sample, respectively, and m is the number of model parameters. The smaller the AIC value, the better the function fit, and then the NSPI can be calculated:

[0036] When 0<F(x w )≤0.5:

[0037]

[0038]

[0039] When 0.5<F(x w )≤1:

[0040]

[0041]

[0042] Among them, c0=2.515517; c1=0.802853; c2=0.010328; d1=1.4132788; d2=0.189269; d3=0.001308.

[0043] As an example, the calculation process of the non-uniform hydrological drought index NSRI in S4 with the meteorological variable as the covariate is as follows: the cumulative runoff sequence y in the ω time scale ω When the Gamma distribution is satisfied, the cumulative runoff sequence y under this time scale is ω The cumulative probability is:

[0044]

[0045] Assuming that the scale parameter α and location parameter β of the Gamma distribution are functions of meteorological variables: precipitation x, evaporation PET, relative humidity w and wind speed v as covariates, the probability density function G of these two parameters can be calculated based on the GAMLSS framework. y Fit the distribution of (y|α,β):

[0046] g1(α)=α 10 +α 11 *x+α 12 *PET+α 13 *w+α 14 *v

[0047] g2(α)=α 20 +α 21 *x+α 22 *PET+α 23 *w+α 24 *v

[0048] Where g1(α) is a link function, which is determined by the domain of the statistical parameter. That is, if the domain of the distribution parameter α is α∈R, then the link function is g1(α)=α. If the distribution parameter α>0, g1(α)=lnα.

[0049] α ij (i=1,2j=0,1,2,…,p or q) are parameter coefficients. The RS algorithm is used to estimate the parameter coefficients and model residuals. The model residuals are analyzed to see whether they approximately obey the normal distribution. The AIC criterion is used to screen the best fitting distribution and then calculate the NSRI:

[0050] When 0<F(y w )≤0.5:

[0051]

[0052]

[0053] When 0.5<F(y w )≤1:

[0054]

[0055]

[0056] Among them, c0=2.515517; c1=0.802853; c2=0.010328; d1=1.4132788; d2=0.189269; d3=0.001308.

[0057] Preferably, the normality test of the NSPI sequence and the NSRI sequence is performed in S5, specifically as follows:

[0058] For a known NSPI sequence Z(x i ), the definition of the test statistic D of the Kolmogorov-Smirnov test method is given as follows:

[0059] D n =max|N(x i )-Z n (x i )|

[0060] Where n is the number of samples, and the empirical distribution curve function of the samples is Z n (x i ), N(x i ) represents the normal distribution curve function. If the significance level a corresponds to the critical value of the test statistic D a Greater than the test statistic D n , then the sample data satisfies the normal distribution, otherwise the sample data does not satisfy the normal distribution;

[0061] Similarly, for a known NSRI sequence Z(y i ), the definition of the test statistic D of the Kolmogorov-Smirnov test method is given as follows:

[0062] D n =max|N(y i )-Z n (y i )|

[0063] Where n is the number of samples, and the empirical distribution curve function of the samples is Z n (y i ), N(y i ) represents the normal distribution curve function. If the significance level a corresponds to the critical value of the test statistic D a Greater than the test statistic D n , then the sample data satisfies the normal distribution, otherwise the sample data does not satisfy the normal distribution.

[0064] Preferably, the specific steps of S6 are as follows:

[0065] When the current meteorological drought NSPI value and the future hydrological drought NSRI drought level satisfy a bivariate normal distribution, the probability of the current NSPI condition transferring to the future NSRI drought level can be deduced as follows:

[0066]

[0067] in Indicates the NSPI value of the current λ month, Identify the NSRI value of λ+M months, C Ms and C Mi Drought level C M The upper and lower limits of the NSPI value are given by the correlation coefficient ρ between the current NSPI value and the future NSRI value. ρ can be calculated as follows:

[0068]

[0069] When the cumulative precipitation series x ω and the cumulative runoff series y ω When all satisfy the normal distribution, after the standardization process, the cumulative precipitation sequence x ω Corresponding NSPI value and the cumulative runoff series y ω Corresponding NSRI value All obey the standard normal distribution, that is:

[0070]

[0071] It is expressed as the covariance of the current NSPI and the NSRI lagged M months later, and is calculated as follows:

[0072]

[0073] Preferably, S7 uses the same conditional probability model as in S6 to calculate the probability of converting the SPI value into SRI levels in different forecast periods without considering the changing environment.

[0074] Preferably, in S8, the results of S7 and S6 are compared to obtain the effect of the changing environment on the probability of drought transmission, as follows:

[0075]

[0076] Where ξ represents the impact of changing environment on the probability of drought transmission, P N To consider the drought transmission probability under the influence of changing environment, P S is the drought transmission probability without considering the influence of the changing environment. The larger the absolute value of ξ, the greater the influence of the changing environment on the drought transmission probability.

[0077] Beneficial effects: This application takes into account the impact of a changing environment on drought transmission, applies the non-uniform drought index to the conditional probability model, further improves the calculation accuracy and credibility of the drought transmission probability, and calculates the drought transmission probability without considering the changing environment, analyzes the impact of the changing environment on the drought transmission probability, and this method can be applied to hydrological drought prediction under a changing environment, further improving the calculation accuracy and credibility of the drought transmission probability. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 It is a flowchart of this application. DETAILED DESCRIPTION

[0079] The present application will be further described below with reference to specific embodiments.

[0080] like Figure 1 FIG. 1 is a flow chart of the present application, which specifically includes the following steps in this embodiment:

[0081] S1: Use the Pettitt test method to analyze the precipitation sequence x=(x1,x2,…x n ) and runoff sequence y=(y1,y2,…y n ) to perform inconsistency test, and combine the Mann-Kendall test method to obtain the trend change of the sample sequence, specifically:

[0082] The Pettitt test method is used to analyze the precipitation sequence x=(x1,x2,…x n )The formula for inconsistency test is as follows:

[0083]

[0084]

[0085]

[0086] Among them U t,n is the test statistic, which indicates the cumulative number of times the value at time t is greater than or less than the value at time i; t0 is the mutation point;

[0087] Similarly, the Pettitt test method is used to analyze the runoff sequence y=(y1,y2,…y n )The formula for inconsistency test is as follows:

[0088]

[0089]

[0090]

[0091] S2: Calculate the cumulative precipitation series x within the time scale ω ω , select the cumulative precipitation sequence x within the time scale of ω in the basin ω Large-scale climate factors with significant correlations CI;

[0092] The large-scale climate factor CI is selected from the sea temperature anomaly Nino3.4SST, Arctic Oscillation Index AO, Pacific Decadal Oscillation Index PDO and North Atlantic Oscillation Index NAO. Specifically, the Pearson correlation test is used to select the cumulative precipitation series x on the ω time scale. ω The large-scale climate factors with good correlation are used as covariates to analyze the large-scale climate factors and the cumulative precipitation series x. ω Teleconnection relationship, using Pearson correlation index λ xc Characterize the cumulative precipitation series x ω The correlation with the large-scale climate factor CI is as follows:

[0093]

[0094] S3: According to the cumulative precipitation sequence x of the basin ω , using the GAMLSS model to calculate the cumulative precipitation series x ω The distribution of is fitted and the non-uniform meteorological drought index NSPI is calculated with the large-scale climate factor CI as the covariate, as follows:

[0095] The cumulative precipitation series x in the time scale ω ω When the Gamma distribution is satisfied, the cumulative precipitation sequence x under this time scale is ω The cumulative probability is:

[0096]

[0097] The scale parameter α and location parameter β here are respectively treated as covariates by p and q large-scale climate factors CI. Based on the GAMLSS framework, the probability density function G of these two parameters can be calculated. x Fit the distribution of (x|α,β):

[0098] g1(α)=α 10 +α 11 *CI1+α 12 *CI2+…+α 1p *CI p

[0099] g2(β)=α 20 +α 21 *CI1+α 22 *CI2+…+α 2q *CI q

[0100] Where g1(α) is a link function, which is determined by the domain of the statistical parameter. That is, if the domain of the distribution parameter α is α∈R, then the link function is g1(α)=α. If the distribution parameter α>0, g1(α)=lnα.

[0101] α ij (i=1,2j=0,1,2,…,p or q) are parameter coefficients. RS algorithm is used to estimate the parameter coefficients and model residuals. The model residuals are analyzed to see whether they approximately obey the normal distribution. The AIC criterion is used to select the best fitting distribution:

[0102]

[0103] AIC=ln(MSE)+2m

[0104] Where n is the sample length, Pe i and P i are the empirical frequency and theoretical frequency of the sample, respectively, and m is the number of model parameters. The smaller the AIC value, the better the function fit, and then the NSPI can be calculated:

[0105] When 0<F(x w )≤0.5:

[0106]

[0107]

[0108] When 0.5<F(x w )≤1:

[0109]

[0110]

[0111] Among them, c0=2.515517; c1=0.802853; c2=0.010328; d1=1.4132788; d2=0.189269; d3=0.001308.

[0112] S4: Based on the meteorological variables of basin precipitation, evaporation, relative humidity and wind speed, the non-uniform hydrological drought index (NSRI) is calculated with meteorological variables as covariates, as follows:

[0113] The cumulative runoff series y within the time scale ω ω When the Gamma distribution is satisfied, the cumulative runoff sequence y under this time scale is ω The cumulative probability is:

[0114]

[0115] Assuming that the scale parameter α and location parameter β of the Gamma distribution are functions of meteorological variables: precipitation x, evaporation PET, relative humidity w and wind speed v as covariates, the probability density function G of these two parameters can be calculated based on the GAMLSS framework. y Fit the distribution of (y|α,β):

[0116] g1(α)=α 10 +α 11 *x+α 12 *PET+α 13 *w+α 14 *v

[0117] g2(α)=α 20 +α 21 *x+α 22 *PET+α 23 *w+α 24 *v

[0118] Where g1(α) is a link function, which is determined by the domain of the statistical parameter. That is, if the domain of the distribution parameter α is α∈R, then the link function is g1(α)=α. If the distribution parameter α>0, g1(α)=lnα.

[0119] α ij (i=1,2j=0,1,2,…,p or q) are parameter coefficients. The RS algorithm is used to estimate the parameter coefficients and model residuals. The model residuals are analyzed to see whether they approximately obey the normal distribution. The AIC criterion is used to screen the best fitting distribution and then calculate the NSRI:

[0120] When 0<F(y w )≤0.5:

[0121]

[0122]

[0123] When 0.5<F(y w )≤1:

[0124]

[0125]

[0126] Among them, c0=2.515517; c1=0.802853; c2=0.010328; d1=1.4132788; d2=0.189269; d3=0.001308.

[0127] S5: The Kolmogorov-Smirnov KS test method is used to test the normality of NSPI sequences and NSRI sequences. For a known NSPI sequence Z(x i ), the definition of the test statistic D of the Kolmogorov-Smirnov test method is given as follows:

[0128] D n =max|N(x i )-Z n (x i )|

[0129] Where n is the number of samples, and the empirical distribution curve function of the samples is Z n (x i ), N(x i ) represents the normal distribution curve function. If the significance level a corresponds to the critical value of the test statistic D a Greater than the test statistic D n , then the sample data satisfies the normal distribution, otherwise the sample data does not satisfy the normal distribution;

[0130] Similarly, for a known NSRI sequence Z(y i ), the definition of the test statistic D of the Kolmogorov-Smirnov test method is given as follows:

[0131] D n =max|N(y i )-Z n (y i )|

[0132] Where n is the number of samples, and the empirical distribution curve function of the samples is Z n (y i ), N(y i ) represents the normal distribution curve function. If the significance level a corresponds to the critical value of the test statistic D a Greater than the test statistic D n , then the sample data satisfies the normal distribution, otherwise the sample data does not satisfy the normal distribution.

[0133] S6: Construct a conditional probability model based on the NSPI and NSRI series to calculate the drought transmission probability of the current NSPI value into NSRI levels of different forecast periods, taking into account the changing environment. The specific calculation process is as follows:

[0134] When the current meteorological drought NSPI value and the future hydrological drought NSRI drought level satisfy a bivariate normal distribution, the probability of the current NSPI condition transferring to the future NSRI drought level can be deduced as follows:

[0135]

[0136] in Indicates the NSPI value of the current λ month, Identify the NSRI value of λ+M months, C Ms and C Mi Drought level C M The upper and lower limits of the NSPI value are given by the correlation coefficient ρ between the current NSPI value and the future NSRI value. ρ can be calculated as follows:

[0137]

[0138] When the cumulative precipitation series x ω and the cumulative runoff series y ω When all satisfy the normal distribution, after the standardization process, the cumulative precipitation sequence x ω Corresponding NSPI value and the cumulative runoff series y ω Corresponding NSRI value All obey the standard normal distribution, that is:

[0139]

[0140] It is expressed as the covariance of the current NSPI and the NSRI lagged M months later, and is calculated as follows:

[0141]

[0142] S7: Based on the consistent drought index SPI series and SRI series, the same conditional probability model as in S6 is used to calculate the probability of the SPI value being converted into the SRI level in different forecast periods without considering the changing environment;

[0143] S8: Compare the drought transmission probabilities of different forecast periods converted from the current NSPI values taking into account the changing environment in S6 and the probabilities of different forecast periods converted from the SPI values taking into account the changing environment in S7 to the SRI levels, and obtain the impact of the changing environment on the drought transmission probability, as follows:

[0144]

[0145] Where ξ represents the impact of changing environment on the probability of drought transmission, P N To consider the drought transmission probability under the influence of changing environment, P S is the drought transmission probability without considering the influence of the changing environment. The larger the absolute value of ξ, the greater the influence of the changing environment on the drought transmission probability.

[0146] Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of the present invention.

Claims

1. A method for calculating the impact of environmental changes on the probability of drought transmission, characterized by: The specific steps include: S1: Use the Pettitt test method to analyze the precipitation sequence x=(x1,x2,…x n ) and runoff sequence y=(y1,y2,…y n ) to conduct inconsistency test, and combine the Mann-Kendall test method to obtain the trend change of the sample sequence; S2: Calculate the cumulative precipitation series x within the time scale ω ω , select the cumulative precipitation sequence x within the time scale of ω in the basin ω Large-scale climate factors with significant correlations CI; S3: According to the cumulative precipitation sequence x of the basin ω , using the GAMLSS model to calculate the cumulative precipitation series x ω The distribution of is fitted and the non-uniform meteorological drought index NSPI is calculated with the large-scale climate factor CI as the covariate; S4: Based on the meteorological variables of basin precipitation, evaporation, relative humidity and wind speed, the non-uniform hydrological drought index (NSRI) is calculated with meteorological variables as covariates; S5: The normality test of NSPI and NSRI sequences was performed using the Kolmogorov-Smirnov KS test method; S6: Construct a conditional probability model based on the NSPI and NSRI series to calculate the drought transmission probability of the current NSPI value into different NSRI levels in the forecast period, taking into account the changing environment. The specific steps are as follows: When the current meteorological drought NSPI value and the future hydrological drought NSRI drought level satisfy a bivariate normal distribution, the probability of the current NSPI condition transferring to the future NSRI drought level can be deduced as follows: in Indicates the NSPI value of the current λ month, Identify the NSRI value of λ+M months, C Ms and C Mi Drought level C M The upper and lower limits of the NSPI value are given by the correlation coefficient ρ between the current NSPI value and the future NSRI value. ρ can be calculated as follows: When the cumulative precipitation series x ω and the cumulative runoff series y ω When all satisfy the normal distribution, after the standardization process, the cumulative precipitation sequence x ω Corresponding NSPI value and the cumulative runoff series y ω Corresponding NSRI value All obey the standard normal distribution, that is: It is expressed as the covariance of the current NSPI and the NSRI lagged M months later, and is calculated as follows: S7: Based on the consistent drought index SPI series and SRI series, the probability of converting the SPI value to the SRI level in different forecast periods without considering the changing environment is calculated; S8: Compare the drought transmission probability of different forecast periods converted from the current NSPI value considering the changing environment in S6 to the NSRI level and the probability of different forecast periods converted from the SPI value without considering the changing environment in S7 to obtain the impact value of the changing environment on the drought transmission probability.

2. The method for calculating the impact of environmental changes on drought transmission probability according to claim 1, characterized in that: In S1, the Pettitt test method is used to analyze the precipitation sequence x=(x1, x2, ...x n )The formula for inconsistency test is as follows: Among them U t,n is the test statistic, which indicates the cumulative number of values at time t that are greater than or less than the value at time i; t0 is the mutation point; Similarly, the Pettitt test method is used to analyze the runoff sequence y=(y1,y2,…y n )The formula for inconsistency test is as follows:

3. The method for calculating the impact of environmental changes on drought transmission probability according to claim 1, characterized in that: The large-scale climate factor CI in S2 is selected from the sea temperature anomaly Nino3.4SST, Arctic Oscillation Index AO, Pacific Decadal Oscillation Index PDO and North Atlantic Oscillation Index NAO. Specifically, the Pearson correlation test is used to select the cumulative precipitation series x on the ω time scale. ω The large-scale climate factors with good correlation are used as covariates to analyze the large-scale climate factors and the cumulative precipitation series x. ω Teleconnection relationship, using Pearson correlation index λ xc Characterize the cumulative precipitation series x ω The correlation with the large-scale climate factor CI is as follows:

4. The method for calculating the impact of environmental changes on drought transmission probability according to claim 1, characterized in that: The calculation process of the non-uniform meteorological drought index NSPI in S3 with the large-scale climate factor CI as a covariate is as follows: The cumulative precipitation series x in the time scale ω ω When the Gamma distribution is satisfied, the cumulative precipitation sequence x under this time scale is ω The cumulative probability is: The scale parameter α and location parameter β here are respectively treated as covariates by p and q large-scale climate factors CI. Based on the GAMLSS framework, the probability density function G of these two parameters can be calculated. x Fit the distribution of (x|α,β): g1(a)=a 10 +a 11 *CI1+a 12 *CI2+…+a 1p *CI p g2(β)=α 20 +a 21 *CI1+a 22 *CI2+…+a 2q *CI q Where g1(α) is a link function, which is determined by the domain of the statistical parameter. That is, if the domain of the distribution parameter α is α∈R, then the link function is g1(α)=α. If the distribution parameter α>0, g1(α)=lnα. α ij (i=1,2j=0,1,2,…,p or q) are parameter coefficients. RS algorithm is used to estimate the parameter coefficients and model residuals. The model residuals are analyzed to see whether they approximately obey the normal distribution. The AIC criterion is used to select the best fitting distribution: Where n is the sample length, Pe i and P i are the empirical frequency and theoretical frequency of the sample, respectively, and m is the number of model parameters. The smaller the AIC value, the better the function fit, and then the NSPI can be calculated: When 0 <F(x w )≤0.5: When 0.5<F(x w )≤1: Among them, c0=2.515517; c1=0.802853; c2=0.010328; d1=1.4132788; d2=0.189269; d3=0.001308.

5. The method for calculating the impact of environmental changes on drought transmission probability according to claim 1, characterized in that: The calculation process of the non-uniform hydrological drought index NSRI in S4 with the meteorological variables as covariates is as follows: the cumulative runoff series y in the ω time scale ω When the Gamma distribution is satisfied, the cumulative runoff sequence y under this time scale is ω The cumulative probability is: Assuming that the scale parameter α and location parameter β of the Gamma distribution are functions of meteorological variables: precipitation x, evaporation PET, relative humidity w and wind speed v as covariates, the probability density function G of these two parameters can be calculated based on the GAMLSS framework. y Fit the distribution of (y|α,β): g1(a)=a 10 +a 11 *x+a 12 *PET+a 13 *w+a 14 *v g2(a)=a 20 +a 21 *x+a 22 *PET+a 23 *w+a 24 *v Where g1(α) is a link function, which is determined by the domain of the statistical parameter. That is, if the domain of the distribution parameter α is α∈R, then the link function is g1(α)=α. If the distribution parameter α>0, g1(α)=lnα. α ij (i=1,2j=0,1,2,…,p or q) are parameter coefficients. The RS algorithm is used to estimate the parameter coefficients and model residuals. The model residuals are analyzed to see whether they approximately obey the normal distribution. The AIC criterion is used to screen the best fitting distribution and then calculate the NSRI: When 0 <F(y w )≤0.5: When 0.5 <F(y w )≤1: Among them, c0=2.515517; c1=0.802853; c2=0.010328; d1=1.4132788; d2=0.189269; d3=0.001308.

6. The method for calculating the impact of environmental changes on drought transmission probability according to claim 1, characterized in that: In S5, the normality test of the NSPI sequence and the NSRI sequence is performed as follows: For a known NSPI sequence Z(x i ), the definition of the test statistic D of the Kolmogorov-Smirnov test method is given as follows: D n =max|N(x i )-Z n (x i )| Where n is the number of samples, and the empirical distribution curve function of the samples is Z n (x i ), N(x i ) represents the normal distribution curve function. If the significance level a corresponds to the critical value of the test statistic D a Greater than the test statistic D n , then the sample data satisfies the normal distribution, otherwise the sample data does not satisfy the normal distribution; Similarly, for a known NSRI sequence Z(y i ), the definition of the test statistic D of the Kolmogorov-Smirnov test method is given as follows: D n =max|N(y i )-Z n (y i )| Where n is the number of samples, and the empirical distribution curve function of the samples is Z n (y i ), N(y i ) represents the normal distribution curve function. If the significance level a corresponds to the critical value of the test statistic D a Greater than the test statistic D n , then the sample data satisfies the normal distribution, otherwise the sample data does not satisfy the normal distribution.

7. The method for calculating the impact of environmental changes on drought transmission probability according to claim 1, characterized in that: S7 uses the same conditional probability model as in S6 to calculate the probability of converting the SPI value into the SRI level of different forecast periods without considering the changing environment.

8. The method for calculating the impact of environmental changes on drought transmission probability according to claim 1, characterized in that: In S8, the results of S7 and S6 are compared to obtain the effect of the changing environment on the probability of drought transmission, which is as follows: Where ξ represents the impact of changing environment on the probability of drought transmission, P N To consider the drought transmission probability under the influence of changing environment, P S is the drought transmission probability without considering the influence of the changing environment. The larger the absolute value of ξ, the greater the influence of the changing environment on the drought transmission probability.

Images