Hydrological probability forecasting method and system based on time series full process processing

Abstract

The present invention provides a hydrological probability forecasting method and system based on the whole process processing of time series, which relates to the technical field of hydrological forecasting. The method includes: obtaining basic data and rainfall forecast data of the target basin, combining with the hydrological model, obtaining forecast flow data, and based on the measured flow and simulated flow data in the historical period, successively using the factor decomposition and data normal transformation methods, establishing the joint distribution function of the different components of the two one by one, and deriving the conditional distribution function of the measured value items with the forecast value items at the future moment as the condition, and finally obtaining the probability forecast results of the flow process at the future moment through random sampling and data reconstruction methods. The method of the present application fully considers the complexity of the composition of the hydrological sequence and the problem that there are obvious differences in the inherent change characteristics of different components, thereby not only improving the post-processing correction accuracy of the probability forecasting method, but also more accurately quantifying the uncertainty of the forecast results.

Description

Technical Field

[0001] The present invention relates to the technical field of hydrological forecasting, and in particular to a hydrological probability forecasting method and system based on full-process processing of time series. Background Art

[0002] Hydrological forecast is a quantitative or qualitative prediction of the state and changing patterns of hydrological elements in a certain period of time in the future based on known measurement or analysis information.

[0003] At present, there are generally two probabilistic forecasting methods at home and abroad. One is the coupling approach of uncertainty factors, which analyzes the main uncertainty factors in each link of the forecasting process, quantifies their uncertainties respectively, and then couples these uncertainties into the flood forecasting model; the other is the total error analysis approach, which does not directly deal with uncertainties such as input, model structure and parameters, but deals with the comprehensive error of hydrological forecasting, that is, starting from the deterministic forecast results, analyzing the total error between the forecast results and the actual values, and constructing the mathematical description equation of the deterministic model output and the actual flood process by using mathematical statistics methods, and deriving the distribution function of the actual value under the condition of the deterministic forecast value to achieve probabilistic forecasting.

[0004] However, existing hydrological probabilistic forecasting methods directly perform post-processing corrections on the deterministic forecast results themselves, using a unified function form to quantify forecast errors and uncertainties. They do not fully consider the complexity of the composition of the hydrological sequence and the obvious differences in the changing characteristics of different components, resulting in the correction accuracy of extreme values ​​in the sequence being much lower than the general value. Summary of the invention

[0005] The purpose of the present invention is to provide a hydrological probability forecasting method and system based on the whole process processing of time series, so as to solve the technical problem of insufficient post-processing correction accuracy of the hydrological probability forecasting method existing in the prior art.

[0006] In a first aspect, the present invention provides a hydrological probability forecasting method based on full-process time series processing, comprising:

[0007] Obtain basic data and rainfall forecast data of the target basin, where the basic data includes hydrological and meteorological data and geological characteristic data;

[0008] Based on the basic data, a hydrological model corresponding to the target basin is established to obtain the measured flow data and simulated flow data of the historical period; the rainfall forecast data is input into the hydrological model to obtain the forecast flow data of the target basin at the future time;

[0009] Based on the simulation results of the hydrological model, the measured flow data and simulated flow data of the target basin are decomposed to obtain multi-order smoothing series and multi-order residual series of the measured flow data and simulated flow data;

[0010] Based on the multi-order smoothed sequences and multi-order residual sequences of the measured flow data and simulated flow data, the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and simulated flow data is established item by item;

[0011] Based on the joint distribution function of each smoothing sequence and residual sequence of measured flow data and simulated flow data, the forecast flow data at future moments is post-processed and corrected to obtain a set of post-processed samples of each item of forecast flow data at future moments in real space;

[0012] The data of each post-processing sample set of the forecast flow data at the future time is reconstructed to obtain the probability forecast result of the flow process at the future time.

[0013] In an optional embodiment, establishing a hydrological model corresponding to the target watershed based on basic data also includes:

[0014] Obtain historical measured flow data for the target watershed;

[0015] The hydrological model is calibrated and verified based on historical measured flow data.

[0016] In an optional embodiment, based on the simulation results of the hydrological model, the measured flow data and the simulated flow data of the target basin in the historical period are decomposed to obtain multi-order smoothing sequences and multi-order residual sequences of the measured flow data and the simulated flow data, including:

[0017] Get the sliding period;

[0018] Based on the sliding period, the measured flow data and the simulated flow data are iteratively decomposed respectively until it is determined that the iteration termination condition is met, thereby obtaining multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data; wherein the decomposition operation includes: based on the sliding period, sliding average processing is performed on the measured flow data sequence and the simulated flow data sequence respectively to obtain the first-order smoothed sequences of the measured flow data sequence and the simulated flow data sequence, and based on the first-order smoothed sequence and the measured flow data sequence or the simulated flow data sequence, the first-order residual sequences of the measured flow data sequence and the simulated flow data sequence are obtained; the first-order residual sequence is used as the measured flow data sequence or the simulated flow data sequence in the next iteration process to perform the decomposition operation.

[0019] In an optional embodiment, the iteration termination condition is:

[0020]

[0021] Where: k j is the sliding period at the end, j is the number of iterations at the end, and it is an integer, k1 is the initial sliding period.

[0022] In an optional embodiment, based on the multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data, the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data is established item by item, including:

[0023] Perform data normal transformation on the multi-order smoothing series and multi-order residual series of measured flow data and simulated flow data;

[0024] Based on the multi-order smoothed sequences and multi-order residual sequences of the transformed measured flow data and simulated flow data, the bivariate normal distribution functions of each order of smoothed sequences and residual sequences are established item by item.

[0025] Based on the maximum likelihood method, the parameters of the bivariate normal distribution function are estimated.

[0026] In an optional embodiment, based on the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data, the future time forecast flow data is post-processed and corrected to obtain a set of post-processed samples of each item of the future time forecast flow data in the real space, including:

[0027] Preprocess the forecast flow data for future moments;

[0028] Based on the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and simulated flow data, the conditional distribution function of each item of the measured flow data value is obtained with the forecast value as the condition. Through random sampling and inverse data transformation, the post-processing sample set of each item of the forecast data in the real space is obtained.

[0029] In an optional embodiment, data reconstruction is performed on each post-processing sample set of the forecast flow data at the future time to obtain a probability forecast result of the flow process at the future time, including:

[0030] Obtain measured flow data for historical periods;

[0031] Based on the multi-order smoothing terms and residual terms after decomposition of measured flow data, a pattern matrix is ​​established;

[0032] Based on the ranking number of each element in the pattern matrix in the ascending series of the corresponding column vector, a permutation rank matrix is ​​obtained;

[0033] Sort each post-processing sample set of the forecast flow data to obtain a sample matrix;

[0034] Taking the permutation rank matrix as the basic structure, reconstruct the elements of each column of the sample matrix to obtain the reconstructed matrix;

[0035] By superimposing the elements of each row of the reconstruction matrix, the probability forecast results of the flow process at future moments are obtained.

[0036] In a second aspect, an embodiment of the present invention provides a hydrological probability forecasting system based on full-process time series processing, including:

[0037] An acquisition module is used to obtain basic data and rainfall forecast data of the target basin;

[0038] The forecast module is used to establish a hydrological model corresponding to the target basin based on basic data, obtain the measured flow data and simulated flow data in the historical period; input the rainfall forecast data into the hydrological model to obtain the forecast flow data of the target basin at the future time;

[0039] A decomposition module is used to decompose the measured flow data and simulated flow data of the target basin based on the simulation results of the hydrological model, and obtain multi-order smoothing sequences and multi-order residual sequences of the measured flow data and simulated flow data;

[0040] Establishing a module, for establishing the joint distribution function of each smoothing sequence and residual sequence of the measured flow data and the simulated flow data item by item based on the multi-order smoothing sequence and multi-order residual sequence of the measured flow data and the simulated flow data;

[0041] A correction module is used to perform post-processing correction on the forecast flow data at future times based on the joint distribution function of each smoothing sequence and residual sequence of the measured flow data and the simulated flow data, so as to obtain a post-processing sample set of each item of the forecast flow data at future times in the real space;

[0042] The reconstruction module is used to reconstruct the data of each post-processing sample set of the forecast flow data at the future time to obtain the probability forecast result of the flow process at the future time.

[0043] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0044] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to execute any one of the methods in the first aspect above.

[0045] The present invention provides a hydrological probability forecasting method and system based on the whole process processing of time series. The basic data and rainfall forecast data of the target basin are obtained, combined with the hydrological model, to obtain the forecast flow data, and based on the measured flow and simulated flow data in the historical period, the factor decomposition and data normal transformation methods are used in turn to establish the joint distribution function of the different components of the two one by one, and the conditional distribution function of the measured value items is derived with the forecast value items at the future moment as the condition, and finally the probability forecast results of the flow process at the future moment are obtained through random sampling and data reconstruction methods. This method fully considers the complexity of the hydrological sequence composition and the obvious differences in the inherent change characteristics of different components, thereby not only improving the post-processing correction accuracy of the probability forecasting method, but also more accurately quantifying the uncertainty of the forecast results. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A schematic flow chart of a hydrological probability forecasting method based on full-process time series processing provided by an embodiment of the present invention;

[0048] Figure 2 A comparison chart of flow processes during the calibration, verification and forecast periods of the hydrological model provided by an embodiment of the present invention;

[0049] Figure 3 A schematic diagram of each component data after factoring the measured flow data provided by an embodiment of the present invention;

[0050] Figure 4 A schematic diagram of a probability forecast result of a flow process at a future moment provided by an embodiment of the present invention;

[0051] Figure 5 A schematic diagram of the structure of a hydrological probability forecasting system based on full-process time series processing provided by an embodiment of the present invention;

[0052] Figure 6 A structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0054] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] At present, the existing hydrological probability methods are all directly post-processing corrections to the deterministic forecast results themselves, using a unified function form to quantify the forecast error and uncertainty, without fully considering the complexity of the hydrological sequence composition and the obvious differences in the change characteristics of different components, thus leading to the technical problem of insufficient correction accuracy. Based on this, the present invention provides a hydrological probability forecasting method and system based on the whole process of time series processing, by collecting basic data such as basin hydrological meteorology, topography, land use, vegetation and soil, and precipitation forecast products, a hydrological model is established, and the flow at future times is predicted, the measured flow in the historical period and the model forecast flow data are factored and the data is normally transformed, the joint distribution function of different components is established, and the conditional distribution function of each item of the measured flow data value is derived with the forecast flow data value at future times as the condition; through the method of random sampling and data reconstruction, the probability forecast result of the flow at future times is obtained, and the present application fully considers the complexity of the hydrological sequence composition and the obvious differences in the intrinsic change characteristics of different components, which not only improves the post-processing correction accuracy, but also more accurately quantifies the uncertainty of the forecast results.

[0056] Figure 1 The present invention provides a flow chart of a hydrological probability forecasting method based on the whole process of time series processing. Figure 1 As shown, the method comprises the following steps:

[0057] Step S110: Obtain basic data and rainfall forecast data of the target watershed.

[0058] The basic data include hydrological and meteorological data and geological characteristic data. The geological characteristic data are the target basin characteristic data such as topography, land use, vegetation and soil in the target basin; the hydrological and meteorological data mainly include rainfall data, temperature data, evaporation data and flow data.

[0059] Furthermore, basic data can be obtained according to the type of hydrological model. For example, conceptual hydrological models such as the Xin'anjiang model and the water tank model can be used to obtain hydrological and meteorological data such as rainfall, runoff, and evapotranspiration. Distributed hydrological models such as the SWAT model and the VIC model can also be used to obtain data on terrain, land use, vegetation, soil, and hydrological and meteorological data such as rainfall, runoff, and evapotranspiration.

[0060] Specifically, rainfall data and temperature data can be collected from representative meteorological stations in the target watershed, and spatial interpolation methods such as bilinear interpolation, Kriging and reciprocal distance square method can be used to obtain rainfall data and temperature data in any area of ​​the target watershed;

[0061] Evaporation data can be obtained from the measured data of evaporation pans in the target watershed;

[0062] Flow data refers to the measured flow at the hydrological station at the outlet section of the target basin, which can be obtained from the hydrological yearbook.

[0063] Rainfall forecast data can be obtained by collecting rainfall forecast products on the TIGGE (The Interactive Grand Global Ensemble) platform. The TIGGE platform integrates forecast products released by major meteorological agencies around the world, including the Australian Bureau of Meteorology (BOM), China Meteorological Administration (CMA), National Centers for Environmental Prediction (NCEP), Japan Meteorological Agency (JMA) and European Center for Medium-Range Weather Forecasts (ECMWF). The type of rainfall forecast data to be obtained is mainly selected in combination with the forecast period. For example, if the forecast period is short, numerical weather forecast products should be used, and if the forecast period is long, climate forecast products should be used.

[0064] In this application, the rainfall forecast product is the GEFS (Global Ensemble Forecast System) data released by the National Center for Environmental Prediction of the United States, which covers information such as rainfall, temperature and humidity around the world, with a spatial resolution of 0.469° and a total of 11 ensemble members. In the specific implementation of this application, the average of the 11 ensemble members is taken, and the forecast precipitation of each area in the target basin is interpolated and calculated in combination with the bilinear interpolation method.

[0065] In some embodiments, step S110 further includes: preprocessing the basic data and rainfall forecast data.

[0066] Specifically, the preprocessing of basic data and rainfall forecast data mainly includes preprocessing in time and space, so that the data format of the basic data and rainfall forecast data conforms to the data format required by the hydrological model to ensure the accuracy and reliability of the hydrological model.

[0067] In terms of time, the basic data and rainfall forecast data are preprocessed to obtain data of the same time scale, such as hour, day, month and year scales;

[0068] In terms of space, the basic data and rainfall forecast data are preprocessed. When a distributed hydrological model is used, it is necessary to combine the multiple spatial units set up in the hydrological model and spatially interpolate basic data such as rainfall, temperature, land use, vegetation and soil to each spatial unit; when a conceptual hydrological model is used, it is necessary to calculate the surface average value of the entire target basin based on the basic data.

[0069] Step S120: Based on the basic data, a hydrological model corresponding to the target basin is established to obtain the measured flow data and model flow data of the historical period; and the rainfall forecast data is input into the hydrological model to obtain the forecast flow data of the target basin at a future time.

[0070] In some embodiments, in step S120, it further includes:

[0071] Obtain historical measured flow data for the target watershed;

[0072] The hydrological model is calibrated and verified based on historical measured flow data.

[0073] Among them, the hydrological model can be calibrated and verified by taking the optimal Nash efficiency coefficient as the objective function:

[0074]

[0075] Where NSE is an indicator for evaluating the prediction accuracy of the hydrological model. is the measured flow data at the i-th moment, is the forecast flow data at the i-th moment, is the mean of the measured flow rate.

[0076] Furthermore, the larger the NSE value is, the more accurate the simulation effect of the hydrological model is.

[0077] Furthermore, the rainfall data of the historical period is input into the established hydrological model to obtain the simulated flow data. The measured flow data and simulated flow data of the historical period are used to determine whether the hydrological model is suitable for the current target basin.

[0078] like Figure 2 As shown in Figure 2, the rainfall forecast data is input into the hydrological model to obtain the forecast flow data of the target basin at future times. Figure 2 It provides hydrological model calibration, verification results and forecast results of flow processes in the future.

[0079] Step S130: Based on the simulation results of the hydrological model, the measured flow data and the simulated flow data of the target basin are decomposed respectively to obtain multi-order smoothing sequences and multi-order residual sequences of the measured flow data and the simulated flow data.

[0080] In some embodiments, step S130 includes:

[0081] Get the sliding period;

[0082] Based on the sliding period, the measured flow data and the simulated flow data are iteratively decomposed respectively until it is determined that the iteration termination condition is met, thereby obtaining multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data; wherein the decomposition operation includes: based on the sliding period, sliding average processing is performed on the measured flow data sequence and the simulated flow data sequence respectively to obtain the first-order smoothed sequences of the measured flow data sequence and the simulated flow data sequence, and based on the first-order smoothed sequence and the measured flow data sequence or the simulated flow data sequence, the first-order residual sequences of the measured flow data sequence and the simulated flow data sequence are obtained; the first-order residual sequence is used as the measured flow data sequence or the simulated flow data sequence in the next iteration process to perform the decomposition operation.

[0083] In some embodiments, in step S130, it further includes:

[0084] The iteration termination condition is:

[0085]

[0086] Where: k j is the sliding period at the end, j is the number of iterations at the end, and it is an integer, k 1 is the initial sliding period.

[0087] Specifically, the measured flow data and simulated flow data of the site in the historical period are collected, wherein the sequence of the measured flow data is: Y = {y 1 ,y 2 ,...,y t}; The sequence of simulated flow data is: X = {x 1 ,x 2 ,...,x t In this application, taking the sequence of measured flow data as an example, the sliding average method is used to decompose the time series to obtain a multi-order smoothing sequence and a multi-order residual sequence of the measured flow data. The specific decomposition process is as follows:

[0088] Step 1: Get the first sliding period k 1 =30, the measured flow data sequence is processed by sliding average to obtain the first-order smoothed sequence of the measured flow data sequence, as follows:

[0089]

[0090] Where Y 1 is the first-order smoothed sequence of measured flow data, T is the matrix transposition symbol, column vector is converted to row vector, w 1 For the corresponding weight, take w 1 ~w 30 =1 / k 1 .

[0091] Step 2: Based on the measured flow data series Y and the first-order smoothing series Y 1 , and obtain the first-order residual value sequence That is, the measured flow data series Y minus the first-order smoothing series Y 1 , as shown in the following formula:

[0092]

[0093] Step 3: For the first-order residual value series Take the second sliding period k 2 =15, the first-order residual value sequence is processed by sliding average to obtain the second-order smoothed sequence of the measured flow data sequence The specific formula is as follows:

[0094]

[0095] In the formula, w 1 ~w 15 For the corresponding weight, take w 1 ~w 15 =1 / k 2

[0096] Step 4: Based on the first-order residual value series With the second-order smooth series Y 2 , and obtain the second-order residual sequence That is, the first-order residual value series Subtract the second-order smoothed series Y 2 , as shown in the following formula:

[0097]

[0098] Step 5: Repeat steps 3 to 4 until the jth sliding average processing. Then stop the iteration.

[0099] In the embodiment of the present application, a total of 5 iterations are performed, corresponding to sliding periods of 30, 15, 8, 4 and 2, respectively, and finally a fifth-order smoothed sequence Y is obtained. 5 and the fifth-order residual sequence

[0100]

[0101] Step 6: Based on steps 1 to 5, the measured flow data sequence can be decomposed into 6 items, such as Figure 3 As shown:

[0102]

[0103] Similarly, referring to the decomposition process of the measured flow data series above, the simulated flow data series can be decomposed into 6 items:

[0104]

[0105] Since the hydrological series is composed of multiple deterministic components and random components, this application uses the moving average method to decompose the measured flow data series and the simulated flow data series in the time series, which can fully reflect the complexity of the sequence composition and the differences in the changes of different components on the time scale. The subsequent post-processing correction of different components will help improve the accuracy of post-processing.

[0106] Step S140: Based on the multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data, establish the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data item by item.

[0107] In some embodiments, in step S140, it includes:

[0108] Perform data normal transformation on the multi-order smoothing series and multi-order residual series of measured flow data and simulated flow data;

[0109] Based on the multi-order smoothed sequences and multi-order residual sequences of the transformed measured flow data and simulated flow data, the bivariate normal distribution functions of each order of smoothed sequences and residual sequences are established item by item.

[0110] Based on the maximum likelihood method, the parameters of the bivariate normal distribution function are estimated.

[0111] Among them, the data normal transformation methods include Box-Cox method, normal quantile transformation method, Log-Sinh method and Yeo-Johnson method. When the data are all positive numbers, any of the above data normal transformation methods can be used directly; when the data contains both positive and negative numbers, the normal quantile transformation method or Yeo-Johnson method should be used to process the data. Among them, in this application, the Yeo-Johnson method is used to perform data normal transformation on the smoothing term and the residual term after the decomposition of the measured flow data in step S130. The method is defined as follows:

[0112]

[0113] In the formula, y and are the data series in real space and normal space respectively.

[0114] Furthermore, according to the data sequence after data transformation, the bivariate normal distribution function of each order smoothing sequence and residual sequence is established item by item, recorded as: F~N(μ,Σ); that is, Taking the first-order smoothing term as an example, the transformed data is recorded as and The corresponding joint distribution function (i.e., bivariate normal distribution function) is as follows:

[0115]

[0116] In the formula, μ is the mean vector, and In the bivariate normal distribution and The corresponding mean parameter, Σ is the covariance matrix, and for and The variance parameter, ρ, is and The correlation coefficient of for and The product of the mean square error parameters.

[0117] Based on the above bivariate normal distribution function form, the likelihood function is constructed as follows, and the maximum likelihood method is used to estimate the parameters of the bivariate normal distribution function:

[0118]

[0119] In the formula, φ BN is the two-dimensional normal distribution density function, is the estimated parameter value.

[0120] Step S150: Based on the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data, the forecast flow data for the future time is post-processed and corrected to obtain a set of post-processed samples of each item of the forecast flow data for the future time in the real space.

[0121] In some embodiments, in step S150, it includes:

[0122] Preprocess the forecast flow data for future moments;

[0123] Based on the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and simulated flow data, the conditional distribution function of each item of the measured flow data value is obtained with the forecast value as the condition. Through random sampling and inverse data transformation, the post-processing sample set of each item of the forecast data in the real space is obtained.

[0124] Specifically, step S150 mainly includes three aspects: conditional distribution function derivation, random sampling and data inverse normal transformation.

[0125] The predicted flow data x at the future time t+1 For example, the data x t+1 After factor decomposition and data normal transformation, combined with the joint distribution function established in step S140, the forecast flow data x is derived. t+1 The conditional distribution function of each item of measured flow data is as follows:

[0126]

[0127] Where: is the conditional distribution function, and is the corresponding parameter of the conditional distribution function.

[0128] Based on this derivation process, other item conditional distribution functions can be derived.

[0129] Furthermore, in this application, the conditional distribution function of each item is sampled by using the equidistant quantile sampling method to obtain a post-processing sample set of each item in the normal space, and the sample number Num=100.

[0130] Furthermore, the extracted sample set is subjected to an inverse normal transformation, which can be achieved by converting it from the normal space to the real space, that is, a post-processing sample set of each item of the forecast flow data at the future moment in the real space.

[0131] Step S160: reconstruct the data of each post-processing sample set of the forecast flow data to obtain the probability forecast result of the flow process at the future moment.

[0132] In some embodiments, in step S160, it includes:

[0133] Obtain measured flow data for historical periods;

[0134] Based on the multi-order smoothing terms and residual terms after decomposition of measured flow data, a pattern matrix is ​​established;

[0135] Based on the ranking number of each element in the pattern matrix in the ascending series of the corresponding column vector, a permutation rank matrix is ​​obtained;

[0136] Sort each post-processing sample set of the forecast flow data to obtain a sample matrix;

[0137] Taking the permutation rank matrix as the basic structure, reconstruct the elements of each column of the sample matrix to obtain the reconstructed matrix;

[0138] By superimposing the elements of each row of the reconstruction matrix, the probability forecast results of the flow process at future moments are obtained.

[0139] Specifically, step S160 mainly includes three aspects: screening and reconstructing pattern data, statistical arrangement rank matrix, and post-processing sample set reconstruction of each item.

[0140] Among them, to obtain the measured flow data in the historical period, the measured flow data in the adjacent time period of the forecast flow data to be post-processed and the measured flow data in the adjacent time period of the same historical period should be screened, and data of equivalent magnitude should be selected.

[0141] In this application, 100 measured data are selected from the screening range, and all the screened measured flow data are recorded as y i,模 (i=1,2,...,100), and its smoothing term and residual term data form the pattern matrix Y 模 , as follows:

[0142]

[0143] Where: Y 模 The pattern matrix contains 100 rows, each row corresponding to the smoothing term and residual term of the filtered data.

[0144] Statistical pattern matrix Y 模 The sorting numbers of each element in the ascending series of the corresponding column vector form a new matrix Rank (i.e., permutation rank matrix), as follows:

[0145]

[0146] Where: for In vector The serial number in ascending order, the same applies to other elements of the matrix.

[0147] Arrange the post-processing sample set obtained in step S150 in ascending order to form a post-processing sample matrix Y 后处理 , as follows:

[0148]

[0149] in, The same goes for the other columns.

[0150] Then, based on the permutation rank matrix Rank, the post-processing sample matrix Y 后处理 Each column element is reconstructed to form a new reconstruction matrix Y 重构 , as follows:

[0151]

[0152] Finally, the reconstructed matrix Y 重构 The elements of each row are superimposed to obtain the probability forecast result of the future time t+1 period, which is recorded as Y t+1 ={y t+1 (1),y t+1 (2),...,y t+1 (100)}, where y t+1 (1) The calculation is as follows:

[0153]

[0154] Similarly, other samples refer to y t+1 (1) Calculation.

[0155] For the post-processing result set Y t+1 The average value or median can be taken as the corrected deterministic forecast result, and the confidence interval corresponding to any significance level α can also be analyzed, which is the corresponding uncertainty interval.

[0156] like Figure 4 As shown, according to the method of the present invention, the post-processing result set Y t+1 The median of the forecasts was taken as the deterministic forecast outcome, and the corresponding 90% uncertainty interval was extracted.

[0157] In summary, the present invention provides a hydrological probability forecasting method and system based on the whole process processing of time series. By collecting basic data such as basin hydrometeorology, topography, land use, vegetation and soil, as well as precipitation forecast products, a hydrological model is established, and the flow at future times is predicted. On this basis, the measured flow and simulated flow data in historical periods are factored and the data is normally transformed, and the joint distribution function of different components is established. The conditional distribution function of each item of the measured flow data value is derived with each item of the forecast flow data value at future times as the condition; through random sampling and data reconstruction methods, the probability forecast results of the flow at future times are obtained. This method fully considers the complexity of the composition of the hydrological sequence and the obvious differences in the change characteristics of different components. While improving the accuracy of the post-processing correction of the probability forecasting method, it can more accurately quantify the uncertainty of the forecast results.

[0158] Figure 5 The present invention provides a schematic diagram of a hydrological probability forecasting system based on the whole process of time series processing. Figure 5 As shown, the system includes:

[0159] An acquisition module 210 is used to acquire basic data and rainfall forecast data of a target watershed;

[0160] The forecast module 220 is used to establish a hydrological model corresponding to the target basin based on the basic data, obtain the measured flow data and simulated flow data of the historical period; input the rainfall forecast data into the hydrological model to obtain the forecast flow data of the target basin at the future time;

[0161] A decomposition module 230 is used to decompose the measured flow data and the simulated flow data of the target basin based on the simulation results of the hydrological model, respectively, to obtain a multi-order smoothing sequence and a multi-order residual sequence of the measured flow data and the simulated flow data;

[0162] Establishing module 240, for establishing joint distribution functions of each smoothing sequence and residual sequence of the measured flow data and the simulated flow data item by item based on the multi-order smoothing sequence and multi-order residual sequence of the measured flow data and the simulated flow data;

[0163] The correction module 250 is used to perform post-processing correction on the future forecast flow data based on the joint distribution function of each smoothing sequence and residual sequence of the measured flow data and the simulated flow data, so as to obtain a post-processing sample set of each item of the future forecast flow data in the real space;

[0164] The reconstruction module 260 is used to reconstruct data of each post-processing sample set of the forecast flow data at the future time to obtain the probability forecast result of the flow process at the future time.

[0165] In some embodiments, the forecast module 220 is specifically used to:

[0166] Obtain historical measured flow data for the target watershed;

[0167] The hydrological model is calibrated and verified based on historical measured flow data.

[0168] In some embodiments, the decomposition module 230 is specifically configured to:

[0169] Get the sliding period;

[0170] Based on the sliding period, the measured flow data and the simulated flow data are iteratively decomposed respectively until it is determined that the iteration termination condition is met, thereby obtaining multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data; wherein the decomposition operation includes: based on the sliding period, sliding average processing is performed on the measured flow data sequence and the simulated flow data sequence respectively to obtain the first-order smoothed sequences of the measured flow data sequence and the simulated flow data sequence, and based on the first-order smoothed sequence and the measured flow data sequence or the simulated flow data sequence, the first-order residual sequences of the measured flow data sequence and the simulated flow data sequence are obtained; the first-order residual sequence is used as the measured flow data sequence or the simulated flow data sequence in the next iteration process to perform the decomposition operation.

[0171] In some embodiments, the decomposition module 230 is specifically configured to:

[0172] The iteration termination condition is:

[0173]

[0174] Where: k j is the sliding period at the end, j is the number of iterations at the end, and it is an integer, k 1 is the initial sliding period.

[0175] In some embodiments, the establishment module 240 is specifically used to:

[0176] Perform data normal transformation on the multi-order smoothing series and multi-order residual series of measured flow data and simulated flow data;

[0177] Based on the multi-order smoothed sequences and multi-order residual sequences of the transformed measured flow data and simulated flow data, the bivariate normal distribution functions of each order of smoothed sequences and residual sequences are established item by item.

[0178] Based on the maximum likelihood method, the parameters of the bivariate normal distribution function are estimated.

[0179] In some embodiments, the correction module 250 is specifically used to:

[0180] Preprocess the forecast flow data for future moments;

[0181] Based on the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and simulated flow data, the conditional distribution function of each item of the measured flow data value is obtained with the forecast value as the condition. Through random sampling and inverse data transformation, the post-processing sample set of each item of the forecast data in the real space is obtained.

[0182] In some embodiments, the reconstruction module 260 is specifically used to:

[0183] Obtain measured flow data for historical periods;

[0184] Based on the multi-order smoothing terms and residual terms after decomposition of measured flow data, a pattern matrix is ​​established;

[0185] Based on the ranking number of each element in the pattern matrix in the ascending series of the corresponding column vector, a permutation rank matrix is ​​obtained;

[0186] Sort each post-processing sample set of the forecast flow data to obtain a sample matrix;

[0187] Taking the permutation rank matrix as the basic structure, reconstruct the elements of each column of the sample matrix to obtain the reconstructed matrix;

[0188] By superimposing the elements of each row of the reconstruction matrix, the probability forecast results of the flow process at future moments are obtained.

[0189] The implementation principle and technical effects of the system provided in the embodiment of the present application are the same as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the system embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0190] like Figure 6 As shown, an electronic device 600 provided in an embodiment of the present application includes: a processor 601, a memory 602 and a bus, the memory 602 stores machine-readable instructions executable by the processor 601, when the electronic device is running, the processor 601 and the memory 602 communicate through the bus, and the processor 601 executes the machine-readable instructions to perform the steps of the hydrological probability forecasting method based on the whole process of time series processing as mentioned above.

[0191] Specifically, the above-mentioned memory 602 and processor 601 can be general-purpose memories and processors, which are not specifically limited here. When the processor 601 runs the computer program stored in the memory 602, it can execute the above-mentioned hydrological probability forecasting method based on the whole process processing of time series.

[0192] The processor 601 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 601. The above processor 601 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module may be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 602, and the processor 601 reads the information in the memory 602 and completes the steps of the above method in combination with its hardware.

[0193] Corresponding to the above-mentioned hydrological probability forecasting method based on the whole process processing of time series, an embodiment of the present application also provides a computer-readable storage medium, which stores machine-executable instructions. When the computer-executable instructions are called and executed by the processor, the computer-executable instructions prompt the processor to execute the steps of the above-mentioned hydrological probability forecasting method based on the whole process processing of time series.

[0194] The hydrological probability forecasting system based on the whole process of time series processing provided in the embodiment of the present application can be specific hardware on the device or software or firmware installed on the device. The device provided in the embodiment of the present application, its implementation principle and the technical effect produced are the same as those in the aforementioned method embodiment. For the sake of brief description, the parts not mentioned in the device embodiment can refer to the corresponding contents in the aforementioned method embodiment. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can all refer to the corresponding processes in the aforementioned method embodiment, and will not be repeated here.

[0195] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0196] For another example, the flowchart and block diagram in the accompanying drawings show the possible architecture, functions and operations of the apparatus, method and computer program product according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and a part of the module, program segment or code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0197] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0198] In addition, each functional unit in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0199] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions for an electronic device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the vehicle marking method of each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0200] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0201] Finally, it should be noted that the above embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application. They should all be covered within the protection scope of the present application.

Claims

1. A hydrological probability forecasting method based on full-process time series processing, characterized in that: include: Obtaining basic data and rainfall forecast data of the target watershed, wherein the basic data includes hydrological and meteorological data and geological characteristic data; Based on the basic data, a hydrological model corresponding to the target watershed is established to obtain measured flow data and simulated flow data in the historical period; the rainfall forecast data is input into the hydrological model to obtain forecast flow data of the target watershed at a future time; Based on the simulation results of the hydrological model, the measured flow data and the simulated flow data of the target basin are decomposed respectively to obtain multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data; wherein, obtaining the multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data includes: obtaining a sliding period; based on the sliding period, iteratively decomposing the measured flow data and the simulated flow data respectively until it is determined that the iteration termination condition is met, and obtaining the multi-order smoothed sequences of the measured flow data and the simulated flow data. The decomposition operation comprises: based on the sliding period, respectively performing sliding average processing on the sequence of the measured flow data and the sequence of the simulated flow data to obtain the first-order smoothed sequence of the measured flow data sequence and the simulated flow data sequence, and based on the first-order smoothed sequence and the measured flow data sequence or the simulated flow data sequence, obtaining the first-order residual sequence of the measured flow data sequence and the simulated flow data sequence; the first-order residual sequence is used as the measured flow data sequence or the simulated flow data sequence in the next iteration process to perform the decomposition operation; Based on the multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data, establishing the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data item by item; Based on the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data, post-processing correction is performed on the future time forecast flow data to obtain a post-processing sample set of each item of the future time forecast flow data in the real space; Data reconstruction is performed on each post-processing sample set of the forecast flow data at the future time to obtain a probability forecast result of the flow process at the future time.

2. The hydrological probability forecasting method based on time series full process processing according to claim 1 is characterized in that: Based on the basic data, a hydrological model corresponding to the target watershed is established, further comprising: Obtaining historical measured flow data of the target watershed; Based on the historical measured flow data, the hydrological model is calibrated and verified.

3. The hydrological probability forecasting method based on time series full process processing according to claim 1 is characterized in that: The iteration termination condition is: Where: is the sliding period at the end, j is the number of iterations at the end, and it is an integer. is the initial sliding period.

4. The hydrological probability forecasting method based on time series full process processing according to claim 1 is characterized in that: Based on the multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data, a joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data is established item by item, including: Performing data normal transformation on the multi-order smoothing sequence and the multi-order residual sequence of the measured flow data and the simulated flow data; Based on the transformed multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data, bivariate normal distribution functions of the smoothed sequences and residual sequences of each order are established item by item; Based on the maximum likelihood method, parameters of the bivariate normal distribution function are estimated.

5. The hydrological probability forecasting method based on time series full process processing according to claim 1 is characterized in that: Based on the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data, the future forecast flow data is post-processed and corrected to obtain a set of post-processed samples of each item of the future forecast flow data in real space, including: Preprocessing the forecast flow data at the future time; Based on the joint distribution function of each smoothed sequence and residual sequence of the measured flow data and the simulated flow data, the conditional distribution function of each item of the measured flow data value is obtained with each item of the forecast value as the condition, and through random sampling and inverse data transformation, the post-processing sample set of each item of the forecast data in the real space is obtained.

6. The hydrological probability forecasting method based on time series full process processing according to claim 1 is characterized in that: Reconstructing data of each post-processing sample set of the forecast flow data at the future time to obtain a probability forecast result of the flow process at the future time, including: Obtain measured flow data for historical periods; Establishing a pattern matrix based on the multi-order smoothing terms and residual terms after decomposition of the measured flow data; Based on the ranking number of each element in the pattern matrix in the ascending series of the corresponding column vector, a permutation rank matrix is ​​obtained; Sorting each post-processing sample set of the forecast flow data to obtain a sample matrix; Taking the permutation rank matrix as the basic structure, reconstructing the elements of each column of the sample matrix to obtain a reconstructed matrix; The elements of each row of the reconstruction matrix are superimposed to obtain the probability forecast result of the flow process at the future moment.

7. A hydrological probability forecasting system based on full-process time series processing, characterized in that: include: An acquisition module is used to obtain basic data and rainfall forecast data of the target basin; A forecast module is used to establish a hydrological model corresponding to the target basin based on the basic data to obtain measured flow data and simulated flow data in the historical period; input the rainfall forecast data into the hydrological model to obtain forecast flow data of the target basin at a future time; A decomposition module is used to decompose the measured flow data and the simulated flow data of the target basin respectively based on the simulation results of the hydrological model to obtain multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data; wherein obtaining the multi-order smoothed sequences and multi-order residual sequences of the measured flow data and the simulated flow data includes: obtaining a sliding period; based on the sliding period, iteratively decomposing the measured flow data and the simulated flow data respectively until it is determined that the iteration termination condition is met, and obtaining the measured flow data and the simulated flow data. Multi-order smoothing sequence and multi-order residual sequence; wherein the decomposition operation comprises: based on the sliding period, respectively performing sliding average processing on the sequence of the measured flow data and the sequence of the simulated flow data to obtain the first-order smoothing sequence of the measured flow data sequence and the simulated flow data sequence, and based on the first-order smoothing sequence and the measured flow data sequence or the simulated flow data sequence, obtaining the first-order residual sequence of the measured flow data sequence and the simulated flow data sequence; the first-order residual sequence is used as the measured flow data sequence or the simulated flow data sequence in the next iteration process to perform the decomposition operation; An establishment module is used to establish the joint distribution function of each smoothing sequence and residual sequence of the measured flow data and the simulated flow data item by item based on the multi-order smoothing sequence and multi-order residual sequence of the measured flow data and the simulated flow data; A correction module, for performing post-processing correction on the future forecast flow data based on the joint distribution function of each smoothing sequence and residual sequence of the measured flow data and the simulated flow data, to obtain a post-processing sample set of each item of the future forecast flow data in real space; The reconstruction module is used to reconstruct the data of each post-processing sample set of the forecast flow data at the future time to obtain the probability forecast result of the flow process at the future time.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions cause the processor to execute the method according to any one of claims 1 to 6.

Images