A method for analyzing flood evolution into reservoir based on abrupt change test of monthly runoff

Abstract

The application discloses a kind of based on monthly runoff mutation test's flood evolution analysis method, belong to the technical field of flood control.The long-term daily reservoir inflow sequence and monthly reservoir inflow sequence of power station are obtained, and the key time node of runoff mutation is determined, to the internal contrast demarcation point of day scale time sequence demarcation, the long-term daily reservoir inflow sequence is divided into multiple independent hydrological stages;Calculate the reservoir inflow flood discrimination threshold and reservoir inflow flood independent interval of each month;Get the monthly daily reservoir inflow sequence of independent hydrological stage, identify and divide independent reservoir inflow flood event;With independent reservoir inflow flood event as unit, characteristic index is collected to form reservoir inflow flood characteristic data set;Reservoir inflow flood characteristic data set is used for quantitative analysis of the driving mechanism of runoff mutation on reservoir inflow flood spatio-temporal evolution.The application provides data support and technical basis for scheduling scheme optimization, flood forecast precision improvement and hydrological regime evolution research.

Description

Technical Field

[0001] This invention belongs to the technical field of flood control, and in particular relates to a method for analyzing the evolution of inflow floods based on monthly runoff mutation testing. Background Technology

[0002] Most of my country's cascade hydropower stations are located on rivers with complex runoff relationships and variable inflow conditions. Reservoir operation is controlled by the dynamic changes in upstream inflow and floodwaters. In recent years, influenced by the combined disturbances of global climate change, basin development and modification, and the joint regulation and storage of cascade reservoirs, the inflow process no longer maintains a naturally stable state, exhibiting significant stage-specific abrupt changes, particularly pronounced in monthly inflow abrupt changes. These abrupt changes in runoff patterns directly reshape the basin runoff generation and confluence formation mechanism of inflow floods, the evolution and attenuation patterns of reservoir areas, and the annual spatiotemporal distribution patterns. This leads to systematic shifts in the size of inflow flood peaks, flood duration, flood volume distribution, and peak timing, directly impacting flood control scheduling decisions for cascade hydropower stations, accurate inflow flood forecasting, and the safe and stable operation of reservoir projects.

[0003] Current technologies in runoff evolution and flood characteristic research generally neglect the specific analytical needs at the reservoir inflow level, often treating natural river runoff and reservoir inflow as a single, independent analysis. On the one hand, conventional runoff mutation testing only involves monthly-scale, global statistical analysis, which can only roughly determine the timing of runoff variations and cannot accurately match the continuous process of daily-scale inflow floods, making it difficult to meet the refined analytical requirements at the reservoir inflow level. On the other hand, traditional inflow flood event identification and characteristic statistics often adopt fixed thresholds and uniform time-sharing models across the entire region, failing to break down hydrological stages based on key nodes of inflow runoff mutations. This ignores the differentiated evolution of inflow flood generation and confluence conditions and reservoir confluence characteristics before and after mutations, resulting in the inability to accurately identify and quantify the intrinsic evolutionary patterns of inflow floods before and after mutations.

[0004] In addition, the existing analysis system lacks specific criteria for classifying inflow floods, making it difficult to define the classification of inflow flood events that span multiple time periods and have long durations. This can easily lead to confusion in the classification of inflow flood events and inconsistent standards for the collection of characteristic indicators. At the same time, the lack of standardized statistical units centered on inflow flood events makes the comparative analysis of inflow flood characteristics before and after abrupt changes highly subjective and unreliable. It is impossible to quantitatively reveal the intrinsic driving mechanism of runoff abrupt changes on the spatiotemporal evolution of inflow floods, and it is difficult to support the practical engineering application needs of inflow flood forecasting correction for cascade hydropower stations, optimization of flood control scheduling schemes, and long-term hydrological situation assessment of river basin reservoir groups. Summary of the Invention

[0005] To address the technical problems existing in the background art, this invention provides a method for analyzing the evolution of inflow floods based on monthly runoff mutation testing.

[0006] This invention adopts the following technical solution: a method for analyzing the evolution of inflow floods based on monthly-scale runoff mutation testing, comprising the following steps: Based on the basic and operational parameters of cascade hydropower stations, long-term daily and monthly inflow sequences of the power stations were obtained. Abrupt change test is performed on the monthly inflow runoff sequence to identify one or more key time points of runoff abrupt change, wherein the key time points are the months in which the abrupt change occurs at the monthly scale; The key time points are mapped to long-term daily inflow runoff sequences and converted into daily time-series boundary points. ; using the aforementioned daily-scale time series boundary point To establish internal contrast, the long-term daily inflow runoff sequence is divided into multiple independent hydrological stages that are connected end to end and do not overlap. For each independent hydrological stage Break down runoff data by month and calculate the results for each month separately. Inflow flood discrimination threshold Independent intervals between the inflow of floodwater and the reservoir ; Obtain the monthly daily inflow sequence for independent hydrological phases, and determine the threshold based on inflow floods. Independent intervals between the inflow of floodwater and the reservoir Identify and classify independent inflow flood events; Characteristic indicators are collected by using independent inflow flood events as units to form an inflow flood characteristic dataset for the corresponding section; the inflow flood characteristic dataset is used to quantitatively analyze the driving mechanism of runoff mutation on the spatiotemporal evolution of inflow flood.

[0007] In a further embodiment, the basic parameters include: the catchment area of ​​the hydropower station, reservoir capacity curve data, characteristic water levels at various levels, watershed topographic parameters, and watershed river system distribution parameters; The operating parameters include the real-time inflow to the hydropower station, the outflow from the reservoir, the operating water level in the reservoir area, the water storage capacity during the period, and the water flow for power generation by the generating units.

[0008] In a further embodiment, the process for determining the key time nodes is as follows: Define the monthly inflow runoff sequence as Initial screening identified candidate mutation nodes. For each mutation candidate node Perform the following operations: With the mutation candidate nodes Using this as a boundary, the monthly inflow runoff sequence Divided into preorder subsequences and subsequent subsequences ; Based on the preceding subsequence and subsequent subsequences Constructing mutation candidate nodes sliding Test statistic ; Set significance levels and corresponding critical values, based on the sliding... Test statistic For mutation candidate nodes Perform a significance test; if the test passes, the node is identified as a critical time node for mutation; if the test fails, the candidate node is removed.

[0009] In a further embodiment, the conversion process of the diurnal time series boundary point is as follows: The month containing the critical time point of the mutation is defined as the [number]. The month, the The daily time series interval corresponding to each month in the long-term daily inflow runoff sequence ,in, For the first The position of the first day of the month in the daily sequence. This refers to the position of the last day of the month in the daily timeline; The following formula is used to select the daily time series interval. The middle position is used as the daily time series boundary point. : , The floor rule ensures that the delimiter is an integer date.

[0010] In a further embodiment, the process of dividing the independent hydrological stage is as follows: Let the long-term daily inflow runoff sequence be Given a minimum analysis length ; Diurnal time series boundary point Extending forward and backward to both sides from the center Forming the initial stage interval And obtain the runoff section before the mutation. With the runoff section after the change The pre-mutation runoff section With the runoff section after the change All meet the minimum analysis length requirement; If symmetrical extension cannot be achieved due to the constraints of the beginning and end boundaries of the sequence, then asymmetric allocation is performed.

[0011] In a further embodiment, each independent hydrological stage Each month within Inflow flood discrimination threshold Independent intervals between the inflow of floodwater and the reservoir The calculation process is as follows: Let the completed independent hydrological stage be the first stage. s The hydrological stage corresponds to the following time interval: The time series interval is Split by month to get the first s The first hydrological stage Monthly daily inflow runoff sample set ; Based on the daily inflow runoff sample set Statistical mean of daily runoff and mean square deviation Combined with the basin's flood season and wet / dry season correction coefficient and runoff abrupt change disturbance coefficient , construct the first s Phase 1 Threshold for judging inflow of floodwater in one month Calculation formula: ; Get the The average duration of floodwater inflow over a month and flood accumulation frequency coefficient , construct the first s Phase 1 Independent interval for inflow of floodwaters within one month : .

[0012] In a further embodiment, the process for identifying and classifying the independent inflow flood events is as follows: Pre-defined flood duration threshold traversing the daily inflow sequence of the month If there is no time when it is satisfied Then determine the month. No flooding incidents occurred. If there exists a moment when the condition is met, then it will be met for the first time. The time is recorded as the starting point of the flood. , will continuously satisfy The duration is not less than the flood duration threshold. Then the end time of the corresponding continuous segment is recorded as the flood end point. This forms the initial flood process; For adjacent initial flood events, calculate the flood-free interval duration. :like If so, it is determined to be two independent flood events entering the reservoir; if Multiple initial flood events will be merged into independent inflow flood events until the interval between adjacent events is not less than [a certain value]. .

[0013] In a further embodiment, the process of forming the inflow flood characteristic dataset is as follows: Using the aforementioned daily-scale time series boundary point As the dividing line, the starting point of the flood is based on the independent inflow flood event. It was classified as a pre-mutation runoff section. or runoff sections after abrupt changes ; For each independent inflow flood event, the section is divided in turn, and for each independent inflow flood event within the same section, the corresponding flood characteristic indicators are extracted one by one. Finally, all characteristic indicators in the sections before and after the mutation were summarized and integrated to construct a characteristic dataset of floods entering the reservoir in different zones.

[0014] In a further embodiment, the inflow flood characteristic dataset includes: The characteristic indicators for each independent inflow flood event in the runoff section before and after the sudden change include at least: flood start and end time, peak flow, flood duration, flood volume, rise duration, receding duration, peak time and flood process morphology parameters. The characteristic indicators are stored in a structured manner, with each independent flood event entering the reservoir as a unit.

[0015] The beneficial effects of this invention are as follows: This invention transforms key nodes of monthly inflow runoff mutations into daily time series boundary points, using these points to divide inflow runoff comparison sections before and after mutations. This solves the industry pain points of the disconnect between traditional mutation testing and inflow flood process analysis scales, and the mismatch between natural river channel analysis and reservoir inflow scenarios. It provides a unified and clear time benchmark for refined analysis of inflow flood evolution under runoff mutation disturbances.

[0016] This invention uses independent hydrological stages as units and calculates the threshold for identifying floods entering the reservoir and the independent interval of floods entering the reservoir on a monthly basis. It abandons the traditional extensive mode of uniform and fixed standards across the entire area and can fully adapt to the differences in the abundance and scarcity of runoff entering the reservoir in different hydrological stages, different months, and the differences in the confluence characteristics of the reservoir area. It significantly improves the accuracy of flood event identification and the adaptability to actual application scenarios of reservoirs.

[0017] This invention uses the starting time of the inflow flood as the dividing line, uniquely assigning each independent inflow flood event to the inflow runoff section before or after the sudden change, thus completely avoiding the problems of ambiguous attribution and definition of inflow flood event sections. Using independent inflow flood events as the basic statistical unit, various inflow flood characteristic indicators are extracted and collected to construct a standardized regional inflow flood characteristic dataset, ensuring the objectivity and reliability of the comparative analysis of inflow floods before and after the sudden change in terms of statistical units and data scope.

[0018] This invention can systematically reveal the evolutionary impact of sudden changes in inflow runoff on key characteristics such as the start and end times of inflow floods, peak inflow flow, duration of inflow floods, and inflow volume. It is specifically designed to serve the correction of inflow flood forecasts for cascade hydropower stations, the optimization of flood control scheduling schemes, and the assessment of hydrological conditions in reservoir areas, providing accurate data support and reliable technical basis for the safe operation of reservoirs and the regulation of water resources in the basin. Attached Figure Description

[0019] Figure 1 This is a flowchart of a flood characteristic identification method based on runoff monthly runoff abrupt change test, as described in an embodiment. Detailed Implementation

[0020] The present invention will now be further described with reference to the accompanying drawings and embodiments.

[0021] Example 1 like Figure 1 As shown, this embodiment provides a method for analyzing the evolution of inflow floods based on monthly runoff mutation testing, including the following steps: Based on the basic and operational parameters of cascade hydropower stations, long-term daily and monthly inflow sequences of the power stations were obtained. Abrupt change test is performed on the monthly inflow runoff sequence to identify one or more key time points of runoff abrupt change, wherein the key time points are the months in which the abrupt change occurs at the monthly scale; The key time points are mapped to long-term daily inflow runoff sequences and converted into daily time-series boundary points. ; using the aforementioned daily-scale time series boundary point To establish internal contrast, the long-term daily inflow runoff sequence is divided into multiple independent hydrological stages that are connected end to end and do not overlap. For each independent hydrological stage Break down runoff data by month and calculate the results for each month separately. Inflow flood discrimination threshold Independent intervals between the inflow of floodwater and the reservoir ; Obtain the monthly daily inflow sequence for independent hydrological phases, and determine the threshold based on inflow floods. Independent intervals between the inflow of floodwater and the reservoir Identify and classify independent inflow flood events; Characteristic indicators are collected by using independent inflow flood events as units to form an inflow flood characteristic dataset for the corresponding section; the inflow flood characteristic dataset is used to quantitatively analyze the driving mechanism of runoff mutation on the spatiotemporal evolution of inflow flood.

[0022] The basic parameters described in this embodiment can be obtained through the compilation of watershed hydrological data, hydropower station engineering design reports, and interpretation of digital elevation models (DEMs). Specifically, these include: the catchment area of ​​the hydropower station, reservoir capacity curve data, characteristic water levels at various levels, watershed topographic parameters, and watershed river system distribution parameters.

[0023] The operating parameters can be collected in real time through the hydropower station's SCADA system, hydrological automatic monitoring system, gate and unit monitoring system, etc., specifically including the hydropower station's real-time inflow, outflow, reservoir operating water level, water storage during the period, and unit power generation water flow.

[0024] Based on the above basic and operational parameters, a long-term daily inflow runoff sequence is compiled on a daily scale, and a monthly inflow runoff sequence is statistically generated on a monthly scale. After removing missing data and outliers and completing consistency correction, reliable basic data support is provided for subsequent inflow runoff mutation verification and inflow flood characteristic identification.

[0025] Existing methods for detecting sudden changes in runoff mostly employ a single statistical model covering the entire region, failing to consider the unique seasonal confluence, reservoir regulation disturbances, and inter-regional inflow interference characteristics of inflow. Furthermore, they lack specific screening for sudden change nodes tailored to reservoir inflow scenarios. This results in insufficient accuracy in sudden change identification and low computational efficiency. More importantly, the sudden change nodes are severely out of sync with the scale of the inflow flood process, exhibiting poor scenario adaptability and making it difficult to accurately capture the key sudden change points that trigger substantial shifts in the reservoir's hydrological situation. Therefore, the process for determining the key time nodes of inflow runoff sudden changes described in this embodiment is as follows: Define the monthly inflow runoff sequence as , For the first Average monthly inbound volume (unit: ), To monitor the total number of months; Preliminary screening of candidate mutation nodes , The selection method is as follows: based on the long-term evolution of inflow runoff in cascade reservoirs, the annual wet and dry cycle of inflow, and the fluctuation range of inflow runoff anomalies, the superimposed effects of watershed runoff generation and runoff, tributary water supply, and upstream reservoir regulation disturbances on inflow runoff are comprehensively considered. Several time points with abnormal inflow runoff characteristics are initially selected and delineated in the overall inflow runoff sequence as candidate boundary points for abrupt changes to be tested. It is not necessary to perform a traversal and statistical verification of all time points in the inflow runoff sequence. While ensuring the reliability of inflow runoff abrupt change node identification, the workload of model calculation and data processing time are greatly reduced.

[0026] For each mutation candidate node Perform the following operations: using the mutation candidate nodes Using this as a boundary, the monthly inflow runoff sequence Divided into preorder subsequences (Sample size is) ) and subsequent subsequences (Sample size is) ); Synchronous computation of the preceding subsequence and subsequent subsequences The mean and variance of the preceding subsequence; for ease of understanding, the mean of the preceding subsequence described in this embodiment is... and the mean of the preceding subsequence The calculation formulas are as follows: , , , ; variance of preceding subsequence and the mean squared error of the preceding subsequence The calculation formulas are as follows: , ; Based on this, the combined mean square error and moving average are constructed using the following formula. Test statistic: , To construct the combined mean squared error; , Candidate nodes for mutation sliding Test statistic.

[0027] Set significance levels and corresponding critical values ​​for mutation candidate nodes. A significance test is performed; if the test passes, the node is identified as a critical time point for runoff mutation; if the test fails, the candidate node is removed.

[0028] The specific procedure for significance testing is as follows: set the significance level. (Usually taken as 0.05), check The distribution table yields the corresponding degrees of freedom as follows: Two-sided critical value ; The constructed sliding Test statistic With two-sided critical value Comparative analysis: like If the mean of the runoff sequence before and after the node is statistically significant, the hydrological situation has undergone a substantial change, and the node is identified as the critical time node of the abrupt change. like If the node is identified as a short-term runoff fluctuation, it will be excluded.

[0029] All verified key time points of mutations are arranged in chronological order, without overlap, for subsequent hydrological stage division and comparative analysis of flood characteristics before and after the mutations.

[0030] The critical time points for abrupt changes identified above were obtained through abrupt change testing on monthly inflow runoff sequences. Since the time unit is "month," these points can only identify the specific month in which the abrupt change occurred and cannot be directly used for time series division of daily runoff sequences. Therefore, this embodiment provides the conversion process for the corresponding daily time series boundary points: The month containing the critical time point of the mutation is defined as the [number]. The month, the The daily time series interval corresponding to each month in the long-term daily inflow runoff sequence ,in, For the first The position of the first day of the month in the daily sequence. This refers to the position of the last day of the month in the daily timeline; The following formula is used to select the daily time series interval. The middle position is used as the daily time series boundary point. : , The floor rule ensures that the delimiter is an integer date.

[0031] Based on the assumption of uniformity of intra-month runoff, the impact of the month of abrupt change is symmetrically distributed to the preceding and following time periods, ensuring the symmetry and comparability of samples before and after the abrupt change. This completes the one-to-one correspondence conversion from monthly key nodes to daily boundary points, providing a clear and unified daily-scale benchmark for the subsequent division of independent hydrological stages.

[0032] Traditional methods, which symmetrically extract fixed-time intervals centered on abrupt change nodes, fail to consider the unique characteristics of inflow runoff, such as reservoir regulation and uneven annual distribution of runoff within intervals. When abrupt change nodes are near the beginning or end of the inflow runoff sequence, insufficient effective sample length on one side and a small number of statistical samples of inflow floods before and after the abrupt change are easily encountered, failing to meet the minimum sample size requirements for comparative analysis of inflow flood characteristics, thus distorting the results of inflow flood evolution pattern analysis. Furthermore, abrupt change nodes at the beginning and end of the runoff sequence are difficult to analyze symmetrically. Existing conventional methods often directly discard boundary abrupt change nodes or blindly reduce the analysis range, wasting long-term inflow runoff measurement data resources and failing to achieve comprehensive analysis of inflow runoff abrupt change characteristics and inflow flood evolution patterns across the entire time series.

[0033] Therefore, this embodiment provides the following independent hydrological stage division process: Let the long-term daily inflow runoff sequence be ,in, For the first Daily inflow runoff (unit: ), This is the total length of the daily runoff sequence (i.e., the total number of monitoring days). Given minimum analysis length (Unit: days), the minimum analysis length described in this embodiment The following two water conservancy statistics principles must be met simultaneously: Representativeness principle of flood samples: It should be long enough to cover a sufficient number of flood events to avoid statistical distortion due to insufficient sample size; The principle of timeliness affected by mutations: It should not be too long, otherwise it will introduce other interfering factors before and after the mutation, affecting the accuracy of mutation effect analysis.

[0034] Based on engineering experience in analyzing inflow floods, Typically, the inflow length of the day corresponding to the entry into the reservoir is taken as 5 to 10 years (approximately 1825 to 3650 days), which can both ensure the sample size and focus on the impact of the mutation itself.

[0035] The diurnal time series boundary points determined above Divide the project into phases based on the baseline: Diurnal time series boundary point Extending forward and backward to both sides from the center Forming the initial stage interval And obtain the runoff section before the mutation. With the runoff section after the change The pre-mutation runoff section With the runoff section after the change All meet the minimum analysis length requirement.

[0036] If symmetrical extension cannot be achieved due to the constraints of the beginning and end boundaries of the sequence, then asymmetric allocation is performed. The specific allocation process is as follows: when When that happens, the stage interval will be adjusted to Compress the front-end extension length to ensure that the length of the runoff section after the abrupt change is not less than Generally, inflow runoff sequences often exhibit abrupt inflection points in the first half of the time series. These are often caused by early water conservancy projects within the basin, the construction and operation of upstream reservoirs, and other human-induced control projects, which continuously disturb the natural runoff and inflow replenishment patterns from the initial monitoring stage. At this point, the effective observation duration at the beginning of the sequence is insufficient, failing to meet the requirements for symmetrical interval layout. This method effectively avoids the problem of distorted statistical comparison of inflow flood characteristics caused by the scarcity of unilateral samples by compressing the analysis span at the beginning and prioritizing the integrity of inflow runoff and inflow flood samples after the abrupt change.

[0037] when When that happens, the stage interval will be adjusted to Compress the length of the rear extension to ensure that the length of the runoff section before the sudden change is not less than If abrupt changes occur in the inflow runoff monitoring sequence at the end of the time series, it is often due to factors such as the frequent occurrence of extreme weather events in the basin in recent years, changes in land use and underlying surface patterns, which cause a systematic shift in the natural runoff generation and concentration mechanisms and the intra-annual distribution pattern of inflow runoff. At this time, the effective observation data duration at the tail end of the sequence is insufficient, making it difficult to establish symmetrical analysis intervals. This method ensures that standardized and effective comparative analysis of inflow flood characteristics can still be carried out before and after the abrupt change by appropriately compressing the analysis span at the end of the time series and prioritizing the integrity of inflow runoff and inflow flood samples before the abrupt change.

[0038] Existing technologies, when determining the threshold for inflow flood identification and the independent interval of inflow floods, generally adopt a fixed single value standard across the entire area. This fails to divide hydrological stages based on the abrupt changes in inflow runoff patterns, nor does it consider the differences in the seasonal rhythms of inflow and the confluence characteristics of the reservoir area across different months. This makes it difficult to adapt to the significant variations in hydrological conditions before and after abrupt changes in inflow runoff, and the combined effects of reservoir regulation. This easily leads to missed or misjudged inflow flood events, and results in significant deviations in the extraction of flood characteristic indicators, making it difficult to accurately depict the actual frequency and evolution patterns of inflow floods in different hydrological stages and months before and after abrupt changes. Therefore, each independent hydrological stage... Each month within Inflow flood discrimination threshold Independent intervals between the inflow of floodwater and the reservoir The calculation process is as follows: Let the completed independent hydrological stage be the first stage. s The hydrological stage corresponds to the following time interval: The time series data for this stage is split into months to obtain the first... s The first hydrological stage Monthly daily inflow runoff sample set ;in, For the first s The start time of the hydrological phase For the first s The end time of the hydrological phase.

[0039] Unlike existing technologies that use a uniform threshold for identifying floods entering the reservoir and a fixed flood interval across the entire region, this method uses the differences in hydrological patterns caused by abrupt changes in runoff as the basis for division and solves differentiated parameters using independent hydrological stages as the basic unit.

[0040] Based on the daily inflow runoff sample set Statistical mean of daily runoff and mean square deviation Combined with the basin's flood season and wet / dry season correction coefficient and runoff abrupt change disturbance coefficient , construct the first s Phase 1 Threshold for judging inflow of floodwater in one month Calculation formula: Generally, the watershed flood season adjustment coefficient... The coefficient for runoff abrupt change disturbance is 1.5–2.0 during the flood season and 1.0–1.2 during the non-flood season. During the high-water season, the value is 1.1 to 1.3, and during the low-water season, it is 0.7 to 0.9.

[0041] Get the The average duration of floodwater inflow over a month (Unit: days) and flood accumulation frequency coefficient , construct the first s Phase 1 Independent interval for inflow of floodwaters within one month : Generally, the frequency coefficient of flood accumulation The values ​​are: 0.8~1.0 during periods of high flood incidence and 1.2~1.5 during periods of low flood incidence.

[0042] Therefore, the process for identifying and classifying independent inflow flood events is as follows: Pre-defined flood duration threshold traversing the daily inflow sequence of the month If there is no time when it is satisfied Then determine the month. No flooding incidents occurred. If there exists a moment when the condition is met, then it will be met for the first time. The time is recorded as the starting point of the flood. , will continuously satisfy The duration is not less than the flood duration threshold. Then the end time of the corresponding continuous segment is recorded as the flood end point. This forms the initial flood process; For adjacent initial flood events, calculate the flood-free interval duration. :like If so, it is determined to be two independent flood events entering the reservoir; if Multiple initial flood events will be merged into independent inflow flood events until the interval between adjacent events is not less than [a certain value]. .

[0043] In this embodiment, the threshold for the duration of inflow floodwaters is... The value is expressed in days and can be a positive integer or a fixed decimal, typically ranging from 3 to 7 days. Commonly used values ​​are 3, 5, and 7 days, used to match the average duration of inflow processes in different watersheds. This parameter does not use a fixed value across the entire watershed but is selected based on the underlying surface conditions of the watershed, the hydrological stage of the inflow runoff, and the differences in monthly inflow characteristics. For basins with rapid confluence and frequent short-duration inflow floods, a smaller threshold (e.g., 3 days) is selected to prevent short-duration inflow floods from being missed. For basins with slow confluence and long evolution cycles of large-scale inflow floods, a larger threshold (e.g., 5-7 days) is selected to eliminate invalid processes caused by short-term flow fluctuations and ensure that the identified inflow flood events are typical.

[0044] This method combines the rule of determining the interval of floods entering the reservoir without the flood inflow period to independently divide or merge adjacent initial flood inflow processes. This solves the problem that traditional methods easily split multi-peak floods and continuous floods entering the reservoir into multiple independent events, ensuring the independence and process integrity of flood inflow events. It also lays a solid foundation of accurate statistical units for the standardized collection of flood inflow characteristic indicators.

[0045] Based on the confirmation of independent inflow flood events, the process of forming the inflow flood feature dataset described in this embodiment is as follows: Using the aforementioned daily-scale time series boundary point As the dividing line, the starting point of the flood is based on the independent inflow flood event. It was classified as a pre-mutation runoff section. or runoff sections after abrupt changes ; For each independent inflow flood event, the section is divided in turn, and for each independent inflow flood event within the same section, the corresponding flood characteristic indicators are extracted one by one. Finally, all characteristic indicators in the sections before and after the mutation were summarized and integrated to construct a characteristic dataset of floods entering the reservoir in different zones.

[0046] Taking the inflow sequence of a hydropower station as an example, let the daily time series boundary point be set. Minimum analysis length as of July 15, 2010 Then: the runoff section before the mutation was: The total duration is 150 days, and the runoff section after the mutation is as follows: A total of 150 days. For example, in independent flood event A: the flood starting point is June 10, 2010, located in... Within this range, it was therefore classified as the pre-mutation runoff zone; For example, in independent flood event B: the flood initiation date is August 20, 2010, located in... Within this range, it was therefore classified as the post-abrupt runoff zone; Independent flood event C: The starting point of the flood was July 15, 2010 (i.e. the dividing point day). It does not belong to any section and is not included in this comparative analysis.

[0047] The final constructed regional flood inflow feature dataset is classified into pre- / post-mutation categories, with each independent flood inflow event as a unit. All feature indicators are stored in a structured manner, providing standardized data for subsequent comparative analysis.

[0048] The inflow flood feature data set includes: feature indicators of each independent inflow flood event in the runoff section before and after the sudden change. The feature indicators include at least: flood start and end time, peak flow, flood duration, flood volume, rise duration, receding duration, peak time, and flood process morphology parameters. The characteristic indicators are stored in a structured manner, with each independent flood event entering the reservoir as a unit.

[0049] For independent inflow flood event A in the section before the mutation, extract its peak flow (e.g., 1200). ), duration of the flood (e.g., 5 days), flood volume (e.g., 35 million cubic meters) Indicators such as ) For independent inflow flood event B in the section following the mutation, extract its peak flow (e.g., 900). ), duration of the flood (e.g., 3 days), flood volume (e.g., 18 million cubic meters) Indicators such as ) Finally, the inflow flood characteristic dataset was used to quantitatively analyze the driving mechanism of runoff mutation on the spatiotemporal evolution of inflow floods. By comparing the statistical distribution differences of key indicators such as peak flow, flood duration, flood volume, peak occurrence time, and rise and fall process morphology of each independent inflow flood event before and after the mutation, the impact of runoff mutation on flood magnitude, frequency, process morphology, and spatiotemporal distribution patterns was quantified. This provides data support and decision-making basis for optimizing flood control scheduling schemes for cascade hydropower stations, calibrating flood forecasting models, and assessing the evolution of watershed hydrological conditions.

Claims

1. A method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change tests, characterized in that, Includes the following steps: Based on the basic and operational parameters of the cascade hydropower stations, the long-term daily and monthly inflow sequences of the power stations were obtained. Abrupt change test is performed on the monthly inflow runoff sequence to identify one or more key time points of runoff abrupt change, wherein the key time points are the months in which the abrupt change occurs at the monthly scale; The key time points are mapped to long-term daily inflow runoff sequences and converted into daily time-series boundary points. ; Using the aforementioned daily-scale time series boundary point To establish internal contrast, the long-term daily inflow runoff sequence is divided into multiple independent hydrological stages that are connected end to end and do not overlap. For each independent hydrological stage Break down runoff data by month and calculate the results for each month separately. Inflow flood discrimination threshold Independent intervals between the inflow of floodwater and the reservoir ; Obtain the monthly daily inflow sequence for independent hydrological phases, and determine the threshold based on inflow floods. Independent intervals between the inflow of floodwater and the reservoir Identify and classify independent inflow flood events; Characteristic indicators are collected by using independent inflow flood events as units to form an inflow flood characteristic dataset for the corresponding section; the inflow flood characteristic dataset is used to quantitatively analyze the driving mechanism of runoff mutation on the spatiotemporal evolution of inflow flood.

2. The method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change detection as described in claim 1, characterized in that, The basic parameters include: the catchment area of ​​the hydropower station, reservoir capacity curve data, characteristic water levels at all levels, topographic and geomorphological parameters of the basin, and water system distribution parameters of the basin. The operating parameters include the real-time inflow to the hydropower station, the outflow from the reservoir, the operating water level in the reservoir area, the water storage capacity during the period, and the water flow for power generation by the generating units.

3. The method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change detection as described in claim 1, characterized in that, The process for determining the key time nodes is as follows: Define the monthly inflow runoff sequence as Initial screening identified candidate mutation nodes. For each mutation candidate node Perform the following operations: With the mutation candidate nodes Using this as a boundary, the monthly inflow runoff sequence Divided into preorder subsequences and subsequent subsequences ; Based on the preceding subsequence and subsequent subsequences Constructing mutation candidate nodes sliding Test statistic ; Set significance levels and corresponding critical values, based on the sliding... Test statistic For mutation candidate nodes Perform a significance test; If the test passes, it is identified as a critical time point for mutation. If the test fails, the candidate node is removed.

4. The method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change detection as described in claim 1, characterized in that, The conversion process of the diurnal time series boundary point is as follows: The month containing the critical time point of the mutation is defined as the [number]. The month, the The daily time series interval corresponding to each month in the long-term daily inflow runoff sequence ,in, For the first The position of the first day of the month in the daily sequence. This refers to the position of the last day of the month in the daily timeline; The following formula is used to select the daily time series interval. The middle position is used as the daily time series boundary point. : , The floor rule ensures that the delimiter is an integer date.

5. The method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change detection according to claim 1, characterized in that, The process of dividing the independent hydrological stage is as follows: Let the long-term daily inflow runoff sequence be Given a minimum analysis length ; Diurnal time series boundary point Extending forward and backward to both sides from the center. Forming the initial stage interval And obtain the runoff section before the mutation. With the runoff section after the change The pre-mutation runoff section With the runoff section after the change All meet the minimum analysis length requirement; If symmetrical extension cannot be achieved due to the constraints of the beginning and end boundaries of the sequence, then asymmetric allocation is performed.

6. The method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change detection as described in claim 1, characterized in that, Each independent hydrological stage Each month within Inflow flood discrimination threshold Independent intervals between the inflow of floodwater and the reservoir The calculation process is as follows: Let the completed independent hydrological stage be the first... s The hydrological stage corresponds to the following time interval: The time series interval is Split by month to get the first s The first hydrological stage Monthly daily inflow runoff sample set ; Based on the daily inflow runoff sample set Statistical mean of daily runoff and mean square deviation Combined with the basin's flood season and wet / dry season correction coefficient and runoff abrupt change disturbance coefficient , construct the first s Phase 1 Threshold for reclamation of floodwater inflow over a month Calculation formula: ； Get the The average duration of floodwater inflow over a month and flood accumulation frequency coefficient , construct the first s Phase 1 Independent interval for inflow of floodwaters within one month : 。 7. The method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change detection according to claim 1, characterized in that, The process for identifying and classifying independent inflow flood events is as follows: Pre-defined flood duration threshold traversing the daily inflow sequence of the month If there is no time when it is satisfied Then determine the month. No flooding incidents occurred. If there exists a moment when the condition is met, then it will be met for the first time. The time is recorded as the starting point of the flood. , will continuously satisfy The duration is not less than the flood duration threshold. Then the end time of the corresponding continuous segment is recorded as the flood end point. This forms the initial flood process; For adjacent initial flood events, calculate the flood-free interval duration. :like If so, it is determined to be two independent flood events entering the reservoir; if Multiple initial flood events will be merged into independent inflow flood events until the interval between adjacent events is not less than [a certain value]. .

8. The method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change detection as described in claim 1, characterized in that, The process of forming the inflow flood characteristic dataset is as follows: Using the aforementioned daily-scale time series boundary point As the dividing line, the starting point of the flood is based on the independent inflow flood event. It was classified as a pre-mutation runoff section. or runoff sections after abrupt changes ; For each independent inflow flood event, the section is divided in turn, and for each independent inflow flood event within the same section, the corresponding flood characteristic indicators are extracted one by one. Finally, all characteristic indicators in the pre- and post-mutation sections are summarized and integrated to construct a regional flood inflow characteristic dataset.

9. The method for analyzing the evolution of inflow floods based on monthly-scale runoff abrupt change detection as described in claim 1, characterized in that, The inflow flood characteristic dataset includes: The characteristic indicators for each independent inflow flood event in the runoff section before and after the sudden change are as follows: the start and end times of the flood, peak flow, flood duration, flood volume, rise duration, receding duration, peak time and flood process morphology parameters related to the inflow. The characteristic indicators are stored in a structured manner, with each independent flood event entering the reservoir as a unit.

Images