Flood forecasting and dispatching implementation system based on rain measuring radar and implementation method of flood forecasting and dispatching implementation system

Abstract

The invention relates to the technical field of flood forecasting and dispatching, in particular to a flood forecasting and dispatching system based on a rain-measuring radar and an implementation method of the flood forecasting and dispatching system. The method comprises the following steps: recording historical records and rainfall measurement radar observation data of each hydrological station, constructing a distributed runoff production and convergence model by identifying and removing rainfall echo interference data and combining the hydrological historical records and anti-interference observation data so as to calculate a runoff production sequence of each sub-basin in the future, and setting the runoff production sequence to predict a flood disaster sequence. And according to the flood disaster sequence, identifying a flood risk high-incidence area, and in combination with preset reservoir water regimen data, making a corresponding flood discharge scheduling scheme, and according to the flood discharge scheduling scheme, carrying out reservoir resource scheduling in real time. The flood forecasting and dispatching implementation method based on the rain measuring radar is more accurate, has higher timeliness and is more accurate in early warning.

Description

Technical Field

[0001] This invention relates to the field of flood forecasting and scheduling technology, and in particular to a flood forecasting and scheduling implementation system and method based on rainfall radar. Background Technology

[0002] Traditional flood forecasting and scheduling methods often rely on historical hydrological data analysis, lacking effective integration with real-time meteorological data. This limits the timeliness and accuracy of forecasts, especially when facing complex extreme precipitation events. Existing technologies cannot promptly identify and clarify early warning data from rainfall radar, thus affecting the quality of flood control data and resulting in low flood forecast accuracy. In practical applications, especially in high-risk flood areas, this can lead to errors in flood control scheduling and resource allocation, increasing flood risk. Existing runoff generation and confluence models rely heavily on static historical data, failing to fully consider the uncertainties of climate change and local precipitation. They lack the ability to dynamically calculate accurate future runoff sequences and often cannot effectively integrate the latest meteorological observation data when forecasting floods, resulting in inaccurate runoff sequence calculations. This fails to provide timely and effective decision support for reservoir scheduling, especially in the face of sudden and complex events. Existing models are slow to respond to emergencies and cannot achieve real-time response. Summary of the Invention

[0003] Therefore, it is necessary to provide a flood forecasting and scheduling system and its implementation method based on rainfall radar to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a flood forecasting and scheduling method based on rainfall radar includes the following steps: Step S1: Record the historical records of each hydrological station and the observation data of the rainfall radar; identify precipitation echo interference data in the rainfall radar observation data, clear the precipitation echo interference data, and confirm the anti-interference observation data; Step S2: Construct a distributed runoff generation and confluence model by combining historical hydrological records and disturbance prevention observation data; Step S3: Simulate and calculate the future runoff sequence of each sub-basin using a distributed runoff generation and confluence model; sequentially merge the future runoff sequences of each sub-basin into peak flows to predict flood disaster sequences; Step S4: Identify high-risk flood areas based on flood disaster sequences; for high-risk flood areas, determine pre-discharge scheduling plans by combining flood sequences with pre-set reservoir hydrological data; Step S5: Real-time scheduling of reservoir resources according to the pre-planned flood discharge schedule to prevent flood risks.

[0005] This invention also provides a flood forecasting and scheduling implementation system based on rainfall radar, used to execute the flood forecasting and scheduling implementation method based on rainfall radar as described above. The flood forecasting and scheduling implementation system based on rainfall radar includes: The data preprocessing module is used to record historical data of each hydrological station and rainfall radar observation data; identify precipitation echo interference data in the rainfall radar observation data, remove precipitation echo interference data, and confirm the anti-interference observation data. The model building module is used to build a runoff generation and confluence model by combining historical data and disturbance prevention observation data; The sequence prediction module is used to simulate and calculate the future runoff sequences of each sub-basin through rainfall model; the future runoff sequences of each sub-basin are sequentially merged into peak flow to predict flood disaster sequences; The scheduling design module is used to identify high-risk flood areas based on flood disaster sequences; for high-risk flood areas, it determines a pre-discharge scheduling plan by combining the flood sequence with preset reservoir hydrological data. The scheduling and prevention module is used to schedule reservoir resources in real time according to the pre-planned flood discharge schedule in order to prevent flood risks.

[0006] This invention, through the effective operation of the data model and the accurate recording of historical records and rainfall radar observation data, ensures the elimination of data deviations and interference. The removal of interfering data improves the accuracy of the anti-interference observation data, laying the foundation for subsequent analysis. The model module, combining historical records and anti-interference observation data, effectively constructs a runoff generation and confluence model, enhancing the simulation capability of the impact on precipitation processes and improving the accuracy and timeliness of predictions. The sequence prediction module, through simulation calculations of runoff generation sequences in various watersheds, can generate future flood disaster sequences in real time, providing a scientific basis for flood risk assessment. The prediction of runoff peaks makes flood event warnings more accurate. The scheduling design module, based on the flood disaster sequence to identify high-risk areas and combined with reservoir hydrological data, can formulate reliable flood discharge and dispatch plans. The proposed solution enhances the scientific nature of reservoir management. During high-risk floods, the real-time reservoir resource scheduling capability of the scheduling and prevention module ensures rapid response to sudden floods and reduces the risk of overflow. The overall system design, through its upgraded architecture, improves the efficiency between various functional modules, facilitating subsequent maintenance and upgrades. The flood scheduling implementation method based on rainfall radar, through the comprehensive application of multi-source data, ensures information flow, promotes data sharing and collaborative decision-making. The system has strong interoperability and can dynamically adjust according to different meteorological conditions and geographical locations, enhancing the flexibility and reliability of flood environments. Overall, the flood disaster scheduling implementation method and system based on rainfall radar provide an innovative solution for flood management and promote the construction of standardized water conservancy systems. Attached Figure Description

[0007] Figure 1 This is a flowchart illustrating the steps of a flood forecasting and scheduling method based on rainfall radar. Figure 2 for Figure 1 A flowchart illustrating step S2; Figure 3 This is a schematic diagram of a distributed generation-convergence model; Figure 4 Water level-discharge relationship curve.

[0008] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0009] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0010] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0011] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0012] To achieve the above objectives, please refer to Figures 1 to 4 A flood forecasting and scheduling method based on rainfall radar includes the following steps: Step S1: Record the historical records of each hydrological station and the observation data of the rainfall radar; identify precipitation echo interference data in the rainfall radar observation data and remove the precipitation echo interference data to obtain interference-free observation data; Step S2: Combine historical records and anti-interference observation data to construct a distributed generation and merging model; Step S3: Calculate the runoff sequence for each sub-basin in the future using a distributed runoff generation and confluence model; then, sequentially merge and superimpose the runoff sequences for each sub-basin in the future to predict the flood forecast sequence. Step S4: Identify high-risk flood areas based on flood forecast sequences; for high-risk flood areas, determine pre-discharge scheduling plans by combining flood forecast sequences with preset reservoir hydrological data; Step S5: Conduct real-time flood control scheduling of the reservoir according to the pre-planned flood discharge schedule to prevent the risk of overflow.

[0013] In this embodiment, in some embodiments, historical data from various hydrological stations and real-time observation data from rainfall radar are recorded to form a basic hydrological dataset containing multi-source observation elements. The historical data includes long-term observation indicators such as hourly rainfall, runoff, water level change sequences, and evaporation, while the rainfall radar observation data includes volume scan parameters such as reflectivity factor (Z), radial velocity (V), and spectral width (W). These two types of data differ significantly in temporal resolution and spatial distribution density; therefore, they need to be uniformly formatted and subjected to interference identification processing before subsequent analysis. Specifically, the rainfall radar volume scan data is first parsed. Based on the elevation angle sequence and scanning period information, the observed precipitation echo interference data is identified according to the scanning system. False echo signals are eliminated by comparing the echo similarity characteristics at different elevation angle levels. Abnormal echo segments are screened by combining the correlation threshold between meteorological radar echo intensity and ground-measured precipitation data. The residual artifact area is corrected by the spatial consistency constraint algorithm, thus obtaining the anti-interference observation data after precipitation echo interference elimination. This data has spatial consistency and temporal continuity and can serve as a reliable basis for rainfall input. Combined with standardized hydrological station historical data, the input set for subsequent distributed runoff generation and confluence models is constructed.

[0014] like Figure 3 As shown, based on the historical records and disturbance prevention observation data, a distributed runoff generation and confluence model is constructed to reflect the rainfall-runoff response characteristics within different sub-basins. The model is divided into grids according to topography, land use type, soil permeability coefficient, and initial water content parameters. Each grid cell is represented in the form of hydrological element parameters. By establishing a rainfall-runoff-confluence chain calculation relationship, the rainfall input is transformed into a grid-level runoff sequence. During the model training phase, historical rainfall and measured runoff are used for inversion calibration to determine the permeability parameters, lag coefficient, and confluence coefficient. A spatiotemporal weighted average algorithm is used to spatially integrate the grid runoff results to generate the future time period runoff sequence for each sub-basin.

[0015] Using the future runoff sequences of each sub-basin calculated by the distributed runoff generation and confluence model as input data, runoff is superimposed based on the hydrological connectivity between the sub-basins. First, the direction of the runoff path and the location of the basin outlet are calculated based on the topographic elevation model (DEM). Then, a propagation delay function is introduced in the time dimension to compensate for the time transmission of flow in adjacent sub-basins, and a flow time-series alignment matrix is ​​established. The overall flood forecast sequence is formed by a step-by-step superposition method. During the superposition process, the sub-basin area weighting and runoff contribution rate are used for weighted fusion to ensure the rationality of spatial distribution and the conservation of total flow, and finally, a time-continuous flood forecast sequence is obtained.

[0016] Based on the flood forecast sequence, high-risk flood areas are identified. Specifically, the flood forecast sequence is compared with historical flood level thresholds and warning level data to calculate the probability distribution of water level exceeding limits in each area during the future period. High, medium, and low-risk areas are divided using a risk classification model. For high-risk areas, corresponding reservoir water information data is further extracted, including real-time reservoir capacity, inflow, outflow, and early warning scheduling curve parameters. Combined with the peak arrival time of the flood in the corresponding time period in the flood forecast sequence, a pre-discharge scheduling plan is generated using a multi-objective optimization algorithm.

[0017] Reservoir resources are allocated in real time according to the pre-planned flood discharge schedule to prevent the risk of overflow. During the scheduling process, control strategies are implemented in time segments. The control values ​​of the flood discharge flow for each time period are determined by the water level scheduling curve. The changes in the inflow flow are monitored in real time, and the scheduling parameters are dynamically adjusted according to the flood forecast sequence to ensure that the outflow flow matches the downstream carrying capacity. The scheduling execution results are fed back to the scheduling center through the telemetry system to form scheduling status record data, which is compared with the flood forecast sequence in a closed loop to evaluate the scheduling effect and the accuracy of the model forecast. Thus, a complete closed-loop flood forecasting and scheduling process from radar observation and model prediction to real-time scheduling is realized.

[0018] Of particular importance, step S2 includes: Step S21: Establish a watershed spatial grid using historical rainfall and flow records, dividing the watershed into grid units with hydrological response characteristics. Each grid unit records rainfall, topographic slope, and confluence direction, forming a watershed spatial grid. Step S22: Project the disturbance prevention observation data into the watershed spatial grid, compare the historical rainfall records with the disturbance prevention observation data, and generate comparison data; Step S23: Correct the rainfall intensity and temporal distribution within each grid cell by comparing the data to obtain rainfall correction data; Step S24: Construct a distributed runoff generation and confluence structure comprising three parts: cloud rain, merged rainfall, and runoff generation and confluence, based on rainfall correction data; Step S25: Iteratively couple the distributed generation and merging structures to construct a distributed generation and merging model.

[0019] In some embodiments, a watershed spatial grid is established using historical rainfall and flow records. By spatializing historical hydrological monitoring data, the entire watershed is divided into several grid cells with independent hydrological response characteristics based on a topographic elevation model (DEM). Each grid cell contains attribute information such as rainfall, topographic slope, runoff direction, soil type, and land use characteristics. This grid structure uses latitude and longitude as the spatial reference and time series as the attribute index, comprehensively characterizing the spatial distribution of rainfall and surface response characteristics within the watershed. Based on this, interference-free rainfall radar anti-interference observation data is projected onto the watershed spatial grid. According to the elevation angle, azimuth angle, and resolution parameters of the radar scan data, a geographic resampling algorithm is used to convert the radar volume scan spatial data into a projected dataset matching the grid cells.

[0020] Specifically, using the geographic coordinates of the center point of each grid cell in the watershed DEM as a reference, the corresponding spatial pixels in the radar data are extracted, and their reflectivity factor (Z) and estimated rainfall values ​​are calculated. These are then combined with measured rainfall and flow information from historical rainfall records for temporal matching and difference analysis to construct rainfall comparison data. This rainfall comparison data is used to characterize the deviation features between the radar-retrieved rainfall field and the ground observation field. The difference distribution includes two parts: a rainfall intensity deviation field and a temporal deviation field.

[0021] Furthermore, the rainfall intensity and temporal distribution within each grid cell of the watershed are corrected using rainfall comparison data. Spatiotemporal interpolation methods are employed to align the timing of rainfall peak advances and lags, and a spatially weighted smoothing algorithm is used to correct the rainfall gradient change rate, thereby generating rainfall correction data with continuity and spatial consistency. Based on this, a distributed runoff generation structure is constructed, comprising three core modules: cloud-in-rain, fused rainfall, and runoff generation. The cloud-in-rain module quantifies the cloud precipitation formation process based on the relationship between radar reflectivity and rainfall rate. The fused rainfall module fuses radar rainfall and ground observation data using a weighted minimum deviation criterion. The runoff generation module calculates surface runoff and groundwater infiltration based on topographic slope, surface runoff coefficient, and underlying surface characteristics. Through parameter linkage optimization and temporal feedback updates of these three modules, a complete distributed runoff generation structure is established.

[0022] Furthermore, the simulation structure is subjected to multiple rounds of iterative coupled computation. In each round of iteration, historical measured flow data is introduced to perform error inversion and parameter correction on the model output results until the mean square error between the flow sequence output by the model and the measured value meets the convergence threshold condition, thereby obtaining a stable and reliable distributed runoff generation and confluence model, realizing high-precision simulation and prediction of the watershed hydrological response process under different rainfall scenarios.

[0023] Of particular importance, step S4 includes: Peak flow and cumulative water level of each sub-basin in future time periods are extracted from the flood forecast sequence to construct a flood evolution distribution map; Identify the spatiotemporal distribution of floods using flood evolution distribution maps; Calculate the direction and velocity of flood peak propagation based on the spatiotemporal distribution of floods, and identify areas of rapid flow convergence; Based on the pre-defined topography and river network structure, identify high-risk flood areas within the rapid flow convergence zone; Spatial matching is performed between high-risk flood areas and preset reservoir hydrological data to extract the reservoir capacity, water level and discharge capacity of the corresponding areas, and reservoir response data is generated. The time window for early flood discharge is calculated based on the peak time difference between reservoir response data and flood forecast sequence, and the corresponding flood discharge flow and scheduling sequence are planned to generate a pre-discharge scheduling plan.

[0024] In some embodiments, by performing hourly analysis on the flood forecast sequence to extract peak flow and cumulative water level data for each sub-basin in the future period, and spatially mapping the peak parameters based on the geographic grid coordinate system, a flood evolution distribution map with spatiotemporal attributes is constructed. The flood evolution distribution map describes the flow change process in the time dimension and the water level accumulation area in the spatial dimension. It expresses the evolution law of floods of different intensities in time and space through color gradients and contour lines, thereby realizing a visual characterization of the propagation trend of flood peak within the basin.

[0025] Specifically, during the construction process, peak flow curves for future periods in each sub-basin of the flood forecast sequence are detected, the time and location parameters corresponding to the peak flow are recorded, and the cumulative water level change is calculated. A time-series water level change matrix is ​​formed by overlaying multiple time periods. Further, based on the peak migration trajectories at different time layers in the flood evolution distribution map, the direction and velocity of flood peak propagation are calculated. A peak tracking algorithm based on spatial gradients is used to extract the change vector of the flood peak center coordinates at adjacent times, and the propagation velocity field is obtained through time interval division to identify areas of rapid flow convergence where the rate of flow change increases significantly along the propagation path. Spatial clustering and gradient threshold segmentation of the propagation velocity field are performed to distinguish the differences in flood concentration intensity in different regions, thereby obtaining the concentrated development areas of the flood in both time and space dimensions.

[0026] Furthermore, by combining pre-set watershed topographic data and river network structure data, and using river slope, tributary confluence density, and confluence path as constraints, spatial constraints are applied to areas of rapid flow convergence. Finally, high-risk flood areas are determined by judging hydrodynamic characteristic thresholds.

[0027] Specifically, using slope distribution and river connectivity in the digital elevation model (DEM) as core parameters, the hydraulic gradient and cumulative runoff indices in areas with high flood risk are calculated to identify high-risk sections that may experience localized overflows under extreme rainfall conditions.

[0028] Furthermore, high-risk flood areas are spatially matched with pre-defined reservoir hydrological data. By comparing the locations of risk areas and reservoirs using geographic coordinate indexing, reservoir attribute information for the corresponding areas is extracted, including current reservoir capacity, real-time water level, maximum scheduled water level, and discharge capacity parameters, forming a reservoir response dataset. By analyzing the peak time relationship between reservoir response data and flood forecast sequences, the time window for advance flood discharge is calculated based on the peak arrival time difference. Then, based on the discharge capacity curves and scheduling constraints of each reservoir, a multi-objective optimization algorithm is used to plan the flood discharge flow allocation and execution sequence, generating a pre-discharge scheduling scheme that conforms to the flood peak transmission pattern.

[0029] Preferably, step S3, which calculates the runoff sequence for future periods in each sub-basin using a distributed runoff generation and confluence model, includes: A water level-discharge relationship curve was plotted based on historical rainfall data and historical water level data. Dynamic response data is generated by extracting the time difference characteristics of water level changes relative to flow rate changes from the water level-flow relationship curve. By mapping dynamic response data to a distributed runoff generation and confluence model and comparing the rainfall input time and water level change time of each sub-basin in the model, the time interval between rainfall occurrence and water level response is calculated to obtain the rainfall response delay. The delay measurement data of rainfall measurement radar observation data for each time period is calculated based on the rainfall response delay. Extract the delay time unit from the rainfall response delay; The actual rainfall data observed by the rain-measuring radar after the delay time unit is calculated by using delayed measurement data. By mapping actual rainfall data to a distributed runoff generation and confluence model, the runoff generation sequence of each sub-basin in the time delay unit is obtained. By changing the time delay unit, the delayed measurement data of rainfall data in each subsequent future time period is obtained, and the runoff generation sequence of each sub-basin in the future time period is generated.

[0030] In some embodiments, such as Figure 4 As shown, a distributed runoff generation and confluence model is used to calculate the runoff generation sequence for future periods in each sub-basin, so as to achieve dynamic coupling prediction of rainfall input and basin response process. The process establishes a water level-discharge relationship curve based on historical rainfall and water level monitoring data. By comparing the rainfall and corresponding water level change sequences in different periods, the lag characteristics of water level change relative to flow change are extracted, thereby generating dynamic response data describing the hydrological response speed of the basin.

[0031] Specifically, when establishing the water level-discharge relationship curve, historical rainfall sequences are used as independent variables, and water level observations for the corresponding time periods are used as dependent variables. Polynomial fitting or piecewise linear fitting is employed to obtain the mapping function between water level and discharge, and the phase difference of this function on the time axis is calculated to characterize the response time shift of water level changes relative to rainfall input. By conducting differential analysis on the water level-discharge curves of different sub-basins, the unique dynamic response characteristics of each region are extracted, and a time difference matrix is ​​formed as the input parameter.

[0032] Furthermore, the dynamic response data is mapped to a distributed runoff generation and confluence model. Within the model, the correspondence between rainfall input time and water level change time in each sub-basin is compared, and the time interval between rainfall occurrence and water level response is calculated to form a rainfall response delay dataset. This data is used to characterize the hydrological lag effect in each sub-basin. By analyzing the rainfall response delay, the rainfall estimates for each time period in the rainfall radar observation data are delayed and corrected to obtain the delay measurement data.

[0033] Specifically, using radar observation time series as a benchmark, the delay time is applied as an offset variable to the rainfall estimate at each observation time, thereby generating a time-corrected rainfall data series to eliminate the time difference between radar measurements and actual water level response.

[0034] Furthermore, delayed time units are extracted from the rainfall response delay dataset. These time units reflect the average response time of different sub-basins under different rainfall events. Using these units as time-sliding windows, the actual rainfall data observed by the rain-measuring radar after the delayed time unit is extrapolated from the delayed measurement data. This actual rainfall data is then used to update the model input field. Subsequently, the corrected actual rainfall data is remapped to the distributed runoff generation and confluence model. Runoff generation calculations are performed for each sub-basin during the time period corresponding to the delayed time unit to obtain the runoff generation sequence for that time period. By changing the length of the delayed time unit and the sliding step size, the rainfall input and flow response for multiple future time periods are progressively extrapolated, thereby generating a complete runoff generation sequence for each sub-basin during future time periods.

[0035] Preferably, the analysis of rainfall response delay based on the water level-discharge relationship curve and the distributed runoff generation and confluence model includes: Extract the time difference between water level changes and flow rate changes from the water level-flow rate relationship curve to obtain dynamic response data; Identify the correlation between water level rise and flow rate changes under different rainfall intensities based on dynamic response data; Based on the correspondence and the distributed runoff generation and confluence model, the water level change time of different sub-basins during the same rainfall process is calculated to obtain the water level response time of the sub-basin. The time difference between the water level change in each sub-basin and the preset rainfall start point is calculated by using the sub-basin water level response time to obtain the rainfall response delay.

[0036] In some embodiments, the rainfall response delay is analyzed based on the water level-discharge relationship curve and a distributed runoff generation and confluence model to quantitatively characterize the time lag characteristics of water level changes relative to rainfall input, thereby providing a dynamic time-series correction basis for future flow forecasts. Specifically, by synchronously pairing and analyzing the time-series differences between water level and flow sequences in historical observation data, the time difference characteristics of water level changes relative to flow changes in the water level-discharge relationship curve are extracted, and dynamic response data is formed based on this. This dynamic response data records the offset of the peak water level change relative to the peak flow in the form of a time difference matrix, and its physical meaning reflects the watershed's ability to store and retain rainfall input.

[0037] Furthermore, based on the dynamic response data, the correlation between water level rise and flow change under different rainfall intensities is identified. By analyzing the rainfall intensity classification and corresponding water level change rates of multiple rainfall events, a rainfall intensity-water level response function model is established to calculate the influence of rainfall intensity changes on the rate of water level rise. The slope parameter of the fitted function characterizes the coupling strength between rainfall and water level. This model can distinguish the differences in water level response under three scenarios: weak rainfall, continuous moderate rainfall, and strong convective rainfall, providing partition weights for subsequent sub-basin response calculations.

[0038] Furthermore, based on the rainfall intensity-water level response correspondence and the hydrological response structure of the distributed runoff generation and confluence model, calculations are performed on different sub-basins during the same rainfall process. Combining the topographic slope, runoff path length, and surface infiltration parameters of each sub-basin, the water level change time in each region is determined, thereby obtaining the water level response time distribution of the sub-basin.

[0039] Specifically, in the simulation model, the rainfall start time is used as a unified time reference. The peak water level times of the runoff generation and confluence calculation units within each sub-basin are statistically analyzed. By comparing the response time series of multiple grid units within the sub-basin, the regional average water level response time is calculated, and a spatialized response time distribution map is formed to represent the overall hydrological response delay characteristics of the basin.

[0040] Furthermore, the time difference between the water level change in each sub-basin and the preset rainfall initiation time is calculated using the sub-basin water level response time to obtain the rainfall response delay. This process obtains the delay time unit for each region by subtracting the rainfall initiation time from the sub-basin water level response time, and quantifies the lag degree of the hydrological system by the length of the time unit. By aggregating the delay time units of all sub-basins to form a rainfall response delay dataset, the response differences between different regions can be analyzed at a spatial scale, thereby providing accurate time-series control parameters for delay correction of rainfall radar observation data and flood process prediction.

[0041] Preferably, calculating the water level change time in different sub-basins during the same rainfall process based on the correspondence and the distributed runoff generation and confluence model includes: The correspondence is mapped to the spatial grid of the distributed runoff generation and confluence model, and the starting time of the water level rise is obtained by identifying the time node when the water level first rises significantly in each sub-basin. The water level rise start time is superimposed with the water level change rate corresponding to the spatial grid point, and the time difference is calculated to generate the water level change time of different sub-basins during the same rainfall process.

[0042] In some embodiments, the dynamic temporal mapping characteristics between rainfall input and water level response within a watershed are characterized by calculating the water level change time of different sub-basins during the same rainfall process based on the correspondence and a distributed runoff generation and confluence model. Specifically, by mapping the correspondence between rainfall intensity and water level change rate to the spatial grid points of the distributed runoff generation and confluence model, a correlation parameter table of rainfall input and water level change rate is established within each grid cell. Combined with the watershed topographic elevation model (DEM) and river network topology, a hydrological response mapping structure with spatial constraints is formed. Under this mapping structure, the water level time series of each sub-basin grid cell in the model calculation results is extracted. By monitoring the trend of the first derivative change of the water level curve, the time node of the first significant rise is identified, and the timestamp corresponding to the node is taken as the starting time of the water level rise in that cell.

[0043] Furthermore, the starting time of the water level rise is superimposed with the water level change rate data in the corresponding spatial grid points to obtain a joint characteristic curve of time and rate during the water level rise process. The time difference integral of the curve is then calculated to obtain the water level change time of each sub-basin during the entire rainfall event.

[0044] Specifically, the rate of change function for each spatial grid point during the water level rise. Perform time integration to calculate from the start of the ascent. At peak time The cumulative water level change, and in The time difference represents the duration of water level changes. By performing the above calculations grid-by-grid in the distributed runoff generation and confluence model, and summarizing the time difference distribution results of each grid point, a temporal and spatial distribution map of water level changes in different sub-basins during the same rainfall process is obtained.

[0045] Furthermore, by calculating the statistical mean and standard deviation of water level change time in sub-basins, the concentration and differences in response time in different regions can be analyzed, thereby providing high-precision time-series input parameters for subsequent rainfall response delay analysis and realizing a joint spatiotemporal quantitative characterization of water level evolution within the basin.

[0046] Preferably, the process of superimposing the initial moment of water level rise with the rate of water level change corresponding to the spatial grid points and calculating the time difference includes: By superimposing the water level rise start time and the water level change rate of that grid point on each spatial grid point in the distributed runoff generation and confluence model, a time-rate coupled trajectory is formed. The time difference corresponding to the inflection point of the time-rate coupled trajectory is used as the water level change time, thereby obtaining the water level change time of different sub-basins during the same rainfall process.

[0047] In this embodiment, in some embodiments, by superimposing the water level rise start time and the water level change rate of that grid point on each spatial grid point in the distributed runoff generation and runoff model, a time rate coupling trajectory reflecting the relationship between temporal changes and dynamic processes is constructed, thereby realizing a quantitative characterization of the water level evolution process in different sub-basins.

[0048] Specifically, in the grid structure of the distributed runoff generation and confluence model, each spatial grid point is used as the analysis unit to extract the corresponding water level rise start time. and water level change rate sequence And by overlaying the timeline, a dynamic trajectory of water level changes is formed. This allows the water level evolution at each grid point to be continuously traceable in the time dimension. The trajectory function can simultaneously reflect both the instantaneous rate of change and the cumulative change in water level, thus providing a data foundation for calculating the time of water level changes.

[0049] Furthermore, by analyzing the changing characteristics of the curve morphology in the time-rate coupled trajectory, the boundary between the rising and stable water levels, i.e., the inflection point of the trajectory, is identified. Specifically, the second derivative of the trajectory curve is used... As an indicator, when it changes from positive to negative, it indicates that the water level change has transitioned from an accelerating phase to a decelerating phase; this time point corresponds to... This was determined to be the end of the rising water level phase. The time of this inflection point was calculated. With the start of the water level rise Time difference between This time difference is used as the water level change time for that spatial grid point.

[0050] Furthermore, the water level change times of all spatial grid points are aggregated to form a spatial distribution matrix. Spatial aggregation is then performed based on sub-basin boundaries to calculate the average water level change time and its spatial standard deviation for each sub-basin, thus obtaining the water level change time distribution results for different sub-basins during the same rainfall process. Using this method, with time-rate coupled trajectories as the core analysis object, adaptive detection and temporal feature extraction of water level change processes can be achieved without relying on external empirical parameters, effectively improving the temporal resolution of distributed runoff generation and confluence models for hydrological response processes.

[0051] Preferably, in the distributed runoff generation and confluence model, the following steps are taken: superimposing the water level rise start time and the water level change rate at each spatial grid point: In the distributed runoff generation and confluence model, the starting time of water level rise for each spatial grid point is determined, and the water level change sequence of the corresponding grid point before and after the starting time is recorded. Extract the rate of water level change in adjacent time periods from the water level change sequence, pair the rate of water level change with the starting time of the water level rise, and generate gridded time rates containing the correspondence between time and rate. Construct a time rate curve using grid point time rates; Extract the inflection points in the time rate curves, and based on all inflection points, combine the curve segments between each inflection point in layers to construct a time rate coupled trajectory.

[0052] In this embodiment, in some embodiments, the time series characteristics of water level changes are analyzed and coupled trajectories are constructed by superimposing the water level rise start time and the water level change rate of that grid point on each spatial grid point in the distributed runoff generation and confluence model.

[0053] Specifically, in the distributed runoff generation and confluence model, the starting time of water level rise is determined for each spatial grid point, and the water level change sequence before and after the starting time of that grid point is recorded to obtain the complete water level change process. The water level change rate of adjacent time periods in the water level change sequence is extracted, and each water level change rate is paired with the corresponding starting time of water level rise to generate a grid point time rate dataset containing the correspondence between time and rate, so as to reflect the dynamic characteristics of water level change at each grid point.

[0054] Furthermore, a time rate curve is constructed using grid point time rates to continuously display the rate of water level change at grid points over different time periods. The curve reflects the trend and acceleration or deceleration phases of water level change. Subsequently, key inflection points in the time rate curve are extracted, including the time points when the rate of change increases or decreases significantly or changes. Based on all grid point inflection points, the curve segments between each inflection point are layered and combined. A time rate coupling trajectory is constructed through a hierarchical structure to reflect the temporal correlation and propagation characteristics of water level changes at different spatial grid points.

[0055] It should be noted that this coupled trajectory can be used for subsequent calculation of water level change propagation velocity and analysis of flood response in sub-basins, and ensures the accuracy of curve segment combination through precise timestamp and rate matching.

[0056] Preferably, in step S3, the runoff-generating sequences of each sub-basin for future periods are sequentially merged and superimposed to predict the flood forecast sequence, including: Based on the spatial distribution of flow changes in the runoff generation sequence of each sub-basin in future time periods, the direction of the confluence path is redefined; Reconstruct the confluence channel based on the direction of the confluence path; Calculate the time propagation delay of the flow on the reconstructed confluence channel and establish the time correspondence of the flow; Based on the time correspondence of flow, the runoff generation sequences of each associated sub-basin for future periods are superimposed step by step to obtain continuous flow transmission results of multiple nodes; Flood forecast sequences are predicted by analyzing the transmission trends of continuous flow transmission results across multiple nodes.

[0057] In some embodiments, based on the spatial distribution of flow changes in the runoff sequence of each sub-basin in future time periods, the flow direction and topographic slope characteristics between each sub-basin are analyzed, and the direction of the confluence path is re-determined to ensure that the order of flow convergence is consistent with the actual hydraulic conditions.

[0058] Furthermore, based on the determined confluence path direction, the confluence channel is reconstructed to clarify the confluence nodes, divergence points, and confluence sequence of the flow in each sub-basin, providing a spatial reference for subsequent flow superposition. Subsequently, the propagation delay of the flow in time is calculated on the reconstructed confluence channel. By analyzing hydraulic parameters such as the length, slope, and friction coefficient of each channel segment, a flow-time correspondence is established to ensure that the runoff sequences of different sub-basins can be accurately matched in time.

[0059] Furthermore, based on the established flow-time correspondence, the runoff sequences of each associated sub-basin for future periods are superimposed level by level to generate continuous flow transmission results across multiple nodes. At the same time, node-level flow information and timestamps are recorded to facilitate flow tracking in space and time. Finally, through the transmission trend analysis of the continuous flow transmission results across multiple nodes, flood forecast sequences are predicted, and the peak flow time and duration are marked, providing a reliable basis for downstream water conservancy scheduling and flood warning.

[0060] It should be noted that the process must ensure that the runoff sequence is aligned in time and fully consider the runoff confluence delay and path coupling characteristics between sub-basins.

[0061] Preferably, calculating the time propagation delay of the flow on the reconstructed confluence channel and establishing the flow-time correspondence includes: Identify and reconstruct local accumulation nodes for flow propagation in the confluence channel; Using local accumulation nodes as boundaries, the reconstructed confluence channel is divided into multiple propagation zones; Calculate the average propagation time for each propagation partition and construct the propagation time series for each partition; Based on the partitioned propagation time series, the peak position of traffic propagation is tracked segment by segment from upstream to downstream, the arrival time of traffic peaks at each node is recorded, and traffic arrival time distribution data along the route is generated. By combining the arrival time distribution data of the flow along the route and reconstructing the confluence channel, upstream and downstream differential calculations are performed to obtain the propagation delay time of each segment; The time nodes corresponding to each segment in the partition propagation time series are matched according to the propagation delay time of each segment to establish the traffic time correspondence.

[0062] In some embodiments, by calculating the time propagation delay of traffic on the reconstructed confluence channel, an accurate traffic time correspondence is established, and local nodes where traffic may accumulate in the reconstructed confluence channel are identified, i.e., local accumulation nodes. These nodes are usually located at channel changes or confluences and may cause lag in traffic transmission.

[0063] Furthermore, using local accumulation nodes as boundaries, the reconstructed confluence channel is divided into multiple propagation zones, each corresponding to an independent flow propagation unit from upstream input to downstream output. Subsequently, the average propagation time of each propagation zone is calculated, taking into account hydraulic parameters such as zone length, slope, channel cross-sectional characteristics, and friction coefficient, and a zone propagation time series is constructed to reflect the temporal evolution characteristics of flow in each zone.

[0064] Furthermore, based on the partitioned propagation time series, the peak positions of the traffic flow are tracked segment by segment from upstream to downstream, and the arrival time of the traffic peak at each node is recorded to generate traffic arrival time distribution data along the route, so as to clarify the temporal response order of each node. Subsequently, combined with the traffic arrival time distribution data along the route and the spatial information of the reconstructed confluence channel, differential calculation is performed on adjacent upstream and downstream nodes to obtain the propagation delay time of each segment, thereby quantifying the lag characteristics of traffic propagation in the channel. Finally, the time nodes corresponding to each segment in the partitioned propagation time series are matched according to the propagation delay time of each segment, and the segment delay is aligned with the partitioned sequence to establish a complete traffic time correspondence.

[0065] Preferably, the stepwise superposition of runoff sequences for future periods of each associated sub-basin based on the flow-time correspondence includes: Starting with the local peak and trough characteristics in the future runoff sequence of each sub-basin, the runoff sequence is matched and aligned by combining the flow-time correspondence to obtain local time-series matching data. Based on local time-series matching data, the runoff generation sequences of adjacent sub-basins are superimposed step by step according to time delay to form multi-node continuous flow data; Calculate the gradient change data of continuous traffic data across multiple nodes; By removing abnormally abrupt node traffic data from multi-node continuous traffic data based on gradient change data, the continuous traffic transmission results of multi-nodes can be obtained.

[0066] In some embodiments, the runoff sequences of future periods of each associated sub-basin are superimposed step by step based on the flow time correspondence to obtain a continuous flow transmission result for multiple nodes.

[0067] Specifically, starting with the local peak and trough characteristics in the future runoff sequence of each sub-basin, the runoff sequence of each sub-basin is matched and aligned through the flow-time correspondence.

[0068] Furthermore, the obtained local time-series matching data is used to guide the time superposition of runoff sequences in adjacent sub-basins, and multi-node continuous flow data is generated by sequentially superimposing the data according to the time delay between each sub-basin.

[0069] By using continuous traffic data from multiple nodes, the traffic gradient change data at each time node is calculated to reflect the upward or downward trend of traffic over time.

[0070] Furthermore, based on gradient change data, abnormal abrupt nodes in multi-node continuous traffic data are identified and removed, i.e., non-true traffic values ​​caused by local observation errors or model fluctuations, thereby obtaining smooth and continuous multi-node traffic transmission results.

[0071] It should be noted that this method ensures that the runoff generation sequences of each sub-basin are precisely aligned in time, and takes into account the dynamic characteristics of local peaks and troughs, thereby improving the continuity and accuracy of the flow superposition results and providing a reliable basis for the generation of subsequent flood forecast sequences.

[0072] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0073] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A flood forecasting and scheduling implementation method based on rainfall radar, characterized in that, Includes the following steps: Step S1: Record the historical records of each hydrological station and the rainfall radar observation data; Identify and remove precipitation echo interference data from rain-measuring radar observation data to obtain interference-free observation data; Step S2: Construct a distributed runoff generation and confluence model by combining historical hydrological records and disturbance prevention observation data; Step S3: Calculate the runoff sequence for each sub-basin in the future using a distributed runoff generation and confluence model; then, sequentially merge and superimpose the runoff sequences for each sub-basin in the future to predict the flood forecast sequence. Step S4: Identify high-risk flood areas based on flood forecast sequences; for high-risk flood areas, determine pre-discharge scheduling plans by combining flood forecast sequences with preset reservoir hydrological data; Step S5: Real-time scheduling of reservoir resources according to the pre-planned flood discharge schedule to prevent the risk of flood overflow.

2. The flood forecasting and scheduling implementation method based on rainfall radar according to claim 1, characterized in that, Step S3, which calculates the runoff sequence for future periods in each sub-basin using a distributed runoff generation and confluence model, includes: A water level-discharge relationship curve was plotted based on historical rainfall data and historical water level data. Dynamic response data is generated by extracting the time difference characteristics of water level changes relative to flow rate changes from the water level-flow relationship curve. By mapping dynamic response data to a distributed runoff generation and confluence model and comparing the rainfall input time and water level change time of each sub-basin in the model, the time interval between rainfall occurrence and water level response is calculated to obtain the rainfall response delay. The delay measurement data of rainfall measurement radar observation data for each time period is calculated based on the rainfall response delay. Extract the delay time unit from the rainfall response delay; The actual rainfall data observed by the rain-measuring radar after the delay time unit is calculated by using delayed measurement data. By mapping actual rainfall data to a distributed runoff generation and confluence model, the runoff generation sequence of each sub-basin in the time delay unit is obtained. By changing the time delay unit, the delayed measurement data of rainfall data in each subsequent future time period is obtained, and the runoff generation sequence of each sub-basin in the future time period is generated.

3. The flood forecasting and scheduling implementation method based on rainfall radar according to claim 2, characterized in that, Analysis of rainfall response delay based on water level-discharge relationship curves and distributed runoff generation-confluence model includes: Extract the time difference between water level changes and flow rate changes from the water level-flow rate relationship curve to obtain dynamic response data; Identify the correlation between water level rise and flow rate changes under different rainfall intensities based on dynamic response data; Based on the correspondence and the distributed runoff generation and confluence model, the water level change time of different sub-basins during the same rainfall process is calculated to obtain the water level response time of the sub-basin. The time difference between the water level change in each sub-basin and the preset rainfall start point is calculated by using the sub-basin water level response time to obtain the rainfall response delay.

4. The flood forecasting and scheduling implementation method based on rainfall radar according to claim 3, characterized in that, Based on the correspondence and the distributed runoff generation and confluence model, the calculation of water level changes in different sub-basins during the same rainfall process includes: The correspondence is mapped to the spatial grid of the distributed runoff generation and confluence model, and the starting time of the water level rise is obtained by identifying the time node when the water level first rises significantly in each sub-basin. The water level rise start time is superimposed with the water level change rate corresponding to the spatial grid point, and the time difference is calculated to generate the water level change time of different sub-basins during the same rainfall process.

5. The flood forecasting and scheduling implementation method based on rainfall radar according to claim 4, characterized in that, The initial time of water level rise is superimposed with the rate of water level change corresponding to spatial grid points, and the time difference is calculated, including: By superimposing the water level rise start time and the water level change rate of that grid point on each spatial grid point in the distributed runoff generation and confluence model, a time-rate coupled trajectory is formed. The time difference corresponding to the inflection point of the time-rate coupled trajectory is used as the water level change time, thereby obtaining the water level change time of different sub-basins during the same rainfall process.

6. The flood forecasting and scheduling implementation method based on rainfall radar according to claim 5, characterized in that, By superimposing the initial time of water level rise and the rate of water level change at that grid point on each spatial grid point in the distributed runoff generation and confluence model, the following parameters are considered: In the distributed runoff generation and confluence model, the starting time of water level rise for each spatial grid point is determined, and the water level change sequence of the corresponding grid point before and after the starting time is recorded. Extract the rate of water level change in adjacent time periods from the water level change sequence, pair the rate of water level change with the starting time of the water level rise, and generate gridded time rates containing the correspondence between time and rate. Construct a time rate curve using grid point time rates; Extract the inflection points in the time rate curves, and based on all inflection points, combine the curve segments between each inflection point in layers to construct a time rate coupled trajectory.

7. The flood forecasting and scheduling implementation method based on rainfall radar according to claim 1, characterized in that, Step S3 involves sequentially converging and superimposing the runoff sequences for future periods from each sub-basin to predict the flood forecast sequence, including: Based on the spatial distribution of flow changes in the runoff generation sequence of each sub-basin in future time periods, the direction of the confluence path is redefined; Reconstruct the confluence channel based on the direction of the confluence path; Calculate the time propagation delay of the flow on the reconstructed confluence channel and establish the time correspondence of the flow; Based on the time correspondence of flow, the runoff generation sequences of each associated sub-basin for future periods are superimposed step by step to obtain continuous flow transmission results of multiple nodes; Flood forecast sequences are predicted by analyzing the transmission trends of continuous flow transmission results across multiple nodes.

8. The flood forecasting and scheduling implementation method based on rainfall radar according to claim 7, characterized in that, Calculating the time propagation delay of traffic on the reconstructed confluence channel and establishing the traffic-time correspondence includes: Identify and reconstruct local accumulation nodes for flow propagation in the confluence channel; Using local accumulation nodes as boundaries, the reconstructed confluence channel is divided into multiple propagation zones; Calculate the average propagation time for each propagation partition and construct the propagation time series for each partition; Based on the partitioned propagation time series, the peak position of traffic propagation is tracked segment by segment from upstream to downstream, the arrival time of traffic peaks at each node is recorded, and traffic arrival time distribution data along the route is generated. By combining the arrival time distribution data of the flow along the route and reconstructing the confluence channel, upstream and downstream differential calculations are performed to obtain the propagation delay time of each segment; The time nodes corresponding to each segment in the partition propagation time series are matched according to the propagation delay time of each segment to establish the traffic time correspondence.

9. The flood forecasting and scheduling implementation method based on rainfall radar according to claim 7, characterized in that, Based on the temporal correspondence of flow rates, the runoff sequences for future periods of each associated sub-basin are superimposed step by step, including: Starting with the local peak and trough characteristics in the future runoff sequence of each sub-basin, the runoff sequence is matched and aligned by combining the flow-time correspondence to obtain local time-series matching data. Based on local time-series matching data, the runoff generation sequences of adjacent sub-basins are superimposed step by step according to time delay to form multi-node continuous flow data; Calculate the gradient change data of continuous traffic data across multiple nodes; By removing abnormally abrupt node traffic data from multi-node continuous traffic data based on gradient change data, the continuous traffic transmission results of multi-nodes can be obtained.

10. A flood forecasting and scheduling system based on rainfall radar, characterized in that, For executing the flood forecasting and scheduling implementation method based on rainfall radar as described in claim 1, the flood forecasting and scheduling implementation system based on rainfall radar includes: The data preprocessing module is used to record historical data of each hydrological station and rainfall radar observation data; identify precipitation echo interference data in the rainfall radar observation data, remove precipitation echo interference data, and confirm the anti-interference observation data. The model building module is used to construct a distributed runoff generation and confluence model by combining historical hydrological records and disturbance prevention observation data; The sequence prediction module is used to simulate and calculate the runoff sequence of each sub-basin for future periods through a distributed runoff generation and confluence model; the runoff sequences of each sub-basin for future periods are then merged and superimposed to predict the flood forecast sequence. The scheduling design module is used to identify high-risk flood areas based on flood forecast sequences; for high-risk flood areas, it determines a pre-discharge scheduling plan by combining the flood sequence with preset reservoir hydrological data. The scheduling and prevention module is used to schedule reservoir resources in real time according to the pre-planned flood discharge schedule in order to prevent flood risks.

Images