Evolution characteristics of hydrological elements and attribution analysis method of runoff change under changing environment

Abstract

The application discloses a method for analyzing evolution characteristics of hydrological elements and attribution of runoff change under changing environment, comprising the following steps: S1, selecting a monitoring area and collecting measured hydrological data in the area; S2, obtaining precipitation data and performing precision evaluation and correction by using the measured hydrological data; S3, analyzing evolution characteristics of the precipitation and runoff by using a fusion algorithm method; and S4, performing attribution analysis on runoff change of the monitoring area according to the analysis result. The application improves the reliability of the test result by relying on the trend and mutation point of the hydrological sequence, using multiple trend test methods and mutation test methods to comprehensively determine the trend and mutation point of the sequence, and improves the efficiency of hydrological prediction by analyzing the climate factors causing the runoff change from multiple angles.

Description

Technical Field

[0001] This invention relates to the field of hydrological evolution, and more specifically, to methods for analyzing the evolution characteristics of hydrological elements and the attribution of runoff changes under changing environments. Background Technology

[0002] Over the past century, the global natural environment has undergone tremendous changes, primarily characterized by a significant rise in temperature. Climate change has gradually become a hot topic of concern for all countries. To date, the Intergovernmental Panel on Climate Change (IPCC) of the United Nations has released six assessment reports. The most recent report indicates that global surface temperature has risen by 1.1°C compared to pre-industrial levels and may continue to rise by more than 1.5°C over the next 20 years. This shows that global warming is accelerating climate change and making it more severe. At the same time, climate warming is accelerating the hydrological cycle, changing the spatial and temporal distribution of precipitation, and thus affecting the evolution of river runoff. As a result, hydrological and meteorological elements such as temperature, precipitation, runoff, and evapotranspiration in different river basins have shown varying degrees of increase or decrease, and the frequency of extreme events has increased.

[0003] In addition to climate change, human activities are having an increasingly significant impact on the regional water cycle. The construction of water conservancy projects has altered the spatial and temporal distribution of water resources, and human activities have changed the nature of the underlying surface, directly affecting runoff generation and confluence processes. At the same time, human activities have changed the atmospheric composition, exacerbating the greenhouse effect and heat island effect, and indirectly affecting the regional water cycle process.

[0004] In some areas, the combined effects of climate change and human activities have led to severe floods and frequent droughts, resulting in a significant increase in average annual temperature, a slight increase in annual precipitation, and a slight decrease in annual runoff. Against this backdrop, a correct understanding of the evolution characteristics of hydrological elements is of great guiding significance for the scientific management of water resources and provides a scientific theoretical basis for the protection of the ecological environment.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] To address the problems in related technologies, this invention proposes a method for analyzing the evolution characteristics of hydrological elements and the attribution of runoff changes under changing environments, in order to overcome the aforementioned technical problems existing in existing related technologies.

[0007] Therefore, the specific technical solution adopted by the present invention is as follows:

[0008] The evolution characteristics of hydrological elements and the attribution analysis method for runoff changes under changing environments include the following steps:

[0009] S1. Select the monitoring area and collect measured hydrological data within that area;

[0010] S2. Acquire precipitation data and use measured hydrological data for accuracy evaluation and correction;

[0011] S3. The evolution characteristics of precipitation and runoff are analyzed using a fusion algorithm.

[0012] S4. Based on the analysis results, conduct attribution analysis on the runoff changes in the monitoring area.

[0013] Furthermore, the acquisition of precipitation data and the use of measured hydrological data for accuracy evaluation and correction include the following steps:

[0014] S21. Based on the measured precipitation data, the accuracy of the precipitation data is evaluated using five indicators: correlation coefficient, root mean square error, mean absolute error, Nash efficiency coefficient, and Kling-Gupta efficiency coefficient.

[0015] S22. Optimize the BP neural network using the sparrow search algorithm;

[0016] S23. The precipitation data is corrected using the optimized BP neural network.

[0017] Furthermore, the optimization of the BP neural network using the sparrow search algorithm includes the following steps:

[0018] S221. Initialize the population and related parameters, and calculate the fitness value of the initial population;

[0019] S222. Update the discoverer's location, calculated using the following formula:

[0020]

[0021] S223. Update the position of the new member, the calculation formula is:

[0022]

[0023] S224. Update the location of sparrows that are aware of danger. The calculation formula is:

[0024]

[0025] in, This represents the j-th dimension position information of the i-th sparrow in the t-th iteration;

[0026] iter max This represents the maximum number of iterations.

[0027] α is a random number between 0 and 1;

[0028] R2 is the warning value;

[0029] ST is a safe value;

[0030] Q is a random number that follows a normal distribution;

[0031] The worst individual in the t-th iteration;

[0032] X p This is the current position of the best discoverer;

[0033] f i This represents the current fitness value of the individual.

[0034] f g This represents the maximum fitness value of the current individual.

[0035] f w This represents the minimum fitness value of the current individual.

[0036] A is a 1×d matrix, with elements being randomly assigned values ​​of 1 or -1;

[0037] L is a 1×d matrix consisting entirely of 1s;

[0038] d is the dimension of the problem to be optimized;

[0039] β is a random number that follows a normal distribution with a mean of 0 and a variance of 1;

[0040] K is the current maximum fitness value;

[0041] It is the optimal individual in the t-th iteration;

[0042] n is the number of sparrows;

[0043] ε is a constant to avoid the denominator being zero.

[0044] Furthermore, the analysis of the evolution characteristics of precipitation and runoff using the fusion algorithm method includes the following steps:

[0045] S31. Analyze the evolution characteristics of hydrological elements within the monitoring area;

[0046] S32. Analyze the spatial evolution characteristics of precipitation within the monitoring area;

[0047] S33. Analyze the interannual variation characteristics of precipitation runoff in the monitored area;

[0048] S34. Analyze the annual distribution characteristics of precipitation runoff within the monitoring area.

[0049] Furthermore, the analysis of the evolution characteristics of hydrological elements within the monitoring area includes the following steps:

[0050] S311. The trend of hydrological elements in the monitoring area is assessed using trend analysis.

[0051] S312. Use the mutation analysis method to identify the mutation points of hydrological elements in the monitoring area;

[0052] S313. The periodicity of hydrological elements in the monitoring area is assessed using periodicity analysis.

[0053] S314. Predict the future trend of precipitation runoff in the monitored area using the recalibrated range analysis method;

[0054] S315. Analyze the annual distribution characteristics of precipitation and runoff within the monitoring area using multiple indicators.

[0055] Furthermore, the analysis of the spatial evolution characteristics of precipitation within the monitoring area includes the following steps:

[0056] S321. Obtain the distribution map of multi-year average precipitation and multi-year average seasonal precipitation in the monitoring area using the inverse distance weighted difference, and analyze the spatial evolution characteristics of precipitation;

[0057] S322. Use the inverse distance weighted difference to obtain the spatial distribution map of precipitation and seasonal precipitation in the monitoring area, and analyze the spatial evolution characteristics of precipitation trend.

[0058] Furthermore, the attribution analysis of runoff changes in the monitoring area based on the analysis results includes the following steps:

[0059] S41. Analyze the relationship between climate factors and runoff changes within the monitoring area using multiple algorithms;

[0060] S42. Further analyze the causal relationship between climate factors and runoff changes using Granger causality tests;

[0061] S43. Use a joint algorithm to analyze the relationship between climate change, human activities and runoff changes.

[0062] Furthermore, the analysis of the relationship between climate factors and runoff changes within the monitoring area using multiple algorithms includes the following steps:

[0063] S411. Calculate the correlation coefficient between runoff and a single climate factor using partial correlation analysis algorithm;

[0064] S412. Calculate the multiple correlation coefficient between runoff and climate factors using the multiple correlation analysis algorithm;

[0065] S413. Calculate the path coefficients of each climate factor using the path analysis algorithm;

[0066] S414. Based on the correlation coefficient, the multiple correlation coefficient, and the path coefficient, analyze the relationship between climate factors and runoff transformation within the monitoring area.

[0067] Furthermore, the analysis of the causal relationship between climate factors and runoff change using the Granger causality test includes the following steps:

[0068] Autocorrelation plots are used to preliminarily test the stationarity of precipitation, potential evapotranspiration, and runoff at different time scales. If the autocorrelation coefficient decays rapidly to 0 with increasing lag time, it indicates that the time series is stationary. If the time series is not stationary, the non-stationary time series is differencing, and the stationarity of the processed data is quantitatively judged using the unit root test. After the non-stationary time series is processed into a stationary time series, Granger causality test is performed.

[0069] Furthermore, the analysis of the relationship between climate change, human activities, and runoff change using a joint algorithm includes the following steps:

[0070] S431. Calculate the contribution rates of climate change and human activities to annual runoff variation and seasonal runoff variation using the cumulative slope change rate method.

[0071] S432. Use the double cumulative curve method to calculate the contribution rates of climate change and human activities to annual runoff variation and seasonal runoff variation.

[0072] S433. Based on the contribution rate calculation results, analyze the relationship between climate change and human activities on annual runoff variation and seasonal runoff variation.

[0073] The beneficial effects of this invention are as follows:

[0074] This invention improves the reliability of test results by relying on the examination of trends and abrupt change points in hydrological sequences and comprehensively determining these trends and abrupt change points using multiple trend and abrupt change test methods. Simultaneously, it analyzes the climatic factors causing runoff changes from multiple perspectives, enriching and perfecting the research results. In the qualitative study, the Granger causality test is introduced to improve the accuracy of hydrological element analysis, greatly enhancing the accuracy of characteristic evolution analysis and runoff change attribution analysis of hydrological elements, and improving the efficiency of hydrological prediction. Attached Figure Description

[0075] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0076] Figure 1 This is a flowchart of a method for analyzing the evolution characteristics of hydrological elements and the attribution of runoff changes under changing environments according to an embodiment of the present invention.

[0077] Figure 2 This is one of the TPDC precipitation and measured precipitation scatter plots based on the hydrological element evolution characteristics and runoff change attribution analysis method under changing environment according to an embodiment of the present invention.

[0078] Figure 3 This is the second scatter plot of TPDC precipitation and measured precipitation based on the evolution characteristics of hydrological elements and the attribution analysis method of runoff change under changing environment according to an embodiment of the present invention.

[0079] Figure 4 This is the third scatter plot of TPDC precipitation and measured precipitation based on the evolution characteristics of hydrological elements and the attribution analysis method of runoff change under changing environment according to an embodiment of the present invention.

[0080] Figure 5 This is one of the TPDC precipitation and measured precipitation trend diagrams based on the evolution characteristics of hydrological elements and runoff change attribution analysis method under changing environment according to an embodiment of the present invention.

[0081] Figure 6 This is the second of the TPDC precipitation and measured precipitation trend diagrams based on the evolution characteristics of hydrological elements and runoff change attribution analysis method under changing environment according to an embodiment of the present invention.

[0082] Figure 7 This is the third TPDC precipitation and measured precipitation trend diagram based on the evolution characteristics of hydrological elements and runoff change attribution analysis method under changing environment according to an embodiment of the present invention. Detailed Implementation

[0083] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention. The components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0084] According to embodiments of the present invention, a method for analyzing the evolution characteristics of hydrological elements and the attribution of runoff changes under changing environments is provided.

[0085] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figures 1-7 As shown, the method for analyzing the evolution characteristics of hydrological elements and the attribution of runoff changes under changing environments according to an embodiment of the present invention includes the following steps:

[0086] S1. Select the monitoring area and collect measured hydrological data within that area;

[0087] Specifically, measured hydrological data includes measured precipitation data, measured runoff data, and measured evaporation data.

[0088] S2. Acquire precipitation data and use measured hydrological data for accuracy evaluation and correction;

[0089] Specifically, the acquisition of precipitation data and the use of measured hydrological data for accuracy evaluation and correction include the following steps:

[0090] S21. Based on the measured precipitation data, the accuracy of the precipitation data is evaluated using five indicators: correlation coefficient, root mean square error, mean absolute error, Nash efficiency coefficient, and Kling-Gupta efficiency coefficient.

[0091] Specifically, the accuracy evaluation index is constructed based on the actual precipitation data of the stations. Five indicators are used to evaluate the accuracy of the TPDC precipitation data before and after correction: correlation coefficient (CC), root mean square error (RMSE), mean absolute error (MAE), Nash efficiency coefficient (NSE), and Kling-Gupta efficiency coefficient (KGE).

[0092] The correlation coefficient describes the degree of linear correlation between TPDC (Precipitation Data Convergence) and measured precipitation data. Its value ranges from -1 to 1. A correlation coefficient closer to 1 or -1 indicates a stronger positive / negative correlation between the TPDC and measured precipitation data. A correlation coefficient of 0 indicates no correlation between the two. The correlation coefficient is calculated as follows:

[0093]

[0094] Where, x i Represents measured precipitation data;

[0095] y i Represents precipitation data TPDC;

[0096] For x i The average value;

[0097] For y i The average value;

[0098] n represents the total number of samples in the precipitation series.

[0099] The root mean square error (RMSE) describes the fluctuation and error between the total precipitation data (TPDC) and the measured precipitation data. A smaller RMSE value indicates higher accuracy of the TPDC relative to the measured precipitation data, and vice versa. The RMSE is calculated as follows:

[0100]

[0101] Where, x i Represents measured precipitation data;

[0102] y i Represents precipitation data TPDC;

[0103] For x i The average value;

[0104] For y i The average value;

[0105] n represents the total number of samples in the precipitation series.

[0106] Mean absolute error avoids the problem of positive and negative errors canceling each other out, and is the most natural and explicit measure of mean error. Its calculation formula is:

[0107]

[0108] Where, x i Represents measured precipitation data;

[0109] y i Represents precipitation data TPDC;

[0110] For x i The average value;

[0111] For y i The average value;

[0112] n represents the total number of samples in the precipitation series.

[0113] The Nash efficiency coefficient (NSE) is used to verify the quality of precipitation data TPDC (Total Precipitation Data Distribution). A NSE value close to 0 indicates that the TPDC is closer to the average level of measured precipitation. NSE ≤ 0.5 is considered unacceptable, 0.5 < NSE ≤ 0.7 is acceptable, 0.7 < NSE ≤ 0.9 is good, and 0.9 < NSE ≤ 1.0 is excellent. The calculation formula is as follows:

[0114]

[0115] Where, x i Represents measured precipitation data;

[0116] y i Represents precipitation data TPDC;

[0117] For x i The average value;

[0118] For y i The average value;

[0119] n represents the total number of samples in the precipitation series.

[0120] The Kling-Gupta efficiency coefficient is used to assess the quality of precipitation data TPDC (Total Term Distribution Data). Compared to the NSE (Neural Segregation of Precipitation) value, the KGE value has the advantage of addressing the issue of different components of the NSE value coupling and affecting the evaluation results. Generally, a KGE value close to -0.41 indicates that the precipitation data TPDC is close to the average level of observed precipitation. When 0.3 < KGE ≤ 1.0, the data is considered acceptable, and the closer the KGE value is to 1.0, the better the data quality is considered. The calculation formula is as follows:

[0121]

[0122] Where, x i Represents measured precipitation data;

[0123] y i Represents precipitation data TPDC;

[0124] For x i The average value;

[0125] For y i The average value;

[0126] n is the total number of samples in the precipitation series;

[0127] i represents the position of a precipitation data point in the sequence (i = 1, 2, ..., n);

[0128] σ x These are measured precipitation data;

[0129] σ y t is the standard deviation of precipitation data TPDC.

[0130] Specifically, at the seasonal scale, the TPDC precipitation data and observed precipitation data show the strongest linear correlation in winter, with a correlation coefficient of 0.95 and a root mean square error (RMSE) of 37.66. The calculated standard deviation of the observed precipitation data is 92.75. If the RMSE is less than half the standard deviation of the observed data, it is generally considered sufficiently low. At the annual scale, the correlation coefficient between TPDC and observed precipitation data is also high, at 0.85, with a RMSE of 237.33, exceeding half the standard deviation of the observed data. Therefore, the TPDC annual precipitation data may contain significant errors and requires correction. Figure 2 In terms of data correlation, the linear correlation between TPDC precipitation data and measured precipitation data is strongest in winter, followed by the annual scale, then spring and autumn, and weakest in summer, with a correlation of only 0.68. In terms of mean absolute error, the error values ​​are winter < autumn < spring < summer < annual. In terms of root mean square error, only the root mean square error of winter precipitation data is less than half of the standard deviation of measured precipitation, while other scales do not meet this requirement. This means that TPDC precipitation data may have a large bias in other seasons and further correction is needed.

[0131] Using time as the independent variable and the corresponding TPDC (Total Precipitation Data Convergence) and measured precipitation data as the dependent variables, a time trend graph of precipitation changes was plotted to further verify the quality of the TPDC precipitation data. Figure 3 This is a trend chart of two precipitation data at different time scales. The trends are generally consistent, both showing upward and downward fluctuations. Furthermore, the measured precipitation is often higher than the TPDC precipitation data, indicating that the TPDC precipitation is underestimated. Precipitation is mainly concentrated in summer, followed by spring and autumn, with the least precipitation in winter. Figure 2 It can be seen that during periods of low precipitation, the TPDC precipitation data and the measured precipitation data are more consistent, which is consistent with... Figure 2 The conclusions reached are consistent.

[0132] Furthermore, comparing the NSE values ​​reveals that the highest NSE value occurs in winter (0.84), while the lowest is in summer (0.40). Ranking the NSE values ​​as winter > autumn = annual > spring > summer, and using NSE > 0.5 as the standard for judging the quality of TPDC precipitation data, spring and summer precipitation are unacceptable, while annual and autumn precipitation are barely acceptable, with their NSE values ​​slightly exceeding the lower limit of the acceptable range. Winter data quality is good. Comparing the KGE values ​​shows that winter precipitation has the highest KGE value (0.72), while annual, spring, and autumn precipitation KGE values ​​are all between 0.60 and 0.65. Summer precipitation has the lowest KGE value (0.49). Using KGE > 0.3 as the standard for judging the acceptable quality of TPDC precipitation data, TPDC precipitation data is acceptable at different time scales, with winter data having the best quality.

[0133] In summary, the TPDC precipitation data exhibits a tendency to underestimate measured precipitation data, with the underestimation becoming more pronounced as precipitation amounts increase. The trends in TPDC precipitation data largely align with measured precipitation data, with correlation coefficients of 0.68 and above, reaching a maximum of 0.95. Considering correlation coefficients, root mean square error, mean absolute error, NSE, and KGE values, TPDC data demonstrates higher accuracy for winter precipitation. While other precipitation scales show high correlation coefficients, the relationship between root mean square error and the standard deviation of measured data suggests potential significant biases. Furthermore, the assessment of NSE and KGE values ​​necessitates further correction of TPDC precipitation data to enhance its applicability.

[0134] S22. Optimize the BP neural network using the sparrow search algorithm;

[0135] Specifically, the optimization of the BP neural network using the sparrow search algorithm includes the following steps:

[0136] S221. Initialize the population and related parameters, and calculate the fitness value of the initial population;

[0137] Specifically, the input layer accepts external signals as input, denoted as x1, x2, ..., x... i , ..., x n The j-th neuron in the hidden layer receives input from n neurons in the output layer and outputs to the output layer. The input and output expressions of the hidden layer neurons are as follows:

[0138]

[0139]

[0140] In the formula, ω ij Let θ be the weight from the i-th neuron in the input layer to the j-th neuron in the hidden layer. j This represents the deviation of the j-th neuron in the hidden layer. Here is the activation function for the hidden layer;

[0141] The k-th neuron in the output layer receives input from p neurons in the hidden layer and outputs the result. The input and output expressions of the output layer neurons are as follows:

[0142]

[0143]

[0144] In the formula, ω ij Let α be the weight from the j-th neuron in the hidden layer to the k-th neuron in the output layer. kThis represents the bias of the k-th neuron in the output layer. The activation function for the output layer;

[0145] Backpropagation of the error calculates the output error of the output layer and determines whether this error is less than a set error value μ. If it is less than the set error, training ends. If it is greater than the set error, it determines whether the maximum number of training iterations C has been reached. If yes, training terminates; otherwise, training continues until the condition is met. The error function expression is:

[0146]

[0147] In the formula, Let be the expected output value of the k-th neuron in the output layer.

[0148] S222. Update the discoverer's location, calculated using the following formula:

[0149]

[0150] S223. Update the position of the new member. The calculation formula is as follows:

[0151]

[0152] S224. Update the location of sparrows that are aware of danger. The calculation formula is:

[0153]

[0154] in, Iter represents the J-th dimension position information of the i-th sparrow in the t-th iteration; max This represents the maximum number of iterations.

[0155] α is a random number between 0 and 1;

[0156] R2 is the warning value;

[0157] ST is a safe value;

[0158] Q is a random number that follows a normal distribution;

[0159] The worst individual in the t-th iteration;

[0160] X p This is the current position of the best discoverer;

[0161] f i This represents the current fitness value of the individual.

[0162] f g This represents the maximum fitness value of the current individual.

[0163] f wThis represents the minimum fitness value of the current individual.

[0164] A + =A T (AA T ) -1 ;

[0165] A is a 1×d matrix, with elements being randomly assigned values ​​of 1 or -1;

[0166] L is a 1×d matrix consisting entirely of 1s;

[0167] d is the dimension of the problem to be optimized;

[0168] β is a random number that follows a normal distribution with a mean of 0 and a variance of 1;

[0169] K is the current maximum fitness value;

[0170] It is the optimal individual in the t-th iteration;

[0171] n is the number of sparrows;

[0172] ε is a constant to avoid the denominator being zero.

[0173] S23. The precipitation data is corrected using the optimized BP neural network.

[0174] S3. The evolution characteristics of precipitation and runoff are analyzed using a fusion algorithm.

[0175] The analysis of the evolution characteristics of precipitation and runoff using the fusion algorithm method includes the following steps:

[0176] S31. Analyze the evolution characteristics of hydrological elements within the monitoring area;

[0177] Specifically, the analysis of the evolution characteristics of hydrological elements within the monitoring area includes the following steps:

[0178] S311. The trend of hydrological elements in the monitoring area is assessed using trend analysis.

[0179] Specifically, trend analysis is conducted using various methods, including the climate trend rate method, the MK trend test method, the Kendall rank test method, and the Spearman rank test method.

[0180] The climate trend rate method determines the increasing or decreasing trend of a factor over time by establishing a univariate linear regression equation between hydrological variables and time. The formula is as follows:

[0181] x i =a+bt i

[0182]

[0183]

[0184] Where, x i Hydrological elements;

[0185] t i For the corresponding time;

[0186] a is the regression constant;

[0187] b is the regression coefficient;

[0188] n represents the sample size of hydrological elements;

[0189] For element values;

[0190] This is the average over time.

[0191] Specifically, b×10 represents the climate trend rate. If b>0, it indicates that the factor is on an upward trend over time; otherwise, it indicates a downward trend. Finally, the significance level 'a' is determined, and the regression equation is tested for significance to determine whether the trend of change is significant.

[0192] In the MK trend test, the null hypothesis H0 indicates that the hydrological element has no significant trend over time, and the alternative hypothesis H1 indicates that the hydrological element has a significant upward / downward trend over time. The formulas for calculating the MK test statistic S, variance Var(S), and standard MK test statistic Z are as follows:

[0193]

[0194]

[0195]

[0196]

[0197] Where n is the length of the time series, if Z > 0, it indicates that the hydrological element has an upward trend over time, and vice versa. If |Z| < Z α If the null hypothesis is accepted, it indicates that the hydrological series does not show a significant increasing or decreasing trend at the confidence level of a.

[0198] Kendall's rank test is used for time series elements x. i (i = 1, 2, ..., n), first determine (x i x j In (i < j), x i <x jThe number of times a statistic appears is denoted as P, and the formula for calculating the statistic U is:

[0199]

[0200] The formulas for calculating the Spearman rank correlation coefficient rs and the test statistic T are as follows:

[0201]

[0202]

[0203] Where n is the length of the time series, t is the time sequence, and T follows a t-distribution with (n-2) degrees of freedom. The significance level a is determined, and the trend of the series is tested.

[0204] S312. Use the mutation analysis method to identify the mutation points of hydrological elements in the monitoring area;

[0205] Specifically, we analyze the abrupt changes of hydrological elements using multiple methods such as the MK mutation test, the sliding t-test, and ordered clustering.

[0206] The MK burst test formula is: for time series element x i Construction of (i = 1, 2, ..., n):

[0207]

[0208] In the formula, R i For x i >x j The cumulative quantity of (1≤j≤i) is defined as follows:

[0209]

[0210] Specifically, UF1 = 0, E(S) k ), Var(S k S are respectively k The mean and variance of the time series are calculated by repeating the above calculation process in reverse order, and UB is given. k =-UF k (K = n, n-1, ..., 1), UB1 = 0. When UB k and UF k When two curves intersect within the critical line, it indicates that the point is a significant abrupt change.

[0211] The sliding t-test formula is as follows: Divide a continuous n hydrological sequence into subsequences X1 and X2, with sequence lengths n1 and n2, means μ1 and μ2, and variances S1 and S2, respectively. 2 and S2 2Construct a statistic T that follows a t-distribution with n1+n2-2 degrees of freedom, and test its significance to determine whether a significant abrupt change has occurred. The calculation formula is as follows:

[0212]

[0213] Ordered clustering: for time series elements x i For (i = 1, 2, ..., n), the sum of squared deviations before and after the abrupt change point t is minimized. The calculation formula is:

[0214]

[0215] in, These are the mean values ​​before and after the mutation, respectively.

[0216] S313. The periodicity of hydrological elements in the monitoring area is assessed using periodicity analysis.

[0217] Specifically, Morle wavelet analysis is used to assess the periodicity of hydrological elements. The wavelet transform function is calculated using the following formula:

[0218]

[0219] Where a is the scaling factor, T is the time step (i.e., the translation factor), w* is the complex conjugate function, and the formula for calculating the wavelet variance is:

[0220]

[0221] S314. Predict the future trend of precipitation runoff in the monitored area using the recalibrated range analysis method;

[0222] Specifically, the rescaled range analysis method is used to predict future trends in precipitation runoff. First, based on a new hydrological element sequence ξ(t) (t = 1, 2, ...), for any positive integer τ, the mean of this sequence is... The formulas for calculating the cumulative deviation X(t, τ), range R(τ), and standard deviation S(τ) are as follows:

[0223]

[0224]

[0225] R(τ)=maxX(t,τ)-minX(t,τ) 1≤t≤τ

[0226]

[0227]

[0228] lg(R / S) τ=Hlga+HlgH

[0229] Here, H is the Hurst index, whose magnitude reflects the positive / negative persistence of future hydrological elements. When 0 < H < 0.5, it indicates that the future trend of hydrological elements is opposite to that of the past. When 0.5 < H < 1.0, it indicates that the future trend of hydrological elements is the same as that of the past. When H = 0.5, it indicates that the future trend of hydrological elements is unrelated to the past. Further subdividing H, when 0 < H ≤ 0.35, it indicates that the future trend of hydrological elements shows strong negative persistence, and so on, when 0.35 < H < 0.5, it shows strong persistence.

[0230] S315. Analyze the annual distribution characteristics of precipitation and runoff within the monitoring area using multiple indicators.

[0231] S32. Analyze the spatial evolution characteristics of precipitation within the monitoring area;

[0232] Specifically, the analysis of the spatial evolution characteristics of precipitation within the monitoring area includes the following steps:

[0233] S321. Obtain the distribution map of multi-year average precipitation and multi-year average seasonal precipitation in the monitoring area using the inverse distance weighted difference, and analyze the spatial evolution characteristics of precipitation;

[0234] S322. Use the inverse distance weighted difference to obtain the spatial distribution map of precipitation and seasonal precipitation in the monitoring area, and analyze the spatial evolution characteristics of precipitation trend.

[0235] Specifically, the monthly precipitation data of the hydrological collection area was calculated using Thiessen polygons, and the seasons were divided as follows: March to May is spring, June to August is summer, September to November is autumn, and December to February of the following year is winter.

[0236] Specifically, based on daily meteorological data, the daily potential evapotranspiration is calculated using the Penman formula, and the seasons are divided as follows: March to May is spring, June to August is summer, September to November is autumn, and December to February of the following year is winter.

[0237] S33. Analyze the interannual variation characteristics of precipitation runoff in the monitored area;

[0238] S34. Analyze the annual distribution characteristics of precipitation runoff within the monitoring area.

[0239] S4. Based on the analysis results, conduct attribution analysis on the runoff changes in the monitoring area.

[0240] The attribution analysis of runoff changes in the monitoring area based on the analysis results includes the following steps:

[0241] S41. Analyze the relationship between climate factors and runoff changes within the monitoring area using multiple algorithms;

[0242] Specifically, the analysis of the relationship between climate factors and runoff changes within the monitoring area using multiple algorithms includes the following steps:

[0243] S411. Calculate the correlation coefficient between runoff and a single climate factor using partial correlation analysis algorithm;

[0244] Specifically, partial correlation analysis is used to calculate the correlation coefficient between a single factor and runoff, while controlling for other variables. Taking potential evapotranspiration as an example, the partial correlation coefficient ρ between precipitation and runoff is... PR (E) The calculation formula is:

[0245]

[0246] In the formula, ρ PR ρ PE and ρ RE These represent the linear correlation coefficients between precipitation and runoff, precipitation and potential evapotranspiration, and runoff and potential evapotranspiration, respectively.

[0247] S412. Calculate the multiple correlation coefficient between runoff and climate factors using the multiple correlation analysis algorithm;

[0248] Specifically, partial correlation analysis is used to calculate the correlation coefficient between a single factor and runoff, while controlling for other variables. Taking potential evapotranspiration as an example, the partial correlation coefficient ρ between precipitation and runoff is... PR (E) The calculation formula is:

[0249]

[0250] In the formula, ρ PR ρ PE and ρ RE These represent the linear correlation coefficients between precipitation and runoff, precipitation and potential evapotranspiration, and runoff and potential evapotranspiration, respectively.

[0251] S413. Calculate the path coefficients of each climate factor using the path analysis algorithm;

[0252] S414. Based on the correlation coefficient, the multiple correlation coefficient, and the path coefficient, analyze the relationship between climate factors and runoff transformation within the monitoring area.

[0253] S42. Further analyze the causal relationship between climate factors and runoff changes using Granger causality test;

[0254] Specifically, the further analysis of the causal relationship between climate factors and runoff change using the Granger causality test includes the following steps:

[0255] Autocorrelation plots are used to preliminarily test the stationarity of precipitation, potential evapotranspiration, and runoff at different time scales. If the autocorrelation coefficient decays rapidly to 0 with increasing lag time, it indicates that the time series is stationary. If the time series is not stationary, the non-stationary time series is differencing, and the stationarity of the processed data is quantitatively judged using the unit root test. After the non-stationary time series is processed into a stationary time series, Granger causality test is performed.

[0256] Specifically, using each climate factor x1, x2...x n With runoff y as the independent variable and runoff y as the dependent variable, path analysis can decompose the correlation coefficient between a certain climate factor and runoff into two parts—direct effect (direct path coefficient) and indirect effect (indirect path coefficient)—without considering the influence of other factors, thus establishing the correlation between runoff y and climate factors x1, x2…x n The linear regression equation is y = a0 + a1x1 + a2x2 + ... + a n x n Based on the correlation coefficients between climate factors and the correlation coefficients between each climate factor and runoff By performing a mathematical transformation on the linear regression equation to establish a matrix equation, the path coefficients can be obtained. The expression of the matrix equation is as follows:

[0257]

[0258] Among them, a1, a2...a n These are the standardized partial regression coefficients, also known as x1, x2, ..., x. n For the direct path coefficient of runoff, The indirect path coefficient of climate factor x1 on runoff y via x2. Let x2 be the indirect path coefficient of the climate factor x2 through x1 on the runoff y. By analogy, the indirect path coefficients of each factor can be obtained.

[0259] Specifically, in a given time series scenario, the Granger causal relationship between climate factors and runoff change means that predicting current runoff using past information on potential evapotranspiration and runoff is more effective than predicting runoff using only past information on runoff itself. In other words, if past changes in climate factors can effectively explain current runoff changes, then climate factors are considered to be the Granger cause of runoff change.

[0260] Granger causality test table for runoff, precipitation, and potential evapotranspiration

[0261]

[0262] Note: *, ** and *** indicate the existence of Granger causality at significance levels of 0.1, 0.05 and 0.01, respectively.

[0263] For the annual runoff series, throughout the study period, the impact of precipitation on runoff passed the significance test at the 0.1 level, indicating that changes in precipitation were a Granger cause of runoff changes. However, the impact of potential evapotranspiration on runoff did not pass the significance test at the 0.1 level, indicating that changes in potential evapotranspiration were not a Granger cause of runoff changes. During the baseline period, the impacts of both precipitation and potential evapotranspiration on runoff did not pass the significance test at the 0.01 and 0.1 levels, respectively, indicating that both were Granger causes of runoff changes. During the variation period, the impacts of precipitation and potential evapotranspiration on runoff passed the significance tests at the 0.01 and 0.1 levels, respectively, indicating that both were Granger causes of runoff changes.

[0264] Using the same approach to analyze the seasonal runoff series, for the spring runoff series, regardless of the study period, the effects of precipitation and potential evapotranspiration on runoff depth did not pass the significance level test, indicating that neither was a Granger cause of spring runoff changes. For the summer runoff series, during the variation period, both precipitation and potential evapotranspiration were Granger causes of runoff changes; during the second variation period, only precipitation was a Granger cause of runoff changes, while in other periods, neither precipitation nor potential evapotranspiration was a Granger cause of runoff changes. For the autumn runoff series, during the baseline and variation periods, neither precipitation nor runoff was a Granger cause of runoff changes; during the second variation period and throughout the entire study period, only potential evapotranspiration was a Granger cause of runoff changes. For the winter runoff series, regardless of the study period, neither precipitation nor potential evapotranspiration was a Granger cause of runoff changes.

[0265] S43. Use a joint algorithm to analyze the relationship between climate change, human activities and runoff changes.

[0266] Specifically, the analysis of the relationship between climate change, human activities, and runoff change using a joint algorithm includes the following steps:

[0267] S431. Calculate the contribution rates of climate change and human activities to annual runoff variation and seasonal runoff variation using the cumulative slope change rate method.

[0268] Specifically, the runoff series is divided into a baseline period before the abrupt change and a change period after the abrupt change based on the year of the abrupt change. The slope S of the linear relationship between the years before and after the abrupt change and the cumulative runoff depth is calculated for each period. Rb and SRa (mm / a), the slope S of the linear relationship between year and cumulative precipitation. Pb and S Pa (mm / a), the slope S of the linear relationship between year and cumulative potential evapotranspiration. Eb and S Ea If (mm / a), then the contributions of precipitation, potential evapotranspiration, and human activities to runoff variation C P (%), C E (%), C H (%) are respectively:

[0269]

[0270]

[0271] C H =100-C P -C E

[0272] S432. Use the double cumulative curve method to calculate the contribution rates of climate change and human activities to annual runoff variation and seasonal runoff variation.

[0273] Specifically, the contribution rate of each factor to runoff variation is analyzed using the double cumulative curve method. Taking the contribution rate of precipitation as an example, the double cumulative curve of precipitation-runoff depth in the baseline period is first plotted, and the univariate linear regression equation of the curve is determined as follows:

[0274] ∑Q=k∑P+b

[0275] Where Q is the runoff depth; k is the slope; P is the precipitation; and b is the intercept.

[0276] By substituting the cumulative precipitation during the abrupt change period into the theoretical cumulative runoff depth for that period, a double cumulative curve of actual precipitation runoff depth and a double cumulative curve of theoretical precipitation runoff depth for the abrupt change period are plotted. The measured runoff depth before and after the abrupt change and the theoretical runoff depth after the abrupt change are calculated respectively, thereby obtaining the contribution rate of precipitation and non-precipitation factors to runoff change. The same method is used to determine the contribution rate of potential evapotranspiration and non-potential evapotranspiration factors to runoff change. Finally, the contribution rate of human activities to runoff change is calculated based on the principle of water balance.

[0277] S433. Based on the contribution rate calculation results, analyze the relationship between climate change and human activities on annual runoff variation and seasonal runoff variation.

[0278] In summary, by utilizing the above-mentioned technical solutions of this invention, the present invention improves the reliability of the test results by relying on the examination of the trend and abrupt change points of the hydrological sequence and comprehensively determining the trend and abrupt change points of the sequence using multiple trend examination methods and abrupt change examination methods. At the same time, it analyzes the climatic factors that cause runoff changes from multiple perspectives, making the research results richer and more complete. In the qualitative research, the Granger causality test is introduced to improve the accuracy of the analysis of hydrological elements, which greatly improves the accuracy of the characteristic evolution analysis and runoff change attribution analysis of hydrological elements, and improves the efficiency of hydrological prediction.

[0279] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for analyzing the evolution characteristics of hydrological elements and the attribution of runoff changes under changing environments, characterized in that, Includes the following steps: S1. Select the monitoring area and collect measured hydrological data within the area; S2. Obtain precipitation data and use the measured hydrological data for accuracy evaluation and correction; The acquisition of precipitation data and the use of measured hydrological data for accuracy evaluation and correction include the following steps: S21. Based on the measured precipitation data, the accuracy of the precipitation data is evaluated using five indicators: correlation coefficient, root mean square error, mean absolute error, Nash efficiency coefficient, and Kling-Gupta efficiency coefficient. S22. Optimize the BP neural network using the sparrow search algorithm; S23. The precipitation data is corrected using the optimized BP neural network; S3. The evolution characteristics of precipitation and runoff are analyzed using a fusion algorithm. The analysis of the evolution characteristics of precipitation and runoff using the fusion algorithm method includes the following steps: S31. Analyze the evolution characteristics of hydrological elements within the monitoring area; The analysis of the evolution characteristics of hydrological elements within the monitoring area includes the following steps: S311. The trend of hydrological elements in the monitoring area is assessed using trend analysis. S312. Use the mutation analysis method to identify the mutation points of hydrological elements in the monitoring area; S313. The periodicity of hydrological elements in the monitoring area is assessed using periodicity analysis. S314. Predict the future trend of precipitation runoff in the monitored area using the recalibrated range analysis method; S315. Analyze the annual distribution characteristics of precipitation and runoff within the monitoring area using multiple indicators; S32. Analyze the spatial evolution characteristics of precipitation within the monitoring area; The analysis of the spatial evolution characteristics of precipitation within the monitoring area includes the following steps: S321. Obtain the distribution map of multi-year average precipitation and multi-year average seasonal precipitation in the monitoring area using the inverse distance weighted difference, and analyze the spatial evolution characteristics of precipitation; S322. Use the inverse distance weighted difference to obtain the spatial distribution map of precipitation and seasonal precipitation in the monitoring area, and analyze the spatial evolution characteristics of precipitation trend. S33. Analyze the interannual variation characteristics of precipitation runoff in the monitored area; S34. Analyze the annual distribution characteristics of precipitation runoff within the monitoring area; S4. Based on the analysis results, conduct attribution analysis on the runoff changes in the monitoring area; The attribution analysis of runoff changes in the monitoring area based on the analysis results includes the following steps: S41. Analyze the relationship between climate factors and runoff changes within the monitoring area using multiple algorithms; The analysis of the relationship between climate factors and runoff changes within the monitoring area using multiple algorithms includes the following steps: S411. Calculate the correlation coefficient between runoff and a single climate factor using partial correlation analysis algorithm; S412. Calculate the multiple correlation coefficient between runoff and climate factors using the multiple correlation analysis algorithm; S413. Calculate the path coefficients of each climate factor using the path analysis algorithm; S414. Based on the correlation coefficient, the multiple correlation coefficient, and the path coefficient, analyze the relationship between climate factors and runoff transformation within the monitoring area; S42. Further analyze the causal relationship between climate factors and runoff changes using Granger causality test; The further analysis of the causal relationship between climate factors and runoff change using the Granger causality test includes the following steps: The stationarity of precipitation, potential evapotranspiration, and runoff at different time scales is preliminarily tested using autocorrelation plots. If the autocorrelation coefficient can rapidly decay to 0 with the increase of lag time, it indicates that the time series is stationary. If the time series is not stationary, the non-stationary time series is differencing, and the stationarity of the processed data is quantitatively judged using the unit root test method. After the non-stationary time series is processed into a stationary time series, Granger causality test is performed. S43. Analyze the relationship between climate change, human activities and runoff changes using a joint algorithm; The analysis of the relationship between climate change, human activities, and runoff change using a joint algorithm includes the following steps: S431. Calculate the contribution rates of climate change and human activities to annual runoff variation and seasonal runoff variation using the cumulative slope change rate method. S432. Use the double cumulative curve method to calculate the contribution rates of climate change and human activities to annual runoff variation and seasonal runoff variation. S433. Based on the contribution rate calculation results, analyze the relationship between climate change and human activities on annual runoff variation and seasonal runoff variation.

2. The method for analyzing the evolution characteristics of hydrological elements and the attribution of runoff changes under changing environments according to claim 1, characterized in that, The optimization of the BP neural network using the sparrow search algorithm includes the following steps: S221. Initialize the population and related parameters, and calculate the fitness value of the initial population; S222. Update the discoverer's location, calculated using the following formula: S223. Update the position of the new member. The calculation formula is as follows: S224. Update the location of sparrows that are aware of danger. The calculation formula is: in, For the first Only sparrows in the first The th iteration in the Dimensional location information; This represents the maximum number of iterations. A random number between 0 and 1; This is a warning value; This is a safe value; These are random numbers that follow a normal distribution. For the first The worst individual in the next iteration; This is the current position of the best discoverer; This represents the current fitness value of the individual. This represents the maximum fitness value of the current individual. This represents the minimum fitness value of the current individual. A is a 1×d matrix, with elements being randomly assigned values ​​of 1 or -1; 1× for all 1s Matrix; Let be the dimension of the problem to be optimized; These are random numbers that follow a normal distribution with a mean of 0 and a variance of 1. This is the current maximum fitness value; For the first The optimal individual in the next iteration; n is the number of sparrows; To avoid constants with a denominator of zero.

Images