River annual runoff process prediction method and system based on rank theory and medium

Abstract

The invention relates to a river annual runoff process prediction method and system based on a rank theory and a medium, and the method comprises the steps: carrying out the rank transformation of a runoff sequence through an sgn function based on a river multi-year daily average flow process; by calculating the cumulative order sequence of the runoff process, a clustering algorithm is used for recognizing the mutation characteristics of the runoff process for many years and the platform period of the cumulative order sequence in the year; predicting an extreme drought event in combination with separation characteristics of an annual cumulative order sequence and a boundary condition; and establishing a quantitative relationship between the platform period duration time and the runoff volume, and predicting the annual runoff change process. According to the method, the change rule of the river runoff accumulation order sequence is identified by using the clustering algorithm, and a quantitative mutation detection method can be provided according to the annual runoff process distribution form change.

Description

Technical Field

[0001] This application relates to the field of hydrology and water resources, and particularly relates to a method, system, and medium for predicting the annual runoff process of a river based on rank theory. Background Art

[0002] In recent years, the impacts of climate change and human activities on river runoff changes have intensified, and the instability and periodicity of river runoff changes have both changed significantly. Studying the changing laws of the river runoff process and proposing effective prediction methods can provide a theoretical basis for river flood prevention early warning and river channel protection. At the same time, it also has guiding significance for the safe operation of water conservancy projects and increasing power generation benefits. There are three types of characteristics in the river runoff change process, namely volatility, periodicity, and seasonality. In terms of volatility, there is a complex physical adjustment mechanism between river runoff changes and natural elements such as underlying surface conditions, rainfall conditions, and temperature, resulting in the instability of the runoff process over time being difficult to predict. In terms of periodicity, existing research generally believes that the river runoff process has transformation periods on multiple time scales, and the runoff process is the result of the coupling of multiple hydrological cycles. In terms of seasonality, there are obvious peak characteristics in the annual distribution of river runoff changes. Affected by the increase in summer temperature, glacier meltwater, and rainfall, the average daily flow during the flood season is much larger than that during the non-flood season, and the flood season and non-flood season in the annual runoff process alternate. Under the influence of the above multi-faceted factors, the physical mechanism of river runoff changes is still not clear, and the changing laws of the runoff process need to be revealed under extreme climate and reservoir operation conditions.

[0003] Currently, the methods for trend testing of the runoff process include: R / S analysis, wavelet analysis, MK test, linear regression, etc. Among them, the R / S analysis method is mainly used to judge the persistence of runoff change trends, the wavelet analysis method is used to decompose various periods of runoff changes and the occurrence time of each period, and the MK test is based on the order process of annual runoff to test the runoff mutation year. The above methods are all based on annual runoff to judge the river runoff change trend, while extreme climate and reservoir operation mainly change the distribution form of the annual runoff process. Currently, there is a lack of quantitative parameters to characterize the annual runoff process form and effective trend testing methods. Based on different stages of river runoff changes, to predict the annual runoff process, the currently widely applicable methods are watershed runoff generation models (SWAT, SWMM), joint probability distribution models, statistical models, etc. The watershed sediment yield model constructs the relationship between various natural elements and runoff volume, the joint probability distribution model infers the runoff process according to the probability distribution of different times within a year and different runoff conditions, and the statistical model predicts the relationship between various conditions and runoff volume through neural networks, Bayesian models, etc. The above prediction models are all empirical models proposed based on data statistics, and do not pay attention to the continuous characteristics of runoff spatio-temporal transport, and cannot construct a quantitative relationship between the runoff process distribution form and runoff volume, resulting in poor runoff prediction accuracy and inability to be widely promoted to the directions of reservoir operation and hydrological early warning.

[0004] With the intensification of the impact of climate change and human activities on rivers, the distribution pattern of the river runoff process has changed significantly. Currently, there is a lack of quantitative indicators to characterize the distribution pattern of river runoff. The existing trend test methods have low recognition accuracy for the annual runoff variation characteristics of rivers, and the quantitative relationship between river runoff and runoff pattern is not yet clear, resulting in great difficulty in predicting the river runoff process. Summary of the Invention

[0005] The purpose of the embodiments of this application is to overcome the deficiencies of the prior art, and provides a method, system and medium for predicting the annual river runoff process based on the rank theory, establishing a quantitative relationship between the duration of different stages within a year and the runoff volume, for predicting the annual runoff variation process, which can be widely applied in aspects such as basin flood control evaluation, reservoir regulation, and hydrological prediction.

[0006] To achieve the above purpose, this application provides the following technical solutions:

[0007] In the first aspect, the embodiments of this application provide a method for predicting the annual river runoff process based on the rank theory, including the following steps:

[0008] Based on the daily average flow process of the river over the years, use the sgn function to perform rank transformation on the runoff sequence;

[0009] By calculating the cumulative order sequence of the runoff process, use the clustering algorithm to identify the mutation characteristics of the annual runoff process and the plateau period of the annual cumulative order sequence respectively;

[0010] Combined with the separation characteristics of the annual cumulative order sequence and the boundary conditions, predict extreme drought events;

[0011] Establish a quantitative relationship between the duration of the plateau period and the magnitude of its runoff volume, and predict the annual runoff variation process.

[0012] The specific operation of using the sgn function to perform rank transformation on the river runoff sequence is,

[0013]

[0014] Rank(x i ) is the order value corresponding to the daily average flow on the i-th day in the sequence x. When the daily average flow on the i-th day is greater than or equal to that on the (i - 1)-th day, the order value of x i is 1. Conversely, when the daily average flow on the i-th day is less than that on the (i - 1)-th day, the order value of x i is -1.

[0015] The specific calculation of the cumulative order sequence of the runoff process is,

[0016] By calculating the cumulative order series of river runoff, the continuous change characteristics of the runoff process are quantitatively characterized, and the interference of short-term fluctuations in daily average flow on the overall shape of the series is reduced.

[0017]

[0018] Rank’(x i ) is the cumulative order value corresponding to the i-th day of the runoff process. When Rank’(x i ) shows a linear increasing trend over time, the daily average flow of the river continuously increases. Conversely, when Rank’(x i ) shows a linear decreasing trend, the daily average flow of the river continuously decreases. For the runoff change process over a period of time, if the cumulative order value at the end of the series is greater than 0, it means that the rising duration of the river is greater than the falling duration; if the end cumulative order value is less than 0, it means that the falling duration of the river is greater than the rising duration.

[0019] The specific sample deviation of calculating the cumulative value of water and sediment is as follows:

[0020] λ i =S' i -S'1-a0Q' i (5)

[0021] λ i is the deviation of the cumulative sediment transport value of the i-th from the linear regression of the extreme value.

[0022] The principle of using the clustering algorithm to identify the mutation points of the multi-year runoff process is as follows: ① Starting from the head of the cumulative order series of daily average flow, slide the segmentation point and calculate the mean square error of the linear fitting of the two parts on the left and right of the segmentation point respectively;

[0023]

[0024] a=(x n -x1) / n (4)

[0025] σ is the mean square error, a is the linear change rate of the cumulative order series, Rank’(x1) is the initial value of the cumulative order series sub-interval, n is the length of the sub-interval. The smaller the mean square error of the cumulative order series sub-interval, the higher the consistency of the runoff series change trend within the sub-interval.

[0026] σ'=σ1+σ2 (5)

[0027] σ’ is the sum of the mean square deviations of each sub-interval. By determining the position of the segmentation point corresponding to the minimum sum of the mean square deviations, the mutation year of the runoff process is determined.

[0028] The specific method for the clustering algorithm to identify the plateau period of the annual cumulative runoff series is as follows. To accurately distinguish the distribution characteristics of the annual runoff process, the annual runoff cumulative series is calculated for each year, and the clustering algorithm is used to identify the characteristics of the annual runoff variation. Affected by extreme climate and reservoir operation, there are three patterns in the annual runoff cumulative series: ① Under normal conditions, the annual variation process of river runoff is affected by seasonal floods. The cumulative order during the non-flood season shows a continuous downward trend, while it shows an upward trend during the flood season. Therefore, there is a plateau period in the annual runoff cumulative series under normal conditions; ② In the case of extremely dry years, the characteristics of the flood season change are not obvious, and there is no plateau period in the annual runoff cumulative series; ③ Under the condition of reservoir operation, the impoundment of the reservoir flattens the annual runoff process, and the downward trend of the runoff during the non-flood season weakens significantly. The plateau of the annual runoff cumulative series curve is higher than that under normal conditions. Summarize the annual cumulative order curves of these three types in turn.

[0029] To identify the plateau period of the annual runoff cumulative series, the clustering algorithm is used to identify the optimal segmentation points at the start and end of the plateau period. The basic principle is as follows: ① In the annual runoff cumulative series, two segmentation points are set successively; ② Starting from the head of the series, the two segmentation points are slid successively, and the mean square errors of the three sub-intervals are calculated in turn using Equation (3). The sum of the mean square errors of the three is used as the evaluation index for the clustering result.

[0030] σ”=σ1'+σ2'+σ3' (6)

[0031] σ” is the sum of the mean square errors of the annual runoff cumulative series. σ1’, σ2’, and σ3’ are the mean square errors of the annual runoff cumulative order during the initial decline period, plateau period, and secondary decline period respectively. When the sum of the mean square errors of the annual cumulative series is the smallest, it represents the optimal selection position of the plateau period segmentation point.

[0032] The specific method for predicting extreme drought events by combining the separation characteristics of the annual cumulative series and boundary conditions is as follows. First, a linear formula is used to fit the runoff cumulative order curve under extreme drought conditions as the boundary condition of the annual cumulative order curve.

[0033] Rank'(x i )=a×x i (6)

[0034] a is the slope parameter. By calculating the separation difference between the annual cumulative order curve and the boundary condition, the earliest prediction point of extreme drought conditions within the year is judged.

[0035] e i =Rank'(x i )-a×x i (7)

[0036] e iIt is the separation difference between the annual cumulative order curve and the boundary condition. Under general conditions, the separation difference between the annual cumulative order curve and the boundary condition satisfies the S-shaped curve distribution.

[0037] Use the Logistic formula to fit the separation difference curve to obtain the annual variation law of the separation difference under general conditions.

[0038]

[0039] In the above formula, A, B, and C are all dimensionless parameters. Determine the first inflection point where the slope of the Logistic formula increases through Equation (8), that is, the earliest identification point within the year for predicting drought years. Substitute Equation (7) into Equation (8) and simplify to obtain the distribution function of the annual cumulative order series under general conditions.

[0040]

[0041] For the three change stages of the annual cumulative order series of river runoff, use the power function to fit the relationship between the duration of the stage and the runoff volume in turn.

[0042] Q = k1 × n1 (10)

[0043]

[0044] Q = k3 × n3 (12)

[0045] In the above formula, Q is the runoff volume, n1, n2, and n3 are the durations of the initial decline period, the plateau period, and the secondary decline period respectively, k1 and k3 are the slope parameters of the initial decline period and the secondary decline period respectively, and k2 and m2 are the fitting parameters of the plateau period. Use Equations (10) to (12) respectively to predict the annual runoff process.

[0046] In the second aspect, the embodiments of the present application provide a prediction system for the annual runoff process of a river based on the rank theory. The system includes: a memory and a processor. The memory includes a program for the prediction method of the annual runoff process of a river based on the rank theory. When the program for the prediction method of the annual runoff process of a river based on the rank theory is executed by the processor, the following steps are implemented: Based on the daily average flow process of the river over the years, use the sgn function to perform rank transformation on the runoff sequence; by calculating the cumulative order series of the runoff process, use the clustering algorithm to identify the mutation characteristics of the runoff process over the years and the plateau period of the annual cumulative order series respectively; combine the separation characteristics of the annual cumulative order series and the boundary conditions to predict extreme drought events; establish a quantitative relationship between the duration of the plateau period and the magnitude of its runoff volume to predict the annual runoff change process.

[0047] In a third aspect, an embodiment of the present application provides a computer-readable storage medium storing program code, which, when executed by a processor, implements the steps of the above-described method for predicting the annual runoff process based on the rank theory.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0049] (1) By using the clustering algorithm to identify the variation law of the river runoff cumulative order sequence, a quantitative mutation test method can be given according to the change of the annual runoff process distribution pattern.

[0050] (2) By analyzing the annual runoff process distribution pattern through the rank theory, the extreme drought conditions can be effectively identified, and early warnings of extreme climates can be given at the beginning of each year.

[0051] (3) Establish a quantitative relationship between the duration of different stages within a year and the runoff volume, which is used to predict the annual runoff change process and can be widely applied in aspects such as basin flood control evaluation, reservoir regulation, and hydrological prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0053] Figure 1 is a flowchart of the method of the present application;

[0054] Figure 2 is a distribution pattern diagram of the annual cumulative order sequence curve;

[0055] Figure 3 is a separation difference diagram of the annual cumulative order and boundary conditions under general conditions;

[0056] Figure 4 is a relationship diagram between the duration of different stages and the runoff volume;

[0057] Figure 5 is a change process diagram of the runoff cumulative order sequence of the Jinsha River;

[0058] Figure 6 is a change process diagram of the runoff cumulative order sequence of the Jinsha River;

[0059] Figure 7 is a separation difference diagram of the annual runoff cumulative order sequence of the Jinsha River and extreme drought conditions;

[0060] Figure 8It is a graph showing the relationship between the duration and runoff of the Jinsha River at different stages within a year. Detailed implementation manners

[0061] Next, the technical solutions in the embodiments of the present application will be described with reference to the accompanying drawings in the embodiments of the present application. It should be noted that similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0062] The term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device including a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "including a..." does not exclude the presence of additional identical elements in the process, method, article or device including the said element.

[0063] The terms "first", "second", etc. are only used to distinguish one entity or operation from another entity or operation, and cannot be understood as indicating or implying relative importance, nor can it be understood as requiring or implying any actual relationship or order between these entities or operations.

[0064] Please refer to Figures 1 to 4 , an annual runoff process prediction method for rivers based on the rank theory is provided in the embodiments of the present application, including the following steps:

[0065] Based on the daily average flow process of the river over the years, use the sgn function to perform rank transformation on the runoff sequence;

[0066] By calculating the cumulative order sequence of the runoff process, use the clustering algorithm to identify the mutation characteristics of the annual runoff process and the plateau period of the annual cumulative order sequence respectively;

[0067] Combined with the separation characteristics of the annual cumulative order sequence and the boundary conditions, predict extreme drought events;

[0068] Establish a quantitative relationship between the duration of the plateau period and the magnitude of its runoff, and predict the annual runoff change process.

[0069] The specific operation of performing rank transformation on the river runoff sequence using the sgn function is as follows:

[0070]

[0071] Rank(x i ) is the order value corresponding to the daily average flow on the i-th day in the sequence x. When the daily average flow on the i-th day is greater than or equal to that on the (i - 1)-th day, x iThe order value is 1. Conversely, when the average daily flow on the i-th day is less than that on the (i - 1)-th day, x i The order value is -1.

[0072] Specifically, the cumulative order sequence for calculating the runoff process is

[0073] By calculating the cumulative order sequence of river runoff, the continuous change characteristics of the runoff process are quantitatively characterized, and the interference of short-term fluctuations in average daily flow on the overall shape of the sequence is reduced.

[0074]

[0075] Rank’(x i ) is the cumulative order value of the runoff process corresponding to the i-th day. When Rank’(x i ) shows a linearly increasing trend over time, the average daily flow of the river continuously increases. Conversely, when Rank’(x i ) shows a linearly decreasing trend, the average daily flow of the river continuously decreases. For the runoff change process over a period of time, if the cumulative order value at the end of the sequence is greater than 0, it means that the duration of river rising water is greater than that of falling water; if the cumulative order value at the end is less than 0, it means that the duration of river falling water is greater than that of rising water.

[0076] Specifically, the sample deviation for calculating the cumulative value of water and sediment is

[0077] λ i =S' i -S'1 - a0Q' i (5)

[0078] λ i is the deviation of the cumulative value of the i-th sediment transport volume from the linear regression of the extreme value.

[0079] The principle of using the clustering algorithm to identify the mutation points of the multi-year runoff process is as follows: ① Starting from the head of the cumulative order sequence of average daily flow, slide the segmentation point and calculate the mean square error of the linear fitting of the left and right parts of the segmentation point respectively;

[0080]

[0081] a=(x n -x1) / n (4)

[0082] σ is the mean square error, a is the linear change rate of the cumulative order sequence, Rank’(x1) is the initial value of the cumulative order sequence sub-interval, n is the length of the sub-interval. The smaller the mean square error of the cumulative order sequence sub-interval, the higher the consistency of the change trend of the runoff sequence within the sub-interval.

[0083] σ'=σ1 + σ2 (5)

[0084] σ’ is the sum of the mean square deviations of each sub - interval. By determining the position of the segmentation point corresponding to the minimum sum of the mean square deviations, the mutation year of the runoff process is determined.

[0085] The specific method for the clustering algorithm to identify the plateau period of the annual cumulative order sequence is as follows. To refine the distribution characteristics of the annual runoff process, the annual runoff cumulative order sequence is calculated for each year, and the clustering algorithm is used to identify the characteristics of the annual runoff change. Affected by extreme climate and reservoir operation, there are three patterns in the annual runoff cumulative order sequence: ① Under normal conditions, the annual variation process of river runoff is affected by seasonal floods. The cumulative order during the non - flood season shows a continuous downward trend, and an upward trend during the flood season. Therefore, there is a plateau period in the annual runoff cumulative order sequence under normal conditions; ② In the case of extremely dry years, the characteristics of the flood season change of the river are not obvious, and there is no plateau period in the annual cumulative order sequence of the river; ③ Under the condition of reservoir operation, the reservoir's water storage operation makes the annual runoff process flatten, and the downward trend of the runoff during the non - flood season is significantly weakened. The plateau of the annual runoff cumulative order sequence curve is higher than that under normal conditions. Summarize the three types of annual cumulative order curves in turn.

[0086] To identify the plateau period of the annual runoff cumulative order sequence, the clustering algorithm is used to identify the optimal segmentation points at the start and end of the plateau period. The basic principle is as follows: ① In the annual runoff cumulative order sequence, two segmentation points are set successively; ② Starting from the head of the sequence, the two segmentation points are slid successively, and the mean square deviations of the three sub - intervals are calculated using Equation (3) in turn. The sum of their mean square deviations is used as the evaluation index for the clustering result.

[0087] σ” = σ1'+σ2'+σ3' (6)

[0088] σ” is the sum of the mean square deviations of the annual runoff cumulative order sequence. σ1’, σ2’, and σ3’ are the mean square deviations of the annual runoff cumulative order during the initial decline period, the plateau period, and the second decline period respectively. When the sum of the mean square deviations of the annual cumulative order sequence is the smallest, it represents that the selection position of the plateau period segmentation point is the optimal.

[0089] The specific method for predicting extreme drought events by combining the separation characteristics of the annual cumulative order sequence and boundary conditions is as follows. First, a linear formula is used to fit the runoff cumulative order curve under extreme drought conditions as the boundary condition of the annual cumulative order curve.

[0090] Rank'(x i ) = a×x i (6)

[0091] a is the slope parameter. By calculating the separation difference between the annual cumulative order curve and the boundary condition, the earliest prediction point of the occurrence of extreme drought conditions within the year is judged.

[0092] e i =Rank'(x i ) - a×xi (7)

[0093] e i is the separation difference between the annual cumulative order curve and the boundary condition. Under general conditions, the separation difference between the annual cumulative order curve and the boundary condition satisfies the S-shaped curve distribution.

[0094] Use the Logistic formula to fit the separation difference curve to obtain the annual variation law of the separation difference under general conditions.

[0095]

[0096] In the above formula, A, B, and C are all dimensionless parameters. Determine the first inflection point where the slope of the Logistic formula increases through formula (8), that is, the earliest identification point within the year for predicting drought years. Substitute formula (7) into formula (8) and simplify to obtain the distribution function of the annual cumulative order series under general conditions.

[0097]

[0098] For the three change stages of the annual cumulative order series of river runoff, use the power function to fit the relationship between the duration of the stage and the runoff volume in turn.

[0099] Q = k1×n1 (10)

[0100]

[0101] Q = k3×n3 (12)

[0102] In the above formula, Q is the runoff volume, n1, n2, and n3 are the durations of the initial decline period, the plateau period, and the secondary decline period respectively, k1 and k3 are the slope parameters of the initial decline period and the secondary decline period respectively, and k2 and m2 are the fitting parameters of the plateau period. Use formulas (10) to (12) to predict the annual runoff process.

[0103] Use the method for predicting the annual runoff process based on the rank theory proposed in this patent to calculate the mutation years of the runoff process in the Jinsha River Basin and the variation law of the annual runoff process. The area of the Jinsha River Basin is 500,000 km 2 , accounting for 27.8% of the total area of the Yangtze River Basin, and is an important source of water and sediment in the upper reaches of the Yangtze River. With the cascade hydropower development in the lower reaches of the Jinsha River, the runoff process at the basin outlet has changed significantly. The daily average flow process at the Xiangjiaba Station at the downstream outlet is used to characterize the runoff change in the Jinsha River. Calculate the annual runoff cumulative order series of the Jinsha River using formulas (1) and (2), and the results are shown in Figure 5 . The daily average flow of the Jinsha River from 1940 to 2020 is 4528.23 m 3 / s, and the variation range is 956 m 3 / s to 28600 m3 / s. The average order value of the daily average flow is -0.32, indicating that the overall order of the daily flow shows a downward trend during the multi-year change process. The duration of the daily flow decrease each year is much longer than that of the daily flow increase. Under the condition of long-term change, the cumulative order of the daily average flow generally shows a decaying trend, but the decay rate of the cumulative order significantly decreases around 1998. This phenomenon is closely related to the operation of the Ertan Hydropower Station in the Jinsha River Basin in 1998.

[0104] Calculate the cumulative order values of the annual runoff in the Jinsha River each year, and use the clustering algorithm to screen the years with morphological and trend mutation in the annual runoff process of the Jinsha River. The results are shown in Figure 6 . The results show that 1998 is the optimal segmentation point. The average value of the cumulative order from 1940 to 1997 is -148.45, and the average value from 1998 to 2020 is -39.68.

[0105] Based on different stages of the change in the daily average flow, use the clustering algorithm to identify the extremely dry years in the Jinsha River from 1940 to 2020 as 1945, 1946, 1949, and 1987. The linear fitting function of the annual cumulative order in the Jinsha River under extremely dry conditions is as follows:

[0106] Rank'(x i ) = -0.90×x i (13)

[0107] Calculate the separation difference between the annual runoff cumulative order under general conditions and extremely dry conditions according to the above formula. The results are shown in Figure 7 . The separation difference between the cumulative order series of the daily average flow in the Jinsha River from 1940 to 2020 and extremely dry conditions generally satisfies the S-shaped distribution. The separation difference of the cumulative order from 1940 to 1997 is significantly smaller than that from 1998 to 2020, indicating that the plateau period of the annual runoff process significantly extends after the operation of the Ertan Hydropower Station, and at the same time, the seasonality of the annual runoff change process weakens.

[0108] Fit the annual separation difference curve in the lower reaches of the Jinsha River through the Logistic formula to obtain the distribution function of the annual runoff cumulative order under different stage conditions.

[0109]

[0110] Equation (13) is the function of the annual cumulative order curve of the Jinsha River before the operation of the reservoir from 1940 to 1997, and equation (14) is the function of the annual cumulative order curve of the Jinsha River after the operation of the reservoir from 1998 to 2020. The fitting degrees of the above two equations are 0.97 and 0.98 respectively, indicating that the formula has high accuracy.

[0111] According to the annual cumulative curve function of the Jinsha River, different periods of the annual river changes are discriminated, and the quantitative relationships between the durations of each period within the year and the runoff are discussed in turn. The results are as Figure 8 shown. The functional formulas of the runoff change curves in different stages are as follows:

[0112] Q = 13.22×n (16)

[0113] Q = 8.72×10 -6 ×n 3.76 (17)

[0114] Q = 93.47×n (18)

[0115] The above three formulas correspond to the runoff curve functions of the initial decline period, the plateau period, and the secondary decline period in turn. The goodness-of-fit of the three is 0.93, 0.87, and 0.98 respectively, indicating that the fitting accuracy of the formula is relatively high.

[0116] The embodiment of the present application provides a system for predicting the annual runoff process of a river based on the rank theory. The system includes: a memory and a processor. The memory includes a program for the method of predicting the annual runoff process of a river based on the rank theory. When the program for the method of predicting the annual runoff process of a river based on the rank theory is executed by the processor, the following steps are implemented: Based on the daily average flow process of the river over the years, the rank transformation of the runoff sequence is performed using the sgn function; by calculating the cumulative order sequence of the runoff process, the clustering algorithm is used to identify the mutation characteristics of the runoff process over the years and the plateau period of the annual cumulative order sequence respectively; combining the separation characteristics of the annual cumulative order sequence and the boundary conditions, extreme drought events are predicted; a quantitative relationship between the duration of the plateau period and the magnitude of its runoff is established to predict the annual runoff change process.

[0117] The embodiment of the present application provides a computer-readable storage medium. The computer-readable storage medium stores program codes. When the program codes are executed by a processor, the steps of the method for predicting the annual runoff process of a river based on the rank theory as described above are implemented.

[0118] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0119] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and combinations of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing device generate means for implementing the specified functions in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for implementing the specified functions in one or more of the blocks.

[0120] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including instruction means that implement the specified functions in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for implementing the specified functions in one or more of the blocks.

[0121] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the specified functions in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for implementing the specified functions in one or more of the blocks.

[0122] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0123] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0124] A computer-readable medium includes permanent and non-permanent, removable and non-removable media that can implement information storage by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0125] The above are only embodiments of the present application and are not used to limit the protection scope of the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting river annual runoff process based on rank theory, characterized in that: The following steps are involved: Based on the multi-year average daily flow process of the river, the rank transformation of the runoff series was carried out using the sgn function. By calculating the cumulative order series of the runoff process, clustering algorithms are used to identify the mutation characteristics of the multi-year runoff process and the plateau period of the cumulative order series within the year. Combine the separation characteristics of the cumulative order series within the year with the boundary conditions to predict extreme drought events; Establish a quantitative relationship between the duration of the plateau period and its runoff volume to predict the runoff change process within the year.

2. The method for predicting river annual runoff process based on rank theory according to claim 1 is characterized in that: The specific method of using the sgn function to perform rank transformation on the river runoff sequence is: Rank(x i ) is the order value corresponding to the average daily flow of the i-th day in the sequence x. When the average daily flow of the i-th day is greater than or equal to the i-1 day, x i The order value of x is 1. On the contrary, when the average daily flow on the i-th day is less than that on the i-1th day, i The order value is -1.

3. The method for predicting river annual runoff process based on rank theory according to claim 1 is characterized in that: The cumulative order column of the runoff calculation process is specifically, By calculating the cumulative order sequence of river runoff, the continuous change characteristics of the runoff process can be quantitatively characterized, and the interference of short-term fluctuations in daily average flow on the overall morphology of the sequence can be reduced. Rank'(x i ) is the cumulative rank value of the runoff process corresponding to the i-th day. i ) increases linearly with time, the average daily flow of the river continues to increase. On the contrary, when Rank'(x i ) is a linear decreasing trend, the average daily flow of the river continues to decrease. For the runoff change process over a period of time, if the cumulative order value at the end of the sequence is greater than 0, it means that the duration of river flooding is greater than the duration of river falling; If the cumulative order value at the end is less than 0, it means that the duration of river flooding is greater than that of flooding.

4. The method for predicting river annual runoff process based on rank theory according to claim 1 is characterized in that: The sample deviation for calculating the water and sediment accumulation value is specifically: l i =S' i -S'1-a0Q' i (5) λ i is the deviation between the cumulative value of the ith sediment discharge and the extreme value linear regression.

5. The method for predicting river annual runoff process based on rank theory according to claim 1 is characterized in that: The clustering algorithm is used to identify the mutation point of the multi-year runoff process. The principle is as follows: ① Slide the segmentation point from the head of the daily average flow accumulation order sequence, and calculate the mean square error of the linear fitting of the left and right parts of the segmentation point respectively; a=(x n -x1) / n (4) σ is the mean square error, a is the linear change rate of the cumulative order series, Rank'(x1) is the initial value of the cumulative order series subinterval, and n is the length of the subinterval. The smaller the mean square error of the cumulative order series subinterval, the higher the consistency of the runoff sequence change trend in the subinterval. σ'=σ1+σ2 (5) σ' is the sum of the mean square errors of each sub-interval. The year of sudden change in runoff process is determined by determining the split point corresponding to the minimum sum of the mean square errors.

6. The method for predicting river annual runoff process based on rank theory according to claim 1 is characterized in that: The clustering algorithm identifies the platform period of the annual cumulative order sequence. Specifically, in order to finely distinguish the distribution characteristics of the annual runoff process, the runoff cumulative order sequence is calculated each year, and the annual runoff change characteristics are identified by the clustering algorithm. Affected by extreme climate and reservoir operation, there are three modes in the annual runoff cumulative order sequence: ① Under general conditions, the annual change process of river runoff is affected by seasonal floods, the cumulative order in the non-flood season is a continuous downward trend, and the flood season is an upward trend. Therefore, under general conditions, the annual runoff cumulative order sequence has a platform period; ② In extreme drought years, the change characteristics of the river flood season are not obvious, and the annual cumulative order sequence of the river does not have a platform period; ③ Under the reservoir operation conditions, the reservoir water storage operation leads to the flattening of the annual runoff process, and the runoff decline trend in the non-flood season is significantly weakened. The platform of the annual runoff accumulation order series curve is higher than that under general conditions. The three types of annual accumulation order curves are summarized in turn. In order to identify the platform period of the annual runoff accumulation sequence, a clustering algorithm is used to identify the optimal segmentation point for the start of the platform period and planting. The basic principle is: ① In the annual runoff accumulation sequence, two segmentation points are set successively; ② Starting from the head of the sequence, two segmentation points are slid in sequence, and the mean square error of the three sub-intervals is calculated in sequence using formula (3). The sum of the mean square errors of the three sub-intervals is used as the clustering result evaluation index. σ'=σ1'+σ2'+σ3' (6) σ” is the sum of the mean square errors of the cumulative order series of annual runoff, σ1', σ2', and σ3' are the mean square errors of the cumulative order series of annual runoff in the initial decline period, the platform period, and the secondary decline period, respectively. When the sum of the mean square errors of the cumulative order series of the year is the smallest, it means that the location of the platform period segmentation point is optimal.

7. The method for predicting river annual runoff process based on rank theory according to claim 1 is characterized in that: The method of combining the separation characteristics of the annual cumulative order series and the boundary conditions to predict extreme drought events specifically includes firstly fitting the runoff cumulative order curve under extreme drought conditions using a linear formula as the boundary condition of the annual cumulative order curve; Rank'(x i )=a×x i (6) a is the slope parameter. By calculating the difference between the cumulative order curve and the boundary conditions within a year, the earliest prediction point of extreme drought conditions within a year can be determined. e i =Rank'(x i )-a×x i (7) e i is the difference between the accumulated order curve and the boundary conditions in the year. Under normal conditions, the difference between the accumulated order curve and the boundary conditions in the year satisfies the S-curve distribution. The separation difference curve was fitted using the Logistic formula to obtain the annual variation pattern of the separation difference under general conditions. In the above formula, A, B, and C are dimensionless parameters. The first inflection point of the increase in the slope of the Logistic formula is determined by formula (8), which is the earliest identification point in the year for predicting drought years. Substituting formula (7) into formula (8) for simplification, the distribution function of the cumulative order series in the year under general conditions is obtained:

8. The method for predicting river annual runoff process based on rank theory according to claim 1 is characterized in that: For the three changing stages of the annual cumulative order series of river runoff, the power function is used to fit the relationship between the stage duration and runoff. Q=k1×n1 (10) Q=k3×n3 (12) In the above formula, Q is the runoff, n1, n2, and n3 are the duration of the initial decline period, the plateau period, and the secondary decline period, respectively; k1 and k3 are the slope parameters of the initial decline period and the secondary decline period, respectively; k2 and m2 are the fitting parameters of the plateau period. Formulas (10) to (12) are used to predict the annual runoff process.

9. A river annual runoff process prediction system based on rank theory, characterized in that: The system comprises: a memory and a processor, wherein the memory comprises a program of a method for predicting an annual runoff process of a river based on rank theory, and when the program of the method for predicting an annual runoff process of a river based on rank theory is executed by the processor, the following steps are implemented: based on the multi-year average daily flow process of the river, the runoff sequence is rank-transformed using the sgn function; by calculating the cumulative order sequence of the runoff process, a clustering algorithm is used to respectively identify the mutation characteristics of the multi-year runoff process and the plateau period of the annual cumulative order sequence; extreme drought events are predicted by combining the separation characteristics of the annual cumulative order sequence and boundary conditions; a quantitative relationship between the duration of the plateau period and its runoff volume is established to predict the annual runoff change process.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program codes, and when the program codes are executed by a processor, the steps of the method for predicting annual river runoff process based on rank theory as described in any one of claims 1 to 8 are implemented.

Images