No-observation-station river reach flow inversion method, device and equipment based on satellite remote sensing

Abstract

The invention relates to a measurement-station-free river reach flow inversion method, device and equipment based on satellite remote sensing. The method comprises the following steps: respectively determining observation data sets of at least three continuous river reaches according to a grid image data set observed by a satellite, wherein the observation data sets comprise observation data and observation moments corresponding to the observation data; performing optimization processing on the observation data to obtain optimized observation data; according to the optimized observation data, determining a common observation moment of each river reach and target observation data of each river reach at the common observation moment; the hydraulic parameters are calibrated according to the target observation data of each river reach and a mass conservation equation, and calibrated hydraulic parameters are obtained; and for each river reach, determining a flow time sequence of the river reach according to the calibrated hydraulic parameters and the optimized observation data. The method can get rid of dependence on measured data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite remote sensing flow inversion, in particular to a satellite remote sensing based flow inversion method, device and equipment for river sections without gauging stations. BACKGROUND

[0002] Although the traditional hydrological monitoring station network can provide accurate flow observation data for specific river sections, it is difficult to meet the demand for systematic cognition of large-scale regional hydrological processes due to insufficient station distribution density. The new generation of satellite-borne radar altimeter represented by the SWOT satellite provides a new technical means for river monitoring with its extensive observation coverage.

[0003] In related technologies, the river flow inversion method usually uses the flow-stage or flow-width relationship curve, and only uses a single observation variable to estimate the flow. However, this kind of method usually relies on hydraulic parameters that cannot be directly observed by satellite, and thus must be calibrated based on the measured data from ground stations, so it cannot get rid of the dependence on ground stations, resulting in the inability to perform flow inversion on river sections lacking ground stations. SUMMARY

[0004] Therefore, it is necessary to provide a satellite remote sensing based flow inversion method, device and equipment for river sections without gauging stations, which can get rid of the dependence on measured data.

[0005] In a first aspect, the present application provides a satellite remote sensing based flow inversion method for river sections without gauging stations, comprising: determining observation data sets of at least three consecutive river sections according to satellite observed grid image data sets, wherein the observation data sets include observation data and observation time corresponding to the observation data; performing optimization processing on the observation data to obtain optimized observation data; determining common observation time of each river section and target observation data of each river section at the common observation time according to the optimized observation data; calibrating hydraulic parameters according to the target observation data of each river section and the mass conservation equation to obtain calibrated hydraulic parameters; and for each river section, determining the flow time series of the river section according to the calibrated hydraulic parameters and the optimized observation data.

[0006] In one of the embodiments, the observation data includes water level values and slope values of the river, and the optimization of the observation data to obtain the optimized observation data includes: determining abnormal water level values and abnormal slope values of each river section based on a river section change consistency principle; wherein the river section change consistency principle indicates that the change trends of all river sections at a common observation time are the same; the abnormal water level values are removed to obtain the optimized water level values of the river section; and the abnormal slope values are corrected to obtain the optimized slope values of the river section.

[0007] In one of the embodiments, the observation data includes width values of the river, and the optimization of the observation data to obtain the optimized observation data includes: correcting abnormal width values in a time sequence of the width values of each river section to obtain a corrected time sequence of the width values; and smoothing the corrected time sequence of the width values to obtain an optimized time sequence of the width values.

[0008] In one of the embodiments, the observation data includes width values of the river, and the determination of the observation data sets of the at least three continuous river sections from the satellite observation raster image data set includes: extracting a river channel mask of each river section from each raster image in the raster image data set; in the river channel mask, a sliding window with a preset river length is used to traverse each sub-river section, and a target sub-river section is determined according to a stability evaluation index of each sub-river section; and an average river width of each target sub-river section is taken as the width value of the river section to obtain a time sequence of the width values of the river section.

[0009] In one of the embodiments, the calibration of the hydrodynamic parameters according to the target observation data of each river section and the mass conservation equation to obtain the calibrated hydrodynamic parameters includes: constructing a likelihood function of Bayesian inference according to the target observation data and the mass conservation equation; determining a posterior probability distribution of the hydrodynamic parameters according to the likelihood function and a prior probability distribution of the hydrodynamic parameters; and determining the calibrated hydrodynamic parameters according to the posterior probability distribution.

[0010] In one of the embodiments, the construction of the likelihood function of Bayesian inference according to the target observation data and the mass conservation equation includes: determining mass residuals of each river section at each common observation time according to the target observation data and the mass conservation equation, wherein the mass residuals are used to represent the imbalance degree of the water quantity of the river section; aggregating all the mass residuals to obtain an overall mass residual; and constructing the likelihood function of Bayesian inference according to the overall mass residual.

[0011] In one of the embodiments, the abnormal water level value of each river section is determined based on the consistency principle of river section change, including: identifying suspicious water level values in the water level values of the river section based on the water level change in the river section; determining a water level mutation threshold corresponding to each suspicious water level value based on the water level values of other river sections; and determining the suspicious water level value as the abnormal water level value in the case that the difference between the suspicious water level value and the water level value at the adjacent observation time in the river section exceeds the water level mutation threshold.

[0012] In one of the embodiments, the abnormal slope value of each river section is determined based on the consistency principle of river section change, including: identifying suspicious slope values in the slope values of the river section based on the slope change in the river section; determining a slope mutation threshold corresponding to each suspicious slope value according to the slope values of other river sections; and determining the suspicious slope value as the abnormal slope value in the case that the ratio between the suspicious slope value and the median value of the slope of the river section exceeds the slope mutation threshold.

[0013] In the second aspect, the application further provides a satellite remote sensing based un-gauge river section flow inversion device, including: a first determining module configured to determine observation data sets of at least three continuous river sections respectively according to satellite observed grid image data sets, the observation data set including observation data and an observation time corresponding to the observation data; an optimization module configured to perform optimization processing on the observation data to obtain optimized observation data; a second determining module configured to determine a common observation time of each river section and target observation data of each river section at the common observation time according to the optimized observation data; a rating module configured to rate hydraulic parameters according to the target observation data of each river section and a mass conservation equation to obtain rated hydraulic parameters; and a third determining module configured to determine a flow time series of each river section according to the rated hydraulic parameters and the optimized observation data.

[0014] In the third aspect, the application further provides a computer device including a memory and a processor, the memory storing a computer program, and the processor implementing the above method when executing the computer program.

[0015] In the fourth aspect, the application further provides a computer readable storage medium storing a computer program, and the computer program is executed by a processor to implement the above method.

[0016] In the fifth aspect, the application further provides a computer program product including a computer program, and the computer program is executed by a processor to implement the above method.

[0017] The satellite remote sensing based non-gauge river section flow inversion method, device, computer equipment, computer readable storage medium and computer program product, determine at least three continuous river section observation data sets according to the satellite observed grid image data set, the observation data set includes observation data and observation time corresponding to the observation data; the observation data is optimized to obtain the optimized observation data; according to the optimized observation data, the common observation time of each river section and the target observation data of each river section at the common observation time are determined; the hydraulic parameters are calibrated according to the target observation data of each river section and the mass conservation equation, and the calibrated hydraulic parameters are obtained; for each river section, the flow time series of the river section is determined according to the calibrated hydraulic parameters and the optimized observation data. The application embodiment realizes parameter calibration independent of the measured station and the measured data based on the target observation data and the mass conservation equation, and then makes the whole inversion process free from the dependence on the measured station and the measured data, so that the non-gauge flow inversion can be realized, and then the flow inversion of the river area lacking the ground measured station can be realized. In addition, the target observation data is the optimized observation data of the common observation time of each river section, which can ensure the accuracy of parameter calibration and inversion, and ensure the accuracy of the inverted flow time series. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the application embodiments or the related art, the drawings needed to be used in the application embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other related drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 An application environment diagram of a satellite remote sensing based non-gauge river section flow inversion method for an embodiment;

[0020] Figure 2 A flowchart of a satellite remote sensing based non-gauge river section flow inversion method for an embodiment;

[0021] Figure 3 A flowchart of step 202 in an embodiment;

[0022] Figure 4 A flowchart of step 201 in an embodiment;

[0023] Figure 5 A flowchart of step 202 in another embodiment;

[0024] Figure 6 A flowchart of step 204 in an embodiment;

[0025] Figure 7 Fig. 2 is a flowchart of a specific example of a method for estimating the flow of a river section without a gauging station;

[0026] Figure 8 Fig. 3 is a schematic diagram of four consecutive river sections in a specific example;

[0027] Figure 9 Fig. 4 is a graph of the time series of the optimized water level and width values of the river section 801;

[0028] Figure 10 Fig. 5 is a comparison graph of the estimated flow and the measured data of the river section 801;

[0029] Figure 11 Fig. 6 is a block diagram of a satellite remote sensing based device for estimating the flow of a river section without a gauging station in an embodiment;

[0030] Figure 12 Fig. 7 is an internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0031] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0032] It should be noted that the terms "first", "second", and the like used in the present application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "include" and "have" and any variations thereof used in the present application are intended to cover non-exclusive inclusion. The term "multiple" used in the present application refers to two or more. The term "and / or" used in the present application refers to one of the options or any combination of multiple options.

[0033] In the related art, the river flow inversion method mostly uses the flow-water level or flow-width relationship curve, and only uses a single observation variable to estimate the flow. However, such a method usually relies on hydraulic parameters that cannot be directly observed by satellite, and thus must be calibrated based on the measured data from the ground station, so it cannot be used to estimate the flow of the river channel area lacking the ground station. In addition, such a method needs to assume the shape of the river section in advance, which limits the universality of the method, and the calibration result is difficult to apply to the upstream and downstream.

[0034] To achieve remote sensing flow inversion without relying on actual measurement stations and data, this application proposes a station-free river segment flow inversion method based on satellite remote sensing. This method systematically identifies and processes outliers in water level and slope observation data, reducing the impact of outliers while maintaining a sufficient number of observations. The method of extracting river width based on raster imagery significantly reduces noise in width observations and effectively improves the consistency between width and water level. The algorithm introduces a flow conservation relationship through a constructed spatiotemporal observation matrix, constraining parameters in the correct direction and overcoming the limitation of traditional flow inversion methods relying on actual measurement data. This application embodiment achieves accurate flow estimation using only satellite observation data, and has significant application value for flow estimation in watersheds with no or insufficient data. The method of this application embodiment is described in detail below.

[0035] The satellite remote sensing-based method for inverting river flow in stations without monitoring stations provided in this application can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on other network servers. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0036] In one exemplary embodiment, such as Figure 2 As shown, a method for inverting river flow in a stationless section based on satellite remote sensing is provided, and this method is applied to... Figure 1 Taking the server in the example, the explanation includes the following steps 201 to 205. Wherein:

[0037] Step 201: Determine the observation datasets for at least three consecutive river segments based on the raster image datasets from satellite observations. The observation datasets include the observation data and the observation times corresponding to the observation data.

[0038] The satellite can be a wide-swath interferometry satellite, such as the SWOT satellite. The raster image dataset refers to the time series of raster images of the river (raster images and their corresponding observation times). The observation dataset refers to the dataset used to invert river flow, including the time series of observation data (observation data and their corresponding observation times). The observation data can include three types: river water level values, water surface slope values, and width values. Each of at least three consecutive river segments corresponds to one observation dataset. The number of observation times may be the same or different for each river segment. At least some observation times must be shared among the river segments.

[0039] For example, firstly, continuous river segments are selected from a priori river shape dataset as the river segments to be inverted, requiring at least three segments to ensure the effective construction of the water conservation relationship between the upstream and downstream of the river. Then, at least three continuous river segment raster image time series B (B is a subset of A) are selected from the raster image time series A of the river observed by satellite. Finally, based on the raster image time series B of at least three continuous river segments, the observation data time series of at least three continuous river segments are determined, that is, the water level value time series, slope value time series, and width value time series of at least three continuous river segments are extracted from B.

[0040] This strategy of selecting multiple river segments not only helps to enhance the physical constraints of flow estimation, but also reduces the impact of observation errors in a single river segment on the overall estimation results.

[0041] Step 202: Optimize the observation data to obtain optimized observation data.

[0042] The optimization process can include operations such as outlier removal, outlier correction, data cleaning, data replacement, and smoothing. Optimization can be applied to a single observation or the entire observation data sequence.

[0043] For example, optimization processes such as outlier removal and outlier correction are performed on at least one of the observed data of water level, slope and width to make the observed data more accurate and comprehensive, thereby obtaining the optimized time series of water level, slope and / or width values.

[0044] Step 203: Based on the optimized observation data, determine the common observation time for each river segment and the target observation data for each river segment at the common observation time.

[0045] The common observation time refers to the observation time that is the same (overlapping) among the observation data of each segment in the river segment to be inverted; that is, the intersection of the observation times of each segment. The observation data of each segment at each common observation time is called the target observation data, which is used for subsequent flow inversion. The target observation data includes water level, slope, and width values.

[0046] For example, after optimizing the observation data of each river segment in the river segment to be inverted, firstly, for the optimized observation dataset (including the optimized observation data and its corresponding observation time), the intersection of the observation times in each optimized observation dataset is found to obtain the common observation time of each river segment. Then, for each river segment, the optimized observation data at the common observation time is obtained from its optimized observation dataset to obtain the target observation data. For example, if the number of river segments to be inverted is at least three consecutive river segments: W1, W2, and W3, and the observation times in the observation dataset of W1 are t1, t3, t5, t6, and t7, the observation times in the observation dataset of W2 are t1, t2, t3, t6, t7, and t8, and the observation times in the observation dataset of W3 are t1, t3, t4, t5, and t6, then the common observation times of the three river segments W1, W2, and W3 are t1, t3, and t6. Therefore, the target observation data of W1 at times t1, t3, and t6, the target observation data of W2 at times t1, t3, and t6 are obtained from the observation datasets of each river segment.

[0047] Step 204: Calibrate the hydraulic parameters based on the target observation data and mass conservation equation for each river section to obtain the calibrated hydraulic parameters.

[0048] The mass conservation equation states that for any given stretch of river, water cannot be created or destroyed. The difference between the mass of water flowing into and out of a stretch per unit time (the difference in upstream and downstream flow) equals the change in the mass of water stored within that stretch. Therefore, when calibrating hydraulic parameters using the mass conservation equation, at least three consecutive river sections are selected as constraints.

[0049] Hydraulic parameters include the average Manning roughness coefficient and the initial cross-sectional area of ​​the river segment (the cross-sectional area at the initial moment). The average Manning roughness coefficient reflects the influence of riverbed material, vegetation, and channel irregularities on flow resistance; this is a crucial but uncertain parameter in the inversion process and cannot be directly observed. The initial cross-sectional area defines the basic geometric dimensions of the river segment and is also unknown. Therefore, parameter calibration of the average Manning roughness coefficient and the initial cross-sectional area is necessary to ensure that the flow rate derived from them best satisfies the mass conservation equation and observational data.

[0050] For example, the calibration of hydraulic parameters is performed within a Bayesian inference framework, specifically through the Markov chain Monte Carlo method. First, based on observational data and hydrological experience, the prior probability distributions of hydraulic parameters (such as the Manning roughness coefficient and initial cross-sectional area) are determined. Then, the water conservation (mass conservation) equation is introduced into the Bayesian model to construct a likelihood term with upstream and downstream water conservation as the core constraint. Based on this, the Markov chain Monte Carlo method is used to perform multiple iterative sampling in the parameter space, gradually converging the sampling chain to the posterior probability distribution of the parameters. Finally, the optimal estimates of the hydraulic parameters are obtained from this posterior distribution, yielding the calibrated average Manning roughness coefficient and initial cross-sectional area. By introducing the upstream and downstream mass conservation equation, the calibration process becomes independent of actual measurement stations and their data.

[0051] Step 205: For each river segment, determine the flow time series of the river segment based on the calibrated hydraulic parameters and optimized observation data.

[0052] For example, after parameter calibration, for each river segment, the calibrated hydraulic parameters and the optimized observation data for that segment are calculated using the Manning formula to invert and obtain the continuous flow time series for that segment. For instance, for river segment W1, the calibrated average Manning roughness coefficient and initial cross-sectional area, along with the optimized observation data (water level, slope, and width) at times t1, t3, t5, t6, and t7, are substituted into the Manning formula to obtain the river flow at times t1, t3, t5, t6, and t7 for river segment W1, thus obtaining the flow time series.

[0053] The aforementioned satellite remote sensing-based method for station-free river flow inversion achieves parameter calibration independent of actual measurement stations and data based on target observation data and the mass conservation equation. This frees the entire inversion process from dependence on actual measurement stations and data, enabling station-free flow inversion and thus allowing flow inversion in river areas lacking ground-based measurement stations. Furthermore, the target observation data is optimized observation data from the common observation time of all river segments, ensuring the accuracy of parameter calibration and inversion, thereby guaranteeing the accuracy of the inverted flow time series.

[0054] The following describes a specific implementation method for optimizing observation data.

[0055] In one exemplary embodiment, the observation data includes river water level and slope values. In this embodiment, as... Figure 3 As shown, step 202 includes steps 301 to 303. Wherein:

[0056] Step 301: Determine the abnormal water level and abnormal slope values ​​for each river segment based on the principle of consistency in river segment changes. The principle of consistency in river segment changes means that all river segments exhibit the same trend of change at the same observation time.

[0057] For example, the observation dataset for each river segment includes time series of water level values ​​and time series of slope values. For each river segment, firstly, based on the river changes within the segment, at least one suspicious water level value is selected from the water level time series, and at least one suspicious slope value is selected from the slope time series. Then, based on the principle of consistency in river segment changes, anomalous water level values ​​are selected from the suspicious water level values, and anomalous slope values ​​are selected from the suspicious slope values.

[0058] Step 302: Remove abnormal water level values ​​to obtain the optimized water level values ​​for the river section.

[0059] Step 303: Correct the abnormal slope values ​​to obtain the optimized slope values ​​for the river section.

[0060] For example, abnormal water level values ​​are removed from the water level time series of each river segment to obtain an optimized water level time series. Abnormal slope values ​​in the slope time series of each river segment are corrected using an interpolation correction method based on the normal slope values ​​at adjacent times to maintain the continuity of slope values, resulting in a corrected slope time series, which is the optimized slope time series, used for subsequent hydraulic parameter calibration.

[0061] Therefore, in this embodiment, the outlier handling strategy involves directly removing water level values, while slope outliers are corrected using interpolation based on normal slope data from adjacent time periods to maintain data continuity. This differentiated processing fully considers the characteristics of different hydraulic variables and their impact on subsequent analysis.

[0062] Regarding the identification of anomalies in water level and slope values, this application proposes a systematic processing method based on river segment correlation, which mainly includes three stages: screening of obvious anomalies, identification of suspicious values, and judgment of river segment correlation. The following provides a detailed description of the identification of anomalies in water level and slope values.

[0063] In an exemplary embodiment, step 301, determining the abnormal water level value of each river segment based on the principle of consistency of river segment changes, includes: identifying suspicious water level values ​​among the water level values ​​of each river segment based on the water level changes in the river segment; determining the water level change threshold corresponding to each suspicious water level value based on the water level values ​​of other river segments; and determining the suspicious water level value as an abnormal water level value if the difference between the suspicious water level value and the water level value at an adjacent observation time in the river segment exceeds the water level change threshold.

[0064] Other river segments refer to river segments other than the current river segment (the river segment currently undergoing outlier identification). Each suspicious water level value corresponds to a water level mutation threshold, which is used to characterize the degree of mutation of the water level in other river segments at the same observation time as the suspicious water level value.

[0065] For example, for the water level time series of each river segment, firstly, based on the water level changes in the same river segment (such as water level fluctuation and the first median water level), obvious water level anomalies are screened and suspicious water level values ​​are identified. The water level fluctuation can be the difference between different first quantiles and second quantiles of the river segment, where the first quantile is greater than the second quantile. Optionally, firstly, water level values ​​with fluctuations higher than a first preset multiple of the first median water level in the same river segment, or water level values ​​with fluctuations lower than a second preset multiple of the first median water level in the same river segment, are identified as obvious abnormal water level values, where the second preset multiple is less than the first preset multiple. These obvious abnormal water level values ​​are then removed, resulting in the remaining water level time series, thereby achieving the screening of obvious water level anomalies. Then, for each remaining water level value, the adjacent water level values ​​at adjacent observation times are determined, and the difference between the remaining water level value and its adjacent water level value is calculated. If the difference exceeds the water level fluctuation of the third preset multiple, the corresponding remaining water level value is regarded as a suspicious water level value, thereby realizing the identification of suspicious water level values.

[0066] Then, for each suspicious water level value and its observation time, further judgment is made based on the correlation of river segments. At this time, the water level change threshold is determined based on the water level values ​​of other river segments, and the difference between the suspicious water level value and the water level value at adjacent observation times in the same river segment is calculated. If the difference exceeds the water level change threshold corresponding to the suspicious water level value or twice the water level change threshold, the suspicious water level value is determined as a (not obvious) abnormal water level value and is removed. If the difference does not exceed the water level change threshold or twice the water level change threshold, it means that the suspicious water level value is a normal water level value, which conforms to the river segment consistency principle and belongs to the normal hydrological response of the river. It is retained, and the retained water level value is the optimized water level value. Optionally, the suspected water level value A1 of the current river segment (e.g., W1) has a corresponding observation time. The water level value A2 of other river segments (e.g., W2 and W3) at the same observation time t3 as the suspected water level value A1 is calculated. The difference between this water level value A2 and the median second water level of that water level value A2 within a first preset period (e.g., 22 days) at that observation time t3 is calculated, resulting in multiple differences (e.g., two). The median of these differences is taken as the water level mutation threshold a corresponding to the suspected water level value A1. The difference between the suspected water level value A1 and the water level value at adjacent observation times in the same river segment is calculated. If the difference exceeds 2a, A1 is considered an abnormal water level value and is removed from the list.

[0067] For example, let's take the current river segment as W1 and the other river segments as W2 and W3. First, the difference between the 90th and 10th quantiles of the water level time series of W1 is taken as the water level fluctuation of W1. Water level values ​​with fluctuations higher than 1.8 times the median water level of the time series, or lower than 1 times the median water level fluctuation, are identified as obvious abnormal water level values ​​and removed, resulting in the remaining water level time series. Then, for each remaining water level value, the adjacent water level values ​​at adjacent observation times are determined, and the difference between the remaining water level value and its adjacent water level values ​​is calculated. If this difference exceeds 0.5 times the water level fluctuation, the remaining water level value is considered a suspicious water level value, assumed to be A1 at observation time t3. Finally, based on the correlation of river segments, the suspicious water level value A1 is further assessed. At this point, the median second water level over 22 days is determined for the water level values ​​at t3 in other river segments W2 and W3. The difference between the water level value at t3 in W2 and its median second water level in the current observation dataset is calculated, and the difference between the water level value at t3 in W3 and its median second water level in the current observation dataset is also calculated. The median of these two differences is taken as the water level change threshold 'a' corresponding to A1. If the absolute value of the difference between A1 and the median second water level over 22 days is greater than 2a, A1 is determined to be an abnormal water level value and is removed; otherwise, it is considered a normal hydrological response of the river and is retained.

[0068] This embodiment utilizes the principle of consistency in river segment changes. For some water level values ​​that deviate from normal values, but whose deviations in the same direction and magnitude are observed in adjacent river segments at the same time, this indicates a reasonable hydrological response and is therefore retained. By considering the consistency of changes in adjacent river segments, the method determines whether a water level value is normal based on whether its trend is consistent with that of other river segments at the same time, thus avoiding the erroneous rejection of normal water level values.

[0069] The process for handling outliers in slope is basically the same as that for water level, but there are some adjustments in the specific parameter settings. A detailed description follows.

[0070] In an exemplary embodiment, step 301, determining the abnormal slope value of each river segment based on the principle of consistency of river segment changes, includes: identifying suspicious slope values ​​among the slope values ​​of each river segment based on the slope changes in the river segment; determining the slope mutation threshold corresponding to each suspicious slope value according to the slope values ​​of other river segments; and determining the suspicious slope value as an abnormal slope value if the ratio between the suspicious slope value and the median slope of the river segment exceeds the slope mutation threshold.

[0071] Each suspected slope value corresponds to a slope mutation threshold, which is used to characterize the degree of mutation of slope values ​​in other river sections at the same observation time as the suspected slope value.

[0072] For example, for the slope value time series of each river segment, firstly, based on the slope changes within the same river segment (e.g., the first median slope), obvious slope anomalies are screened and suspicious slope values ​​are identified. Optionally, firstly, for each slope value, a first ratio between the slope value and the first median slope of the same river segment is calculated. Slope values ​​with a first ratio less than a first preset ratio, or slope values ​​with a first ratio greater than a second preset ratio, are identified as obvious slope anomalies. Where the first preset ratio is less than the second preset ratio, obvious slope anomalies are removed, resulting in the remaining slope value time series, thereby achieving the screening of obvious slope anomalies. Then, for each remaining slope value, the ratio between the remaining slope value and the first median slope is calculated. If this ratio exceeds a third preset ratio or is greater than a fourth preset ratio, the corresponding remaining slope value is identified as a suspicious slope value. Where the third preset ratio is greater than the fourth preset ratio, suspicious slope values ​​are identified.

[0073] Then, for each suspicious slope value and its observation time, further judgment is made based on the correlation of river segments. At this time, the slope change threshold is determined based on the slope values ​​of other river segments. If the first ratio corresponding to the suspicious slope value exceeds the corresponding slope change threshold or twice the slope change threshold, the suspicious slope value is determined as a (not obvious) abnormal slope value, and all abnormal slope values ​​are corrected by interpolation correction method. If the first ratio does not exceed the slope change threshold or twice the slope change threshold, it means that the suspicious slope value is a normal slope value, which conforms to the river segment consistency principle and belongs to the normal hydrological response of the river. No correction is made, and thus the optimized slope value is obtained. Optionally, the suspected slope value B1 of the current river segment (e.g., W1) has a corresponding observation time. The slope value B2 of other river segments (e.g., W2 and W3) at the same observation time t6 as the suspected slope value B1 is calculated. This is then compared with the median slope value of B2 within a second preset period (e.g., 45 days) at that observation time t6. Multiple second ratios (e.g., two) are obtained. The median of these second ratios is taken as the slope abrupt change threshold b corresponding to the suspected slope value B1. If the first ratio corresponding to the suspected slope value B1 exceeds 2b, then B1 is considered an abnormal slope value, and B1 is corrected using an interpolation correction method.

[0074] For example, let's take the current river segment as W1, and the other river segments as W2 and W3. First, the median of the time series of slope values ​​for W1 is used as the baseline slope value for W1. Slope values ​​that are more than 10 times higher than the baseline slope value, or less than 0.1 times the baseline slope value, are identified as significant abnormal slope values. Then, for each remaining slope value, the first ratio of the remaining slope value to the baseline slope is calculated. If this first ratio exceeds 3 or is less than 1 / 3, the remaining slope value is considered a suspicious slope value, let's assume it's B1 at observation time t3. Finally, based on the river segment correlation, the suspicious slope value B1 is further assessed. At this point, the second median slope values ​​at t3 for other river segments W2 and W3 are determined over 45 days. The second ratio of the slope value at t3 in W2 within the current observation dataset to its second median slope value is calculated, and the second ratio of the slope value at t3 in W3 within the current observation dataset to its second median slope value is also calculated. Two second ratios are obtained, and the median of these two ratios is taken as the slope abrupt change threshold b corresponding to B1. If the first ratio corresponding to B1 is greater than 2b, B1 is determined to be a suspicious slope value and corrected; otherwise, it is considered a normal hydrological response of the river and retained.

[0075] Therefore, this embodiment, based on the principle of consistency of river segment changes, determines whether a slope value is normal if, although it deviates from the normal value, the observation results of adjacent river segments at the same time also show deviations in the same direction and with similar magnitudes, indicating a reasonable hydrological response, and thus no correction is made. By considering the consistency of changes in adjacent river segments, the slope value is determined to be normal based on whether the changing trends of other river segments at the same time are consistent with it, thus avoiding the erroneous rejection of normal slope values.

[0076] Through the above-mentioned refined outlier identification and processing procedures, the previously incorrectly retained abnormal water levels were successfully removed, and the slope data was corrected, significantly improving the spatiotemporal consistency and reliability of hydraulic variable data.

[0077] The following describes the specific implementation method for extracting river width from raster images and performing outlier identification and optimization. To address the problem of high noise levels in existing river width data, this application's embodiments reasonably adopt an automatic river width data extraction strategy:

[0078] That is, in one exemplary embodiment, such as Figure 4 As shown, step 201 includes steps 401 to 403. Wherein:

[0079] Step 401: Extract the river channel mask for each river segment from each raster image in the raster image dataset.

[0080] For example, connectivity and morphology algorithms are used to automatically extract accurate channel masks for each river segment from a raster image dataset. Optionally, firstly, a buffer of 3 times the prior river width is established based on the river centerline of the river segment, and the raster image is cropped according to the buffer. Then, a strict buffer of 0.5 times the prior river width is established within the 3 times prior river width buffer based on the river centerline, and water pixels connected to water pixels within the strict buffer are used as preliminary channel masks. Then, a skeletonization method is used to refine the preliminary channel mask to obtain the accurate channel mask for the river segment, where each raster of the channel mask indicates whether it represents a river.

[0081] Step 402: Within the river channel mask, a sliding window with a preset river length is used to traverse each sub-river segment, and the target sub-river segment is determined based on the stability evaluation index of each sub-river segment.

[0082] The stability evaluation indicators include at least one of the following: satellite multi-orbit observation consistency, relative standard deviation of width, and channel straightness. Satellite multi-orbit observation consistency characterizes the consistency or degree of error of the width values ​​of different sub-river segments observed by satellites from different orbits, measuring the stability and reliability of the sub-river segment observation results. Relative standard deviation of width is the relative standard deviation of the width values ​​of the corresponding sub-river segment, measuring the uniformity of the width within the sub-river segment itself. It is calculated by dividing the standard deviation of the width values ​​by the average river width value within the same sub-river segment and multiplying by 100%. Channel straightness is calculated by dividing the straight-line distance between the two ends of a sub-river segment by the length of that sub-river segment.

[0083] For example, for each river segment, within the channel mask obtained in step 401, a sliding window of a preset river length (e.g., 3 km) is used to traverse all sub-segments of the preset river length for that river segment, sliding from the starting point to the ending point of the channel. The consistency of satellite multi-orbit observations, the relative standard deviation of width, and the straightness of the channel are calculated for each sub-segment. A score is assigned to each sub-segment based on these stability evaluation indicators. After scoring all windows, the sub-segment with the highest score is selected as the target sub-segment, which is the representative sub-segment of the given river segment. Optionally, the squared value of the channel straightness is calculated, and the weighted sum of the squared value, the consistency of satellite multi-orbit observations, and the relative standard deviation of the river width is used as the score. The calculation formula is as follows:

[0084] (1)

[0085] In the formula, Sinu represents the straightness of the river channel, ranging from 0 to 1. The closer the value is to 1, the straighter the river channel; the smaller the value, the more meandering the river channel. Corr is the Pearson correlation coefficient between the width distribution curves of sub-river segments along the river channel observed by satellites from different orbits, ranging from 0 to 1. The larger the value, the higher the stability and reliability of the sub-river segment; the smaller the value, the lower the stability and reliability of the sub-river segment. CV is the relative standard deviation of width. The smaller the value, the more gradual the width change of the sub-river segment; the larger the value, the more drastic the width change of the sub-river segment. 0.5, 0.3, and -0.2 are the weights of each indicator.

[0086] Step 403: Use the average width of each target sub-river segment as the width value of the river segment to obtain the time series of the width value of the river segment.

[0087] Each river segment will be assigned a target sub-segment, and each target sub-segment will have an average river width at different observation times. The width value of this river segment can be multiple.

[0088] For example, for each river segment corresponding to each raster image (corresponding to one observation time), its target sub-segment is taken as its representative sub-segment. The average width of the water coverage area in the target sub-segment of the corresponding raster image is calculated, and this average width is taken as the width value of the river segment at the corresponding observation time, thus obtaining the width value time series corresponding to the raster image dataset (raster image time series). For example, for river segment W1, the target sub-segment with the highest score in the raster image time series is determined, and the width values ​​of the target sub-segments in each raster image are used to construct the width value time series (obtained from multiple satellite observations).

[0089] Therefore, this embodiment uses multi-track consistency, relative standard deviation, and river straightness as reference indicators to select sub-segments with less interference and relatively regular river channels. The average width of the sub-segment is used as the width value of the corresponding river segment in one observation. Thus, the width values ​​observed multiple times and their observation times are used as the width value time series of the river segment, which can make the width value time series of each river segment have high quality and high representativeness.

[0090] Regarding the identification and handling of anomalies in width values, this application employs a filter for outlier optimization. The identification and optimization of width values ​​are described in detail below.

[0091] In one exemplary embodiment, the observation data includes the width value of the river. In this embodiment, as... Figure 5 As shown, step 202 includes steps 501 and 502. Wherein:

[0092] Step 501: Correct the abnormal width values ​​in the time series of width values ​​for each river segment to obtain the corrected width value time series.

[0093] For example, a Hampel filter is used to handle outlier width values. This filter employs a sliding window of length 7 (containing 7 width values). Within each sliding window, the absolute value of the difference between each of the 7 width values ​​and the median width within that window is calculated. The median of these absolute differences is then identified as the median absolute deviation, and three times the median absolute deviation is used as a threshold. If the difference between a width value and the median width within the window exceeds the threshold, the width value is identified as an outlier. The outlier width value is then corrected using the median width of its respective window, for example, by replacing it with the median width, resulting in a corrected width value time series for that river segment.

[0094] Step 502: Smooth the corrected width value time series to obtain the optimized width value time series.

[0095] For example, Gaussian smoothing is used for further filtering, and the calculation formula is as follows:

[0096]

[0097] In the formula, The width value before smoothing at the j-th observation time. t represents the smoothed width value at the i-th observation time, t is the observation time, and t0 is the Gaussian filter parameter, which can be 10 days. i For the i-th observation time, t j The j-th observation time.

[0098] According to formula (2), the time series of the corrected width value in step 501 is smoothed so that the river width series presents a smooth, physically reasonable curve that can reflect the real hydrodynamic process.

[0099] Therefore, this embodiment corrects and smooths out the abnormal width values ​​in the time series of the width values ​​of each river segment, uses filters to remove outliers, and reduces the problem of river width oscillation in a short period of time. This effectively suppresses the jagged edges and small jumps in the series, making the curve smoother and significantly reducing the noise of the width series. The smoothed width changes are more consistent with hydrodynamics, the time series shows a more reasonable trend, and the correlation with the water level changes in the same period is significantly enhanced.

[0100] After processing outliers in water level and slope, and extracting and optimizing width, the intersection of observation times (common observation times) of the selected river segments (at least three consecutive segments) is integrated to construct spatiotemporal two-dimensional observation matrices for water level, slope, and width. For example, for water level, a spatiotemporal two-dimensional observation matrix is ​​constructed based on the water level values ​​at the common observation times of all river segments, with river segments as rows and observation times as columns. This yields the spatiotemporal two-dimensional observation matrices for water level, slope, and width, which are used for calibrating the hydraulic parameters of the model.

[0101] In one exemplary embodiment, such as Figure 6 As shown, step 204 includes steps 601 to 603:

[0102] Step 601: Construct the likelihood function for Bayesian inference based on the target observation data and the mass conservation equation.

[0103] For example, firstly, the following river flow calculation model is constructed based on Manning's formula:

[0104]

[0105] In the formula, the average Manning roughness coefficient of the river section is... and initial cross-sectional area of ​​water passage These are the hydraulic parameters that need to be calibrated using a model; while the change in cross-sectional area of ​​the water passage... River width and water surface slope Variables such as these can be obtained directly from satellite observation data. Let be the river flow rate of river segment r at time t.

[0106] Then, the mass conservation equations of the upstream and downstream are used as physical constraints, and a Bayesian inference likelihood function is constructed based on the target observation data.

[0107] Step 602: Determine the posterior probability distribution of the hydraulic parameters based on the likelihood function and the prior probability distribution of the hydraulic parameters.

[0108] For example, based on target observation data or a two-dimensional spatiotemporal observation matrix, preliminary estimates of hydraulic parameters are calculated using empirical formulas to obtain prior probability distributions of hydraulic parameters. For instance, the Manning roughness can be based on a log-normal distribution according to channel type as a prior. Then, based on Bayesian inference, the posterior probability distribution of the parameters is determined. It is proportional to the product of the likelihood function and the prior probability distribution:

[0109]

[0110] In the formula, This represents a set of observational data on river water level, width, and slope. This represents the hydraulic parameters to be calibrated. For the prior distribution of hydraulic parameters, This represents the likelihood function of the river segment, and the mass conservation equation is introduced in its calculation process.

[0111] The Markov chain Monte Carlo (MCMC) sampling method is used to sample the posterior distribution. Through numerous iterations, the MCMC chain converges to obtain the posterior distribution of the parameters.

[0112] Step 603: Determine the calibrated hydraulic parameters based on the posterior probability distribution.

[0113] For example, samples of the posterior distribution of hydraulic parameters are obtained from the converged MCMC sampling chain. The mean (or maximum a posteriori probability estimate) of this posterior distribution is taken as the final calibrated hydraulic parameters. These parameters are globally optimal, ensuring that the model satisfies the law of water conservation to the greatest extent possible while interpreting all observation data.

[0114] In this embodiment, the likelihood function is obtained by embedding the mass conservation equation into Bayesian inference and then calibrating the parameters so that the calibration process does not depend on the measured data and the calibrated parameters satisfy the mass conservation equation.

[0115] In an exemplary embodiment, step 601 includes: determining the mass residual of each river segment at each common observation time based on the target observation data and the mass conservation equation, wherein the mass residual is used to characterize the degree of imbalance of water volume in the river segment; aggregating all mass residuals to obtain the overall mass residual; and constructing a likelihood function for Bayesian inference based on the overall mass residual.

[0116] For example, the mass conservation equation is:

[0117]

[0118] In the formula, For traffic The rate of change of x along the river channel Let be the rate of change of the cross-sectional area of ​​the water passage with t. Let be the average lateral inflow or outflow of river segment r at observation time t. This value is positive when water flows from the floodplain into the river channel, and negative otherwise. In this embodiment, it is set to 0. Therefore, based on this mass conservation equation, for a river segment without external water inflow, the difference between upstream inflow and downstream outflow is exactly equal to the change in the river segment's own water storage capacity.

[0119] By mathematically transforming the mass conservation equation (5), we obtain the form shown in equation (6). This transformation can be achieved through the mass residual. It provides a direct way to measure the water conservation status of a river section.

[0120]

[0121] In the formula, and Both are vectors of length 6, representing the discretized differential operator and the flow rate, respectively, and their expressions are shown in equations (7) and (8):

[0122]

[0123] ;

[0124] ;

[0125] ;

[0126] when hour, ;

[0127] when hour, .

[0128] in, Let r be the length of the river segment. This represents the total number of selected river segments.

[0129] For each river segment r and each common observation time t, calculate the mass residual of river segment r at the common observation time t based on the target observation data (water level, slope, width) and the current hydraulic parameter sampling values ​​at that time. Optionally, firstly, based on the initial cross-sectional area of ​​the water passage sampled and the cross-sectional area of ​​the water passage at time t, the change in the cross-sectional area of ​​the water passage is calculated. Then, the change in cross-sectional area at time t. Width value Slope value Substituting the sampled Manning roughness into the Manning formula (3) above, the river flow at time t is calculated. and river flow and the change in cross-sectional area of ​​water passage Substituting into the discretized mass conservation equation (6), we obtain the mass residual of river segment r at time t. Its magnitude reflects the degree of imbalance between the changes in flow rate and water storage at that point.

[0130] Then, the local mass residuals calculated for all river segments at all common observation times are aggregated to obtain the global mass residual, which characterizes the degree of imbalance between flow and storage changes in all river segments. Optionally, a weighted sum of all local mass residuals is calculated to obtain the global mass residual.

[0131] Finally, the overall mass residual is used as the core variable to construct the likelihood function in Bayesian inference. Optionally, the likelihood function is made inversely proportional to the overall mass residual; the smaller the overall mass residual, the greater the likelihood of the currently sampled hydraulic parameter; the larger the overall mass residual, the smaller the likelihood of the currently sampled hydraulic parameter.

[0132] After the parameter calibration is completed, the calibrated hydraulic parameters and the optimized observation data (water level, slope and width values) are substituted into equation (3) to calculate the flow inversion results for each river segment.

[0133] Therefore, in this embodiment, the constraint of water conservation in upstream and downstream river sections is introduced, sampling is performed, and finally the posterior hydraulic parameter distribution considering physical conservation is obtained. By constraining the parameters in the correct direction, the limitations of traditional flow inversion methods that rely on measured data are overcome.

[0134] The following is combined with Figure 7 and Figure 8 The method for inverting river flow in a stationless section based on satellite remote sensing, according to an embodiment of this application, is described through a specific example. For instance... Figure 7 As shown, traffic inversion is performed through the following steps:

[0135] Step 701: Select four consecutive river segments from the prior river dataset as the river segments to be inverted.

[0136] like Figure 8 As shown, four consecutive river segments 801, 802, 803 and 804 from the upstream and downstream of the river were selected from the prior river dataset as river segments to be inverted. There are no tributaries or reservoirs within 5 kilometers upstream and downstream of this area, which is an ideal river segment to be inverted.

[0137] Step 702: Obtain a subset of raster images of the river segment to be inverted from the raster image dataset;

[0138] Step 703: Determine the time series of water level, slope, and width values ​​for four consecutive river segments based on the raster image subsets.

[0139] Step 704: Based on the principle of consistency of river section changes, determine the abnormal water level value and abnormal slope value of each river section, remove the abnormal water level value, and correct the abnormal slope value to obtain the optimized water level value time series and slope value time series.

[0140] Step 705: Correct the abnormal width values ​​in the time series of width values ​​of each river segment, and smooth the corrected width value time series to obtain the optimized width value time series.

[0141] The optimized time series of water level and width values ​​for river section 801 are as follows: Figure 9 As shown, the gray dashed line represents the width value, and the black solid line represents the water level value.

[0142] Step 706: Based on the optimized time series of water level, slope, and width values, construct spatiotemporal two-dimensional observation matrices for water level, slope, and width values, respectively.

[0143] For example, there were 41 observations of river section 801 between November 2023 and March 2025, with 21 simultaneous observations of the four river sections used for parameter calibration, meaning there were 21 common observation times.

[0144] Step 707: Calibrate the hydraulic parameters based on the spatiotemporal two-dimensional observation matrix and the mass conservation equation to obtain the calibrated hydraulic parameters;

[0145] Step 708: For each river segment, substitute the calibrated hydraulic parameters and optimized observation data into the Manning formula to obtain the flow time series of the river segment.

[0146] The flow inversion results for section 801 are as follows: Figure 10 As shown by the solid line in the image, Figure 10 The dashed line represents the measured data, and the solid line represents the inversion result. The measured data is only used for accuracy evaluation and is not used for model parameter calibration. The inversion accuracy was verified based on the measured data, and the Nash-Sutcliffe Efficiency (NSE) reached 0.88, indicating that the embodiments of this application have high inversion accuracy; the Normalized Root Mean Square Error (NRMSE) is only 9%, indicating good inversion performance.

[0147] In summary, compared with existing ground observation and satellite flow inversion methods, the embodiments of this application have the following characteristics and technical advantages: (1) Large spatial observation scale and high degree of automation. Based on new spaceborne radar altimeters such as SWOT satellite, accurate monitoring of river flow on a global scale can be achieved, overcoming the problem of sparse spatial distribution of ground observation stations; (2) No dependence on actual measurement stations. By considering the water conservation relationship between river segments, the model parameter calibration is achieved using the Markov chain Monte Carlo method, which does not rely on any actual measurement station data; (3) High accuracy and strong robustness. By optimizing the preprocessing of water level, slope and width observation data, the interference of outliers and noise on the inversion results is reduced. In addition, the selection strategy of continuous river segments further reduces the impact of observation errors of individual river segments on the overall results.

[0148] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0149] Based on the same inventive concept, this application also provides a satellite remote sensing-based non-station river section flow inversion device for implementing the above-mentioned satellite remote sensing-based non-station river section flow inversion method. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more satellite remote sensing-based non-station river section flow inversion device embodiments provided below can be found in the limitations of the method above, and will not be repeated here.

[0150] In one exemplary embodiment, such as Figure 11As shown, a stationless river segment flow inversion device based on satellite remote sensing is provided, comprising: a first determining module 1101, an optimization module 1102, a second determining module 1103, a calibration module 1104, and a third determining module 1105, wherein: the first determining module 1101 is used to determine observation datasets for at least three consecutive river segments based on raster image datasets observed by satellite, the observation datasets including observation data and the observation times corresponding to the observation data; the optimization module 1102 is used to optimize the observation data to obtain optimized observation data; the second determining module 1103 is used to determine the common observation time for each river segment and the target observation data for each river segment at the common observation time based on the optimized observation data; the calibration module 1104 is used to calibrate the hydraulic parameters based on the target observation data and the mass conservation equation for each river segment to obtain calibrated hydraulic parameters; the third determining module 1105 is used to determine the flow time series of each river segment based on the calibrated hydraulic parameters and the optimized observation data.

[0151] In one embodiment, the observation data includes the river's water level and slope values. The optimization module 1102 includes a first determining unit, a eliminating unit, and a first correcting unit, wherein: the first determining unit is used to determine the abnormal water level and abnormal slope values ​​of each river segment based on the principle of consistency of river segment changes; wherein the principle of consistency of river segment changes means that all river segments have the same trend of change at the common observation time; the eliminating unit is used to eliminate the abnormal water level values ​​to obtain the optimized water level values ​​of the river segments; the first correcting unit is used to correct the abnormal slope values ​​to obtain the optimized slope values ​​of the river segments.

[0152] In one embodiment, the observation data includes the width value of the river, and the optimization module 1102 further includes a second correction unit and a smoothing unit, wherein: the second correction unit is used to correct the abnormal width values ​​in the width value time series of each river segment to obtain a corrected width value time series; the smoothing unit is used to smooth the corrected width value time series to obtain an optimized width value time series.

[0153] In one embodiment, the observation data includes the width value of the river. The first determining module 1101 is specifically used to: extract the channel mask of each river segment from each raster image in the raster image dataset; within the channel mask, traverse each sub-river segment using a sliding window with a preset river length, and determine the target sub-river segment according to the stability evaluation index of each sub-river segment; and use the average river width of each target sub-river segment as the width value of the river segment to obtain the time series of the width value of the river segment.

[0154] In one embodiment, the calibration module 1104 includes a construction unit, a second determination unit, and a third determination unit, wherein: the construction unit is used to construct a Bayesian inference likelihood function based on the target observation data and the mass conservation equation; the second determination unit is used to determine the posterior probability distribution of the hydraulic parameters based on the likelihood function and the prior probability distribution of the hydraulic parameters; and the third determination unit is used to determine the calibrated hydraulic parameters based on the posterior probability distribution.

[0155] In one embodiment, the construction unit is specifically used to: determine the mass residual of each river segment at each common observation time based on the target observation data and the mass conservation equation, wherein the mass residual is used to characterize the degree of imbalance of water volume in the river segment; aggregate all mass residuals to obtain the overall mass residual; and construct the likelihood function of Bayesian inference based on the overall mass residual.

[0156] In one embodiment, the first determining unit is specifically used to: identify suspicious water level values ​​among the water level values ​​in the river segment based on the water level changes in the river segment; determine the water level change threshold corresponding to each suspicious water level value based on the water level values ​​in other river segments; and determine the suspicious water level value as an abnormal water level value if the difference between the suspicious water level value and the water level value at an adjacent observation time in the river segment exceeds the water level change threshold.

[0157] In one embodiment, the first determining unit is further configured to: identify suspicious slope values ​​among the slope values ​​of the river segment based on the slope changes in the river segment; determine the slope mutation threshold corresponding to each suspicious slope value based on the slope values ​​of other river segments; and determine the suspicious slope value as an abnormal slope value if the ratio between the suspicious slope value and the median slope of the river segment exceeds the slope mutation threshold.

[0158] The modules in the aforementioned satellite remote sensing-based stationless river flow inversion device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0159] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 12As shown. The computer device includes a processor, memory, input / output interface (I / O), and communication interface. The processor, memory, and I / O interface are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores inversion data. The I / O interface allows the processor to exchange information with external devices. The communication interface allows communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for inverting river flow without monitoring stations based on satellite remote sensing. Those skilled in the art will understand that... Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0160] In one exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the satellite remote sensing-based method for inverting river flow without measuring stations according to the embodiments of this application.

[0161] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements a method for inverting river flow in a stationless section based on satellite remote sensing.

[0162] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements a method for inverting river flow in a stationless section based on satellite remote sensing.

[0163] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this application. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this application. For those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the protection scope of this application.

Claims

1. A method for inverting river flow in a stationless section based on satellite remote sensing, characterized in that, The method includes: Based on the raster image datasets from satellite observations, at least three consecutive river segments are determined, and each observation dataset includes observation data and the observation time corresponding to the observation data. The observation data is optimized to obtain the optimized observation data; Based on the optimized observation data, the common observation time for each river segment and the target observation data for each river segment at the common observation time are determined. Based on the target observation data and mass conservation equations for each river section, the hydraulic parameters are calibrated to obtain the calibrated hydraulic parameters. For each river segment, the flow time series of the river segment is determined based on the calibrated hydraulic parameters and the optimized observation data.

2. The method according to claim 1, characterized in that, The observation data includes river water level and slope values. The optimization processing of the observation data to obtain optimized observation data includes: The abnormal water level and abnormal slope values ​​for each river segment are determined based on the principle of consistency of river segment changes; wherein, the principle of consistency of river segment changes means that the change trends of all river segments are the same at the common observation time; The abnormal water level values ​​are removed to obtain the optimized water level values ​​for the river section. The abnormal slope value is corrected to obtain the optimized slope value of the river section.

3. The method according to claim 1, characterized in that, The observation data includes the width value of the river, and the optimization processing of the observation data to obtain optimized observation data includes: The abnormal width values ​​in the time series of width values ​​of each river segment are corrected to obtain the corrected width value time series; The corrected width value time series is smoothed to obtain the optimized width value time series.

4. The method according to claim 1, characterized in that, The observation data includes the width value of the river, and the observation datasets determined based on the raster image datasets from satellite observations for at least three consecutive river segments include: Extract the channel mask for each river segment from each raster image in the raster image dataset; Within the riverbed membrane, a sliding window with a preset river length is used to traverse each sub-river segment, and the target sub-river segment is determined based on the stability evaluation index of each sub-river segment. The average width of each target sub-river segment is used as the width value of the river segment to obtain a time series of the width values ​​of the river segment.

5. The method according to any one of claims 1-4, characterized in that, The hydraulic parameters are calibrated based on the target observation data and mass conservation equations for each river segment, resulting in calibrated hydraulic parameters, including: Construct a Bayesian likelihood function based on the target observation data and the mass conservation equation; Based on the likelihood function and the prior probability distribution of the hydraulic parameters, determine the posterior probability distribution of the hydraulic parameters; The calibrated hydraulic parameters are determined based on the posterior probability distribution.

6. The method according to claim 5, characterized in that, The step of constructing the Bayesian likelihood function based on the target observation data and the mass conservation equation includes: Based on the target observation data and the mass conservation equation, the mass residual of each river segment at each common observation time is determined, and the mass residual is used to characterize the degree of imbalance of water volume in the river segment; Aggregate all quality residuals to obtain the overall quality residual; The likelihood function for Bayesian inference is constructed based on the overall quality residual.

7. The method according to claim 2, characterized in that, Abnormal water level values ​​for each river segment are determined based on the principle of consistency in river segment changes, including: Based on the water level changes in the river section, suspicious water level values ​​among the various water level values ​​in the river section are identified; The threshold for each suspected water level change is determined based on the water level values ​​of other river sections; If the difference between the suspected water level value and the water level value at an adjacent observation time in the river section exceeds the water level change threshold, the suspected water level value is determined as the abnormal water level value.

8. The method according to claim 2, characterized in that, The abnormal slope value for each of the river segments is determined based on the principle of consistency in river segment changes, including: Based on the slope changes in the river section, suspicious slope values ​​are identified among the various slope values ​​in the river section; Determine the slope abruptness threshold corresponding to each of the suspected slope values ​​based on the slope values ​​of other river sections; If the ratio between the suspected slope value and the median slope of the river segment exceeds the slope abrupt change threshold, the suspected slope value is determined to be the abnormal slope value.

9. A station-free river section flow inversion device based on satellite remote sensing, characterized in that, The device includes: The first determining module is used to determine observation datasets for at least three consecutive river segments based on the raster image datasets observed by satellite. The observation datasets include observation data and the observation time corresponding to the observation data. The optimization module is used to optimize the observation data to obtain optimized observation data; The second determining module is used to determine the common observation time for each river segment and the target observation data for each river segment at the common observation time based on the optimized observation data. The calibration module is used to calibrate the hydraulic parameters based on the target observation data and mass conservation equation for each river section, and obtain the calibrated hydraulic parameters. The third determining module is used to determine the flow time series of each river segment based on the calibrated hydraulic parameters and the optimized observation data.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images