Method and device for remotely estimating water storage capacity of seasonal rivers and lakes

Abstract

This invention provides a remote sensing estimation method and apparatus for seasonal river and lake water storage, belonging to the field of remote sensing hydrological processing technology. The method includes: determining the central axis of the river and lake in the target area based on water body mask data; identifying dry sections and determining flow interruption points; constructing a river and lake network topology map; determining hydraulic distances based on the river and lake network topology map; obtaining an initial water surface elevation field by performing elevation propagation on elevation observation data; obtaining a target three-dimensional water surface elevation field by applying water level constraints; constructing a water storage mapping model based on the target three-dimensional water surface elevation field and water body mask data; and using the water storage mapping model to determine the water storage in the target area. This invention constructs a river and lake network topology map using flow interruption points to determine hydraulic distances that reflect the true hydraulic connectivity characteristics. It uses these hydraulic distances to drive elevation propagation to obtain the target three-dimensional water surface elevation field, and then constructs a water storage mapping model to estimate the water storage, thereby solving the problem of elevation field distortion and achieving accurate estimation of seasonal river and lake water storage.

Description

Technical Field

[0001] This invention relates to the field of remote sensing hydrological processing technology, and in particular to a method and apparatus for remote sensing estimation of seasonal river and lake water storage. Background Technology

[0002] Seasonal rivers and lakes are widely distributed in arid and semi-arid regions worldwide, exhibiting unique hydrological characteristics such as flow interruption during the dry season, flooding during the wet season, and large annual water level fluctuations. In areas lacking surface hydrological station networks, accurate monitoring of their water storage is a crucial requirement for water resource management and flood early warning.

[0003] Currently, in areas lacking ground-based hydrological observation data, satellite elevation observation data is typically used for elevation propagation to reconstruct the water surface elevation field, and then a relational model is established to estimate water storage.

[0004] However, this method relies solely on geometric spatial distance to drive elevation propagation, neglecting the hydraulic connectivity of rivers. This results in severe distortion of the elevation field reconstruction, leading to significant errors in the final calculated seasonal river and lake water storage, which cannot meet the requirements for accurate monitoring of water storage. Summary of the Invention

[0005] This invention provides a remote sensing estimation method and device for seasonal river and lake water storage, which solves the defects of existing technologies that rely solely on geometric spatial distance to drive elevation propagation while ignoring the hydraulic connectivity of rivers, resulting in severe distortion of elevation field reconstruction and large errors in water storage calculation, and achieves accurate monitoring of seasonal river and lake water storage. The core innovations of this invention include three aspects: (1) proposing a hydraulic distance calculation method based on flow interruption sensing, setting the distance between nodes crossing the flow interruption point to infinity, fundamentally solving the physical meaningless elevation error caused by the forced interpolation across dry river sections in existing methods; (2) constructing an elevation field optimization model with joint constraints of longitudinal water surface gradient and cross-sectional water level consistency, realizing three-dimensional water surface reconstruction that conforms to the real hydrodynamic laws; (3) establishing a segmented nonlinear water storage mapping model adapted to the three stages of dry flow interruption, main channel water flow, and floodplain, and introducing a floodplain water storage coefficient correction term based on vegetation canopy height and micro-topographic undulation, significantly improving the accuracy of water storage calculation throughout the year.

[0006] This invention provides a remote sensing method for estimating the seasonal water storage capacity of rivers and lakes, comprising the following steps: Acquire water mask data, elevation observation data, and digital elevation model of the target area; Based on the water body mask data, the central axis of the river and lake in the target area is determined, and the waterless section is identified based on the central axis of the river and lake to determine the flow interruption point. A river-lake network topology map is constructed based on the central axis of the river and lake, the flow interruption point, and the elevation observation data; the hydraulic distance is determined based on the river-lake network topology map. Based on the hydraulic distance, the elevation observation data is propagated to obtain an initial water surface elevation field; based on the digital elevation model, the initial water surface elevation field is constrained by water level to obtain a target three-dimensional water surface elevation field. Based on the target three-dimensional water surface elevation field and the water body mask data, a water storage mapping model is constructed, and the water storage of the target area is estimated using the water storage mapping model.

[0007] According to the remote sensing estimation method for seasonal river and lake water storage provided by the present invention, the step of determining hydraulic distance based on the river and lake network topology map includes: Obtain the flow direction weights and tortuosity coefficients between adjacent nodes in the river and lake network topology diagram; The hydraulic distance between adjacent nodes is determined based on the flow direction weight, the curvature coefficient, and the length of the river / lake central axis between adjacent nodes. The hydraulic distance between adjacent nodes that cross the flow interruption point is set to infinity.

[0008] According to the remote sensing estimation method for seasonal river and lake water storage provided by the present invention, the step of performing elevation propagation on the elevation observation data based on the hydraulic distance to obtain an initial water surface elevation field includes: Obtain the minimum hydraulic distance path from the node to be interpolated in the river and lake network topology map to each of the elevation observation points, wherein the node to be interpolated is a node in the river and lake network topology map that lacks elevation observation data; If the minimum hydraulic distance path includes the flow interruption point, then the elevation observation point closest to the downstream of the node to be interpolated is determined as the propagation starting point, and the hydraulic distance between the propagation starting point and the node to be interpolated is obtained. Elevation propagation is performed based on the hydraulic distance between the propagation starting point and the node to be interpolated to obtain the initial water surface elevation field.

[0009] According to the present invention, a remote sensing estimation method for seasonal river and lake water storage capacity is provided, wherein the step of applying water level constraints to the initial water surface elevation field based on the digital elevation model to obtain the target three-dimensional water surface elevation field includes: Based on the digital elevation model, the longitudinal water surface gradient along the central axis of the river and lake is determined. The longitudinal water level gradient constraint term is determined based on the aforementioned longitudinal water level gradient. Based on the longitudinal water surface gradient constraint, the initial water surface elevation field, and the cross-sectional water level consistency constraint, a joint optimization function is constructed. Solving the joint optimization function yields the target three-dimensional water surface elevation field.

[0010] According to the remote sensing estimation method for seasonal river and lake water storage provided by the present invention, the joint optimization function is obtained based on the following mathematical model: ; in, For the joint optimization function, For the first i The weight of each pixel's elevation observation value The first in the initial water surface elevation field i The elevation observation value of each pixel For the target three-dimensional water surface elevation field to be solved, the first... i The elevation value of each pixel. , and These are the weighting coefficients. For the longitudinal water surface gradient constraint term, This refers to the cross-sectional water level consistency constraint term. This is a spatial smoothing constraint term.

[0011] According to the present invention, a remote sensing estimation method for seasonal river and lake water storage capacity is provided, wherein the water storage capacity mapping model is constructed based on the target three-dimensional water surface elevation field and the water body mask data, including: Based on the target three-dimensional water surface elevation field and the water body mask data, the water level area change rate curve is determined; Obtain the initial water level of the flow interruption, and determine the water level corresponding to the local maximum value of the water level area change rate curve as the initial water level of the floodplain; Based on the initial water level of the flow interruption and the initial water level of the floodplain, the water storage mapping model is constructed. The water storage mapping model includes a sub-model of the flow interruption stage, a sub-model of the main channel water passage stage, and a sub-model of the floodplain stage.

[0012] According to the present invention, a remote sensing method for estimating the seasonal water storage capacity of rivers and lakes includes estimating the water storage capacity of the target area using the water storage mapping model, comprising: If the water level in the target area is lower than the initial water level of the flow interruption, the water storage capacity of the target area is estimated using the flow interruption stage sub-model. If the water level in the target area is greater than or equal to the initial water level of the flow interruption and less than the initial water level of the floodplain, then the water storage capacity of the target area is estimated using the main channel water passage stage sub-model. If the water level in the target area is greater than or equal to the initial water level of the floodplain, the water storage capacity of the target area is estimated using the floodplain stage sub-model. The floodplain stage sub-model includes a floodplain water storage coefficient, which is determined based on the vegetation canopy height and micro-topographic undulation of the target area.

[0013] According to the remote sensing estimation method for seasonal river and lake water storage provided by the present invention, the step of acquiring elevation observation data of the target area includes: Acquire initial elevation observation data of the target area and multispectral remote sensing images that match the acquisition time of the initial elevation observation data; Based on the near-infrared band of the multispectral remote sensing image, the near-infrared gradient characteristics of the location of the initial elevation observation data are determined. Based on the near-infrared gradient features, the hydraulic model deviation features, boundary distance features, and water body proportion features of the initial elevation observation data, the comprehensive anomaly score of the initial elevation observation data is determined. The initial elevation observation data is then processed by removing anomalies based on the comprehensive anomaly score to obtain the elevation observation data.

[0014] According to the remote sensing estimation method for seasonal river and lake water storage provided by the present invention, the step of estimating the water storage of the target area using the water storage mapping model further includes: If water mask data of the target area at the time step to be measured is obtained, the water mask data is input into the water storage mapping model to estimate the water storage at the time step to be measured. If the water mask data of the target area at the time step to be measured is not obtained, then the precipitation, water surface evaporation and outflow of the target area are obtained; Based on the precipitation, the water surface evaporation, the outflow, and the water storage at the previous time step, a water balance equation is constructed, and the water storage at the time step to be measured is estimated by numerically integrating the water balance equation.

[0015] The present invention also provides a device for estimating the seasonal water storage capacity of rivers and lakes, comprising the following modules: The data acquisition module is used to acquire water mask data, elevation observation data, and digital elevation models of the target area; The flow interruption identification module is used to determine the central axis of the river and lake in the target area based on the water body mask data, identify waterless sections based on the central axis of the river and lake, and determine the flow interruption point. The topology construction module is used to construct a river and lake network topology map based on the central axis of the river and lake, the flow interruption point, and the elevation observation data; and to determine the hydraulic distance based on the river and lake network topology map. The elevation reconstruction module is used to perform elevation propagation on the elevation observation data based on the hydraulic distance to obtain an initial water surface elevation field; and to apply water level constraints to the initial water surface elevation field based on the digital elevation model to obtain a target three-dimensional water surface elevation field. The water storage estimation module is used to construct a water storage mapping model based on the target three-dimensional water surface elevation field and the water body mask data, and to estimate the water storage of the target area using the water storage mapping model.

[0016] The present invention provides a method and apparatus for remote sensing estimation of seasonal river and lake water storage. It constructs a river and lake network topology map by using flow interruption points to determine the hydraulic distance that reflects the true hydraulic connectivity characteristics. It uses the hydraulic distance to drive elevation propagation to obtain the target three-dimensional water surface elevation field, and then constructs a water storage mapping model to estimate the water storage. This can solve the problem of elevation field distortion and achieve accurate monitoring of water storage. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the remote sensing estimation method for seasonal river and lake water storage provided by the present invention.

[0019] Figure 2 This is a schematic diagram of the process for determining the initial water surface elevation field provided by the present invention.

[0020] Figure 3 This is a flowchart illustrating the process of determining the three-dimensional water surface elevation field of a target, provided by the present invention.

[0021] Figure 4 This is a schematic diagram of the process for constructing a water storage mapping model provided by the present invention.

[0022] Figure 5 This is a schematic diagram of the process for obtaining elevation observation data of a target area provided by the present invention.

[0023] Figure 6 This is a schematic diagram of the structure of the remote sensing estimation device for seasonal river and lake water storage provided by the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] It should be noted that, in the description of this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0026] To facilitate a full understanding of the technical solution of this application, the following content is hereby introduced: Seasonal rivers and lakes constitute a significant portion of global river and lake networks and are widely distributed across arid and semi-arid regions worldwide. Water resource management, ecological conservation, and flood warning for these rivers heavily rely on accurate monitoring of water storage. However, due to their unique hydrological characteristics—namely, flow interruption during the dry season, flooding during the wet season, and large annual water level fluctuations—accurate monitoring of water storage faces the following core technological challenges: (1) Lack of ground observation network. In areas lacking data, there is a lack of long-term hydrological station network, making it impossible to obtain continuous water level process data, and the traditional water storage estimation method based on hydrological stations is completely ineffective.

[0027] (2) Single remote sensing methods have fundamental limitations. Optical remote sensing can extract the high-precision horizontal range of water bodies, but it cannot directly obtain water surface elevation information; traditional altimetry satellites have long revisit cycles and sparse coverage along the orbit, resulting in a serious lack of temporal resolution over seasonal rivers and lakes; although Synthetic Aperture Radar (SAR) has all-weather observation capabilities, the inversion of water surface elevation from SAR backscattering requires complex physical models and a large amount of ground verification data.

[0028] (3) Modeling the relationship between water level, area and volume is difficult. Seasonal rivers and lakes have obvious dry season flow interruption and the transition from the main channel to the floodplain during the wet season. The three variables have strong nonlinear piecewise characteristics. The traditional single power function relationship model assumes that the water body shape changes continuously, which cannot characterize the flow interruption stage and the sudden change of the floodplain, resulting in a large error in the calculation of water storage.

[0029] For satellite altimetry data, such as surface water and ocean topography (SWOT), existing technologies have the following limitations in performing spatial elevation propagation, outlier removal, and water storage modeling: First, existing elevation interpolation methods neglect hydraulic connectivity. Representative methods such as spatiotemporal kriging use interpolation weights based on a variogram constructed from Euclidean geometric spatiotemporal distances, failing to explicitly characterize the hydraulic connectivity of rivers. These methods cannot identify flow interruption points; when waterless sections exist, the river segments on either side of the interruption point are hydraulically disconnected, yet existing methods still interpolate based on geometric distances, producing physically meaningless elevation results. Furthermore, they lack directional constraints, ignoring the fundamental physical law of river water flowing downhill; and they are also unsuitable for the dynamic changes from flow interruption to floodplain. Methods incorporating hydrodynamic models rely on complete river and lake topographic data, making them unusable in areas with limited data, and are computationally extremely expensive, also failing to address seasonal flow interruptions.

[0030] Second, existing methods for removing water body elevation anomalies have limitations. Current elevation quality control methods typically rely on gross error removal based on product quality indicators or statistical thresholds. Their common drawback is that they only consider the statistical distribution of elevation observation data, failing to distinguish between reasonable elevation changes caused by actual topographic relief and systematic elevation deviations caused by mixed pixels at the land-water boundary. Because the quality assessment of multi-source remote sensing information and radar altimetry data is not integrated, the false positive rate for anomaly identification is high, easily misclassifying actual topographic relief as anomalies.

[0031] Third, existing water storage mapping modeling methods are not applicable to seasonal rivers and lakes. Existing modeling methods primarily target perennial lakes and reservoirs, employing a single power function fit, which cannot cover the dry season and flow interruption phases of seasonal rivers and lakes. None of the existing methods have established piecewise nonlinear models adaptable to the entire process of dry season flow interruption, low-water main channel, and high-water floodplain, and they do not quantitatively consider the impact of vegetation canopy interception and micro-topography on the floodplain water storage coefficient, resulting in significant errors in the calculation of annual water storage.

[0032] In summary, existing technologies have gaps in multiple dimensions. To address these deficiencies, this application proposes a remote sensing method for estimating seasonal river and lake water storage, primarily comprising: elevation propagation based on flow interruption sensing and hydraulic distance; cross-modal elevation anomaly identification through multi-source remote sensing collaboration; elevation field reconstruction with water level constraints based on a digital elevation model; and a piecewise nonlinear water storage mapping model adaptable to the entire process from dry sections to floodplains, thereby achieving accurate monitoring of seasonal river and lake water storage.

[0033] The following is combined Figures 1-6 This invention describes the remote sensing estimation method and apparatus for seasonal river and lake water storage provided by the present invention.

[0034] Figure 1 This is a flowchart illustrating the remote sensing estimation method for seasonal river and lake water storage provided by the present invention, as shown below. Figure 1 As shown, the execution subject of the seasonal river and lake water storage remote sensing estimation method provided by the present invention can be a server, a cloud computing platform, or a computer capable of executing the method of the present invention, etc. Unless otherwise specified, the following embodiments will be described using a server as an example.

[0035] As an optional embodiment, this remote sensing estimation method for seasonal river and lake water storage mainly includes, but is not limited to, the following steps: Step 110: Obtain water mask data, elevation observation data, and digital elevation model for the target area.

[0036] The target area refers to the spatial range of seasonal rivers and lakes where precise monitoring of water storage is required. For example, the target area could be a seasonal river and lake basin in arid or semi-arid regions where long-term ground hydrological station data is lacking.

[0037] Water mask data refers to image data used to characterize the two-dimensional spatial distribution range and boundary features of water bodies within a target area. For example, water mask data can be a binary matrix that identifies whether a pixel belongs to a continuous water body range.

[0038] Water mask data can be obtained by calculating water indexes and thresholding multispectral remote sensing images. For example, water mask data can be obtained by acquiring Sentinel-2 multispectral remote sensing images and extracting them using a dual-index fusion decision tree constructed by combining normalized differential water index and automatic water extraction index.

[0039] An adaptive thresholding method is used to determine the optimal segmentation threshold of the index. When the normalized difference water index and the automatic water extraction index of a pixel are both greater than their respective optimal segmentation thresholds, the pixel is extracted as a water pixel and morphological opening and closing operations are performed to ensure the continuity and integrity of the water boundary.

[0040] Elevation observation data refers to a dataset containing absolute water surface elevation information obtained through satellite radar altimetry payloads. For example, elevation observation data can be a SWOT satellite interferometric radar two-dimensional water surface elevation observation product that includes fields such as water surface elevation value and elevation uncertainty.

[0041] Elevation observation data can be downloaded and preprocessed by accessing publicly available satellite data distribution platforms. For example, SWOT elevation products for a specific revisit period can be obtained through relevant data platforms and then filtered for quality based on quality marker bits and pixel water body proportion thresholds.

[0042] A digital elevation model (DEM) refers to a digital elevation dataset that represents the topographic features of a target area, such as the surface or riverbed. For example, a DEM can be TerraSAR-X add-on for Digital Elevation Measurement (TanDEM-X) or Advanced Land Observing Satellite (ALOS) global digital elevation model data.

[0043] Digital elevation models can be obtained by acquiring publicly available global topographic mapping data and unifying the elevation datum. For example, the downloaded raw digital elevation model data can be pre-converted to a geoid datum that is consistent with the elevation observation data, so as to provide a priori water level constraints for river and lake morphology.

[0044] Step 120: Determine the central axis of the river and lake in the target area based on the water mask data, identify waterless sections based on the central axis of the river and lake, and determine the flow interruption point.

[0045] The central axis of a river or lake refers to the core skeleton curve that characterizes the spatial distribution features and main channel orientation of a river or lake. For example, the central axis of a river or lake can be a smooth, continuous line with a single-pixel width that reflects the topological structure of the geometric center of the river in the target area.

[0046] The central axis of a river or lake can be determined by extracting the skeleton of the acquired two-dimensional water body mask and smoothing it mathematically. For example, the water body mask extracted from multispectral remote sensing images such as Sentinel-2 is processed to obtain an initial single-pixel wide skeleton. Then, noise pruning is performed on the skeleton to remove short burr branches. Subsequently, cubic B-spline interpolation is used for smoothing to determine the final central axis of the river or lake.

[0047] Waterless sections can be identified by scanning along the direction of water flow and combining the spatial continuity of the water body mask. For example, the state of the water body mask can be scanned segment by segment along the extracted smooth river and lake central axis. If a continuous waterless state is detected along the river and lake central axis and the length is greater than or equal to a set threshold (such as 500 meters), the interval is identified as a seasonally dried-up waterless section.

[0048] A flow interruption point refers to the node at the beginning and end of a dry river segment where the hydraulic connectivity of a river is physically interrupted. For example, flow interruption points can be topological nodes set at the beginning and end of identified continuous dry segments. These nodes will be used to characterize the interruption of the hydraulic propagation path in subsequent elevation propagation calculations.

[0049] Step 130: Construct a river and lake network topology map based on the central axis of the river and lake, the flow interruption point, and elevation observation data; determine the hydraulic distance based on the river and lake network topology map.

[0050] A river and lake network topology map refers to graph-structured data that characterizes the key spatial locations of various river systems and their flow directions and connectivity. For example, it can be constructed by taking the intersection and confluence points of the central axis of the river and lake, the effective observation points where the elevation observation data is located, and the identified flow interruption points as the node set of the topology map, and taking the river segments connecting adjacent nodes as the directed edge set. The direction of the directed edges is determined by the aspect field of the digital elevation model to keep it consistent with the actual flow direction.

[0051] Hydraulic distance refers to the non-Euclidean spatial distance measured along the central axis of a river or lake, taking into account the hydraulic connectivity and river morphology. For example, hydraulic distance can be a composite distance metric that combines the actual centerline length of the river section, the weight of the influence of the flow direction on the reliability of information transmission, and the correction coefficient for the curvature of the river or lake. When the path crosses a dry section, the distance is infinite.

[0052] Step 140: Based on hydraulic distance, perform elevation propagation on elevation observation data to obtain an initial water surface elevation field; based on the digital elevation model, apply water level constraints to the initial water surface elevation field to obtain the target three-dimensional water surface elevation field.

[0053] Elevation propagation refers to the spatial expansion process of transmitting elevation values ​​from discrete elevation observation points along a water flow path to surrounding pixels without observation data. For example, in practical implementation, an improved Dijkstra algorithm can be used to transmit elevation information from valid elevation observation points along the path of minimum hydraulic distance to the pixels to be interpolated.

[0054] Specifically, the improved Dijkstra algorithm is based on a pre-constructed river network topology map and uses hydraulic distance as the search weight. Its improvement lies in: real-time retrieval of flow interruption point markers during the search process; if the path to be interpolated crosses a flow interruption point, the propagation path is forced to fail, triggering a recalculation mechanism (i.e., interrupting propagation and recalculating from the nearest downstream water-bearing observation point). Simultaneously, a strict physical constraint is applied during propagation: the downstream pixel elevation must not exceed the upstream pixel elevation minus the product of the riverbed gradient and hydraulic distance. This constraint is then used to perform Bayesian weighted fusion of the elevation results reached from multiple reachable observation points, ultimately achieving accurate reconstruction of the elevation field.

[0055] The initial water surface elevation field refers to the continuous water surface elevation distribution data that initially forms after hydraulic distance propagation, covering the entire connected water body plane. For example, the initial water surface elevation field can be a gridded elevation surface obtained by weighted fusion of multi-source elevation estimation, which has not yet undergone joint physical optimization of river and lake cross sections.

[0056] Water level constraints refer to the process of spatial morphological correction and joint optimization of preliminary elevation results by utilizing the prior physical laws and geometric characteristics of river and lake topography. For example, the longitudinal riverbed gradient of each river and lake segment can be extracted based on the digital elevation model to establish a longitudinal water surface gradient model, and the standard deviation threshold of the water body pixel elevation consistency on the same cross section can be combined to establish a cross section water level consistency constraint. A joint optimization objective function can be constructed and solved iteratively using the conjugate gradient method.

[0057] The target three-dimensional water surface elevation field refers to the continuous three-dimensional water surface morphology reconstruction result obtained after joint optimization with physical constraints, which conforms to the laws of natural water flow dynamics and eliminates unreasonable local fluctuations. For example, the target three-dimensional water surface elevation field can be a high-precision three-dimensional water surface model with centrifugal force superelevation correction and extremely low cross-sectional elevation standard deviation in sharp bends or meandering river sections.

[0058] Step 150: Based on the target three-dimensional water surface elevation field and water body mask data, construct a water storage mapping model, and use the water storage mapping model to determine the water storage of the target area.

[0059] A water storage mapping model refers to a mathematical model that establishes the physical relationship between water surface elevation, water area, and water volume at a specific stage of a hydrological process. For example, a water storage mapping model can be a three-stage piecewise nonlinear water level-area-volume relationship function that includes flood storage correction coefficients and is established for the dry season, the low-water main channel flow period, and the high-water flood period, based on water level observation data extracted from the target three-dimensional water surface elevation field and area data extracted from water body mask data.

[0060] The water storage capacity of the target area refers to the total volume of three-dimensional water bodies actually present in seasonal rivers and lakes at a specific observation time or within a continuous period. For example, the water storage capacity of the target area can be obtained by inputting the corresponding water level and area data into the water storage mapping model for nonlinear least squares parameter calibration and solution calculation, and then jointly reconstructing the high time resolution dynamic water volume by combining the numerical integration of the water balance equation and Kalman filter state update mechanisms.

[0061] The present invention provides a remote sensing estimation method for seasonal river and lake water storage. It constructs a river and lake network topology map by using flow interruption points to determine the hydraulic distance that reflects the true hydraulic connectivity characteristics. The hydraulic distance is used to drive elevation propagation to obtain the target three-dimensional water surface elevation field. Then, a water storage mapping model is constructed to estimate the water storage, thereby solving the problem of elevation field distortion and realizing accurate monitoring of water storage.

[0062] In another embodiment of the present invention, the hydraulic distance is determined based on the river and lake network topology map, including: obtaining the flow direction weight and curvature coefficient between adjacent nodes in the river and lake network topology map; determining the hydraulic distance between adjacent nodes based on the flow direction weight, curvature coefficient and the length of the river and lake central axis between adjacent nodes; wherein the hydraulic distance between adjacent nodes that cross the flow interruption point is set to infinity.

[0063] Flow direction weight refers to a penalty or adjustment coefficient that reflects the impact of water flow direction on the reliability of spatial propagation of elevation information. For example, flow direction weight can be a numerical ratio that distinguishes between downstream and upstream directions. Since the elevation change downstream is dominated by gravity and the physical laws are clear, its value can be set to a smaller baseline value. However, the propagation upstream is affected by the backwater effect and has greater uncertainty, so its value can be set to a penalty value greater than the baseline value.

[0064] The flow direction weight can be obtained by judging the upstream and downstream relative relationship of the water flow direction of adjacent nodes in the river and lake network topology. For example, when the node to be propagated is downstream of the effective observation node, i.e. downstream, the flow direction weight can be obtained as 1.0. When the node to be propagated is upstream of the effective observation node, i.e. upstream, the flow direction weight can be obtained as 1.5 in order to apply uncertainty penalty.

[0065] The tortuosity coefficient refers to a proportional parameter that characterizes the actual meandering degree of a river or lake and is used to correct the length of the physical propagation path between two points. For example, the tortuosity coefficient can be a dimensionless value greater than or equal to 1, reflecting the degree to which the actual physical path of water flow in a meandering river section is longer than the straight line connecting the two points in space.

[0066] The curvature coefficient can be extracted by calculating the quotient of the actual geometric length of a specific river segment and the straight-line distance between its two ends. For example, the curvature correction value that accurately reflects the true hydraulic path length can be obtained by measuring the actual curve length of the corresponding river and lake central axis between adjacent nodes in the topology map and dividing the actual curve length by the straight-line distance between the two topology nodes.

[0067] Specifically, determining the hydraulic distance between adjacent nodes requires the comprehensive superposition and accumulation of various spatial parameters reflecting the physical characteristics of the river. For example, for any two topological nodes with a connection relationship, all directed edges on the shortest directed path connecting the two points can be extracted. The actual length of each directed edge measured along the central axis of the river or lake can be multiplied by its corresponding flow direction weight and curvature coefficient. Finally, the weighted product of all edges can be summed to obtain the hydraulic distance between the two nodes.

[0068] Considering that seasonally dried-up river sections can fundamentally disrupt the physical continuity of water flow, causing the water bodies on both sides of the dried-up river section to no longer possess true hydraulic connectivity, this invention forcibly sets the hydraulic distance between adjacent nodes crossing the point of flow interruption to positive infinity. This allows the algorithm to automatically interrupt the distance-based elevation information propagation process when it encounters a waterless section, effectively avoiding the physically meaningless elevation errors caused by forcibly crossing the point of flow interruption for spatial geometric interpolation in existing technologies, and ensuring the physical rationality of elevation reconstruction.

[0069] The remote sensing estimation method for seasonal river and lake water storage provided by this invention accurately quantifies the hydraulic distance between nodes by comprehensively incorporating the influence of water flow direction, the actual meandering degree of rivers and lakes, and the blocking effect of flow interruption points on hydraulic connectivity. This ensures that the spatial propagation of elevation information strictly follows the gravity-driven and connectivity laws of natural water flow, fundamentally eliminating the physically meaningless errors caused by the forced mathematical interpolation across dry river sections in existing technologies. It also fully corrects the path measurement deviation of meandering river sections, significantly improving the physical authenticity and interpolation accuracy of local elevation reconstruction in complex morphologies and flow interruption sections.

[0070] Figure 2 This is a flowchart illustrating the process of determining the initial water surface elevation field provided by the present invention, as shown below. Figure 2 As shown, as another optional embodiment provided by the present invention, elevation propagation is performed on elevation observation data based on hydraulic distance to obtain an initial water surface elevation field, including but not limited to the following steps: Step 210: Obtain the minimum hydraulic distance path from the node to be interpolated in the river and lake network topology map to each elevation observation point. The node to be interpolated is the node in the river and lake network topology map that lacks elevation observation data.

[0071] It should be noted that the nodes to be interpolated are mainly distributed in the blank areas of water bodies not covered by the satellite altimetry trajectory, or in the positions of invalid observation pixels that were removed due to previous quality control. In order to reconstruct the three-dimensional water surface in these areas, it is necessary to find the physically optimal elevation information transmission channel.

[0072] The minimum hydraulic distance path refers to the connecting route with the minimum cumulative weighted hydraulic distance along the direction of water flow from the node to be interpolated to the node with valid elevation observation data in the river and lake network topology map. For example, it can be obtained by constructing a minimum priority queue with all quality-controlled valid elevation observation points as the initial set, and using an improved Dijkstra algorithm to perform cumulative hydraulic distance step size and relaxation iterative search of adjacent pixels within the water body range for the node to be interpolated.

[0073] Step 220: If the minimum hydraulic distance path includes the flow interruption point, then the elevation observation point closest to the downstream of the node to be interpolated is determined as the propagation starting point, and the hydraulic distance between the propagation starting point and the node to be interpolated is obtained.

[0074] Specifically, during the path search process, the algorithm checks whether the optimal propagation path passes through the river section marked as physically blocked. If it finds that the path contains a flow-breaking edge with a hydraulic distance set to positive infinity, it determines that the original upstream elevation information transmission link has been interrupted. At this time, the algorithm will automatically search for the elevation observation point that is closest to the node to be interpolated and located on a river section with water in the downstream area of ​​the flow-breaking section, and use it as a new elevation calculation source.

[0075] For example, for a pixel to be interpolated located downstream of a dried-up river or lake, since its hydraulic distance is blocked to infinity and it cannot receive the propagation elevation from the upstream observation point, the system will switch its propagation starting point to the nearest effective observation node downstream of the pixel, and recalculate the hydraulic distance to the node to be interpolated based on the new node.

[0076] Considering that the dry river sections that occur during the dry season in seasonal rivers and lakes will cause the water bodies on both sides of the break point to completely lose their hydraulic connectivity, if the geometric distance interpolation is continued, it will inevitably violate the real hydrological connectivity mechanism. Therefore, this invention introduces a propagation interruption and recalculation mechanism based on flow interruption perception, which can realize the independent elevation propagation of each sub-river segment physically isolated by the flow interruption section, and completely eliminate the systematic interpolation error caused by crossing the flow interruption section.

[0077] Step 230: Elevation propagation is performed based on the hydraulic distance between the propagation starting point and the node to be interpolated to obtain the initial water surface elevation field.

[0078] Specifically, after determining the effective propagation starting point and corresponding hydraulic distance that does not cross the flow interruption point, the elevation observation value of the starting point is gradually calculated and transmitted to the node to be interpolated along the searched effective hydraulic path, according to the natural dynamic physical law of water flowing downhill. When the elevation information of multiple observation points converges at the same node to be interpolated, a weighted fusion calculation is performed to finally generate a preliminary continuous elevation distribution surface covering the entire water body.

[0079] For example, by subtracting the product of the average riverbed gradient and hydraulic distance of the path segment from the elevation observation value at the propagation starting point, the physical upper limit of the downstream pixel elevation cannot be strictly limited to not exceeding the upstream pixel elevation. At the same time, for the interpolation nodes reached by multiple reachable observation points, a Bayesian weighted fusion method is used to calculate the final elevation estimate of each interpolation node by using the elevation uncertainty of the elevation observation point itself and the exponential decay factor based on the hydraulic distance as the fusion weight, thereby obtaining a smooth and physically logical initial water surface elevation field.

[0080] As an optional implementation, the calculation process of the hydraulic distance between adjacent nodes and the Bayesian weighted fusion can be implemented using specific mathematical formulas. Specifically, for two hydraulically connected nodes, their hydraulic distance can be expressed as: ; in, For nodes and nodes The hydraulic distance between them For paths containing Directed edge The actual centerline length, For directed edges directional weighting coefficients For directed edges The bending correction factor, For connecting nodes and nodes The path, which consists of several directed edges composition.

[0081] When performing weighted fusion of elevation propagation, Bayesian weighted fusion is used for the interpolation nodes reached by propagation from multiple reachable observation points. The formula for calculating the fusion weight is as follows: in, For the first The fusion weights propagated from reachable observation points to the nodes to be interpolated For the first The elevation uncertainty of an accessible observation point The hydraulic distance attenuation factor, For the first The hydraulic distance from the reachable observation point to the node to be interpolated The characteristic distance is usually one-third of the average length of the river segment during the observation period.

[0082] The remote sensing estimation method for seasonal river and lake water storage provided by this invention, when obtaining the minimum hydraulic distance path, if a flow interruption point is encountered, flexibly switches the propagation starting point to the nearest elevation observation point downstream of the node to be interpolated to recalculate the hydraulic distance and perform elevation propagation. This enables independent elevation calculation for each sub-river segment physically isolated by the dry and waterless section, effectively blocking the transmission of erroneous elevation information across the dry river segment, completely eliminating the physically meaningless interpolation error across the flow interruption section, and further improving the accuracy and physical rationality of the local water surface elevation field reconstruction.

[0083] Figure 3 This is a flowchart illustrating the process of determining the three-dimensional water surface elevation field of a target, as provided by the present invention. Figure 3As shown, as another optional embodiment provided by the present invention, the initial water surface elevation field is constrained based on the digital elevation model to obtain the target three-dimensional water surface elevation field, including but not limited to the following steps: Step 310: Based on the digital elevation model, determine the longitudinal water surface gradient along the central axis of the river and lake in the initial water surface elevation field.

[0084] The longitudinal water surface gradient refers to the descent gradient or spatial rate of change of the river or lake water surface elevation along the direction of water flow. For example, it can be estimated by extracting the riverbed gradient of each analyzed river segment based on a digital elevation model as a priori benchmark.

[0085] Specifically, the river or lake can be divided into different analysis segments along its centerline. The effective elevation observation points within each analysis segment can be used in conjunction with the weighted least squares method to estimate the gradient. For example, when the number of effective elevation observation points within an analysis segment is insufficient, the weighted average gradient of the upstream and downstream analysis segments can be used as the estimated longitudinal water surface gradient for that segment.

[0086] Step 320: Determine the longitudinal water level gradient constraint term based on the longitudinal water level gradient.

[0087] The longitudinal water surface gradient constraint term refers to a mathematical penalty term used in joint optimization to constrain the elevation difference of continuous pixels along the water flow direction to approach the theoretical gradient distribution law. For example, the longitudinal water surface gradient constraint term can be a residual sum of squares function constructed based on the initial water surface elevation, downstream distance, and calculated longitudinal water surface gradient of the segment.

[0088] Specifically, the longitudinal water surface gradient constraint term can also include local perturbation terms used to characterize local topographic effects such as sharp bends, waterfalls, or sandbars. For example, local perturbation terms can be obtained by fitting the residual sequence with cubic B-splines and then incorporated into the overall equation of the longitudinal water surface gradient to accurately construct the longitudinal water surface gradient constraint term.

[0089] Step 330: Construct a joint optimization function based on the longitudinal water surface gradient constraint, the initial water surface elevation field, and the cross-sectional water level consistency constraint.

[0090] Cross-sectional water level consistency constraints refer to penalty constraints used to ensure that the elevation of each water cell on the same cross-section of a river or lake remains at a basic level or conforms to specific hydrodynamic physical laws. For example, a cross-sectional water level consistency constraint can be a mathematical expression that requires the actual standard deviation of the elevation of all water cells on the same cross-section to be less than or equal to a threshold that is adaptively adjusted according to the river width. Furthermore, for meandering river sections, a water surface superelevation correction caused by the centrifugal force of the water flow can be superimposed.

[0091] As an optional embodiment, the threshold for adaptive adjustment based on river width can be determined by the following formula: ; in, The threshold is adaptively adjusted based on the river width. This represents the width of the river section. The coefficient parameters (such as 0.02, 0.03) need to be calibrated based on measured data from the specific study area to ensure the scientific rationality of the threshold setting.

[0092] It should be noted that before constructing the joint optimization function, the river and lake can be divided into multiple analysis segments based on the tortuosity coefficient and the rate of change of river width, so that constraints can be applied separately. The specific classification criteria are shown in Table 1. Table 1. River and Lake Segmentation Types and Identification Criteria type Judgment conditions Typical river and lake morphology Special handling of constraint models Straight section <![CDATA[C s <1.1]]> straight main channel Standard lateral water level consistency constraints micro-bend 1.1≤C_s<1.3 micro-bending Curve overhang correction <3cm curved section 1.3≤C_s<1.8 The main channel is noticeably bent. <![CDATA[Superelevation correction for curve Δh c > Sharp bends C_s≥1.8 Strongly curved / oxbow lake section Ultra-high correction and enhanced cornering centrifugal force weight Segmentation Detected splitter node Braided / Network Rivers and Lakes Branch water volume distribution constraints It should be noted that for the clearly curved main channel and strongly curved / oxbow lake sections in Table 1, the water surface forms a transverse gradient under the action of centrifugal force, with the water level on the outer side being higher than that on the inner side, and the superelevation correction for these bends is... The calculation formula is: in, This is the correction amount for curve superelevation. The cross-sectional average velocity is... For the river width, It is the acceleration due to gravity. The radius of curvature is calculated geometrically from the central axis of the river or lake.

[0093] The cross-sectional average velocity can be calculated using Manning's formula: in, The cross-sectional average velocity is... The roughness coefficient, For hydraulic radius, This refers to the longitudinal water level gradient. When adjusting the cross-sectional elevation constraints, the outer side of the curve is allowed to be higher than the cross-sectional average. The inner side is lower than the cross-sectional average. .

[0094] A joint optimization function refers to a mathematical model that integrates physical boundary conditions and elevation observation data fitting terms from various dimensions into a single overall objective function. For example, a joint optimization function can be a weighted summation objective function that includes data fitting terms, longitudinal water level gradient constraints, cross-sectional water level consistency constraints, and spatial smoothing constraints.

[0095] Step 340: Solve the joint optimization function to obtain the target three-dimensional water surface elevation field.

[0096] Specifically, numerical iterative optimization algorithms can be used to iteratively find the optimal solution for the joint optimization function containing various constraints until the convergence conditions set by the system are met. For example, the L-curve method can be used to automatically determine the optimal weight coefficients of each constraint term in the objective function, and the conjugate gradient method can be used to iteratively solve the objective function. The iteration ends when the maximum elevation change between two adjacent iterations is less than a set minimum threshold, and the final elevation field is output.

[0097] Considering that existing technologies often result in scattered cross-sectional elevations and violate the physical law that natural water surfaces tend to be level by simply interpolating based on observation points, this invention establishes a joint physical constraint model of longitudinal water surface gradient and cross-sectional water level consistency to globally optimize the initial elevation field. This can significantly reduce the standard deviation of cross-sectional elevations and the root mean square error of the overall elevation field, and achieve continuous three-dimensional morphological reconstruction of water surfaces with clear physical meaning and spatial morphology that conforms to the real hydrodynamic characteristics.

[0098] As an optional implementation, the joint optimization function is obtained based on the following mathematical model: ; in, For joint optimization functions, For the first i The weight of each pixel's elevation observation value The first in the initial water surface elevation field i The elevation observation value of each pixel The target three-dimensional water surface elevation field to be solved is the first... i The elevation value of each pixel. , and These are the weighting coefficients. For the longitudinal water surface gradient constraint term, This is a constraint on the consistency of water level across the cross section. This is a spatial smoothing constraint term.

[0099] The remote sensing estimation method for seasonal river and lake water storage provided by this invention determines the longitudinal water surface gradient along the central axis of the river or lake based on a digital elevation model and constructs a longitudinal water surface gradient constraint term. Then, it combines the initial water surface elevation field and the cross-sectional water level consistency constraint term to construct and solve a joint optimization function. This method can deeply integrate the natural hydrodynamic physical law of the longitudinal water flow gradient and the horizontal water surface leveling into the elevation calculation, effectively correcting the local abnormal fluctuations and unreasonable elevation dispersion that are easily generated by single mathematical space interpolation. It achieves high-precision three-dimensional water surface elevation field reconstruction with clear physical meaning and spatial morphology that strictly conforms to the real natural characteristics.

[0100] Figure 4This is a schematic diagram of the process for constructing a water storage mapping model provided by the present invention, as shown below. Figure 4 As shown, as another optional embodiment provided by the present invention, a water storage mapping model is constructed based on the target three-dimensional water surface elevation field and water body mask data, including but not limited to the following steps: Step 410: Based on the target three-dimensional water surface elevation field and water body mask data, determine the water level area change rate curve.

[0101] The water level area change rate curve refers to a mathematical function curve that reflects the rate at which the two-dimensional coverage area of ​​a water body changes with the vertical elevation of the water surface. For example, it can be obtained by numerical difference calculation of the water surface elevation sequence over many years and the water area data of the corresponding period.

[0102] Specifically, elevation and area data from different observation periods can be spatiotemporally matched to form a serialized dataset. For example, a paired dataset can be formed by combining historical water level observation sequences extracted from the target three-dimensional water surface elevation field with water area extracted from water body mask data of the same period. After performing local weighted regression on the scatter points of the paired dataset, the derivative sequence of water area with respect to water level can be calculated, thereby generating a curve of the rate of change of area with water level.

[0103] Step 420: Obtain the initial water level of the flow interruption, and determine the water level corresponding to the local maximum value of the water level area change rate curve as the initial water level of the floodplain.

[0104] The initial water level of flow interruption refers to the critical elevation threshold at which a seasonal river or lake begins to physically interrupt water flow and the water area tends to be at its minimum during the dry season. For example, the initial water level of flow interruption can be the 5th percentile value of the water level corresponding to the water area being greater than the minimum detectable water area threshold in many years of observation.

[0105] The water level corresponding to a local maximum refers to the elevation coordinate value at which the value on the water level area change rate curve suddenly jumps and reaches the local highest point. For example, when the water level rises from the main channel to the floodplain, the floodplain area is much larger than the cross-sectional area of ​​the main channel, and the area increases sharply with the rate of increase of the water level, forming the peak water level.

[0106] The floodplain initiation water level refers to the critical elevation limit at which the river water level begins to overflow and extend into the floodplains or beaches on both banks after crossing the boundary of the main channel. For example, the floodplain initiation water level can be the water surface elevation value directly mapped after locking the maximum value on the rate of change curve using an automated numerical differentiation algorithm.

[0107] Specifically, this application does not rely on measured hydrological data and manual field surveys. It automatically identifies key hydrological nodes based entirely on multi-source satellite remote sensing data. For example, the extracted multi-year water level area pairing data is input into the algorithm program to automatically extract and calculate the starting water level of the flow interruption and the starting water level of the floodplain, providing extremely accurate physical boundary points for subsequent segmented modeling.

[0108] Step 430: Based on the initial water level of the flow interruption and the initial water level of the floodplain, construct a water storage mapping model. The water storage mapping model includes a sub-model for the flow interruption stage, a sub-model for the main channel water passage stage, and a sub-model for the floodplain stage.

[0109] The flow interruption stage sub-model refers to the bottom boundary model that describes the state of rivers and lakes without continuous water surface and with zero water storage. For example, the flow interruption stage sub-model can be a constant function that sets the water area and water storage to zero when the water surface elevation is lower than the water level at the start of the flow interruption.

[0110] The main channel water passage stage sub-model refers to a mathematical model that describes the nonlinear growth relationship between water level and volume when the water flow is restricted to the main channel of a river or lake. For example, the main channel water passage stage sub-model can be a power function model with an empirical power exponent constructed based on the difference between the elevation and the initial water level of the flow interruption.

[0111] The floodplain stage sub-model refers to a high-order relational model that describes the complex water storage mechanism when water overflows the main channel and covers the floodplains on both banks during periods of high water level. For example, the floodplain stage sub-model can be a piecewise function that is based on the water storage volume of the main channel, superimposed with a floodplain volume term that increases linearly with the water level, and corrected by introducing a floodplain water storage coefficient that is jointly parameterized by the vegetation canopy height and the micro-topographic relief.

[0112] Considering the distinct dry season flow interruption and floodplain transition process of seasonal rivers and lakes, the three variables exhibit strong nonlinear piecewise characteristics. Traditional single power function models assume continuous changes in water morphology, which cannot characterize the zero-volume state and floodplain abrupt changes during the flow interruption phase, leading to huge errors in water storage calculation. Therefore, this invention establishes a piecewise nonlinear function adapted to the entire hydrological process based on automatically identified key water level thresholds, and quantitatively considers the impact of floodplain vegetation interception and micro-topography. This enables a complete and accurate characterization of the dynamic changes in water storage throughout the year for seasonal rivers and lakes, significantly reducing the relative root mean square error in the calculation of water storage throughout the entire year.

[0113] The remote sensing estimation method for seasonal river and lake water storage provided by this invention constructs a three-stage sub-model that includes the flow interruption stage, the main channel water passage stage, and the floodplain stage by combining the initial water level of the flow interruption. This model can accurately match the strong nonlinear segmented hydrological characteristics of seasonal rivers and lakes during the dry season flow interruption and the transition from the main channel to the floodplain during the wet season. It completely overcomes the inherent defects of traditional single function models that cannot characterize the zero-volume state of flow interruption and the sudden change of floodplain, thereby significantly reducing the measurement error of the annual water storage of seasonal rivers and lakes.

[0114] In another embodiment of the present invention, the water storage capacity of the target area is determined using a water storage mapping model, including: if the water level of the target area is less than the initial water level of the flow interruption, the water storage capacity of the target area is determined using a flow interruption stage sub-model; if the water level of the target area is greater than or equal to the initial water level of the flow interruption and less than the initial water level of the floodplain, the water storage capacity of the target area is determined using a main channel water passage stage sub-model; if the water level of the target area is greater than or equal to the initial water level of the floodplain, the water storage capacity of the target area is determined using a floodplain stage sub-model.

[0115] Specifically, when performing dynamic estimation of water storage, the system will automatically match the corresponding hydrophysical stage based on the current observed or inverted water level value and select the appropriate mathematical model to solve for the volume.

[0116] For example, when the water level H is lower than the initial water level H of the flow interruption. min At this time, the water body is in the interrupted flow stage, and the water area A(H) = 0 and the water storage volume V(H) = 0. When the water level H is greater than or equal to the initial water level of the flow interruption H min And less than the initial water level H of the floodplain bank At this time, the main channel is in the water-filling stage, and the water area is: A(H)= α A ×(H-H min )^ β A ; The water storage capacity is: V(H)= α V ×(H-H min )^ β V ; When the water level H is greater than or equal to the initial flood level H bank At that time, the area was in the floodplain stage, and the water surface area was: A(H)=A bank + α A2 ×(H-H bank )^ β A2 +A fp(H) ; The water storage capacity is: V(H)=V bank + α V2 ×(H-H bank )^ β V2 + γ fp ×A fp(H) ×(H-H bank ); Among them, A bank and V bank These represent the water area and water storage volume at the initial floodplain water level, A. fp(H) Let H be the area of ​​the floodplain at water level H. α A , β A , α V , β V , α A2 , β A2 , α V2 , β V2 These are all model parameters that need to be calibrated. γ fp The floodplain water storage coefficient.

[0117] The floodplain water storage coefficient is determined based on the vegetation canopy height and micro-topographic undulation of the target area.

[0118] As an optional implementation, the parameters in the water storage mapping model ( α A , β A , α V , β V (etc.) were obtained by constructing a water level-area paired dataset and simultaneously calibrating it using the Levenberg-Marquardt nonlinear least squares algorithm. The objective function for optimization is: ; in, For the common constraint parameter vector, For the observed water area, The area calculated for the model. The observed water storage is calculated by integrating the three-dimensional water surface elevation field of the target. The water storage capacity calculated by the model. This is the weighting coefficient between the water storage residual term and the area residual term.

[0119] It should be noted that the parameterized equation for the floodplain storage coefficient can be expressed as: γ fp = γ base ×(1-k1×h canopy mean )×(1-k2× σ dem ); in, γ fp The floodplain water storage coefficient, γ base The basic water storage coefficient is given by k1, where k is the vegetation interception coefficient, and h is the base water storage coefficient for the flat, bare land floodplain. canopy mean is the mean height of the vegetation canopy, and k2 is the micro-topography correction factor. σ dem This refers to the degree of micro-topographic relief.

[0120] To further improve the efficiency and accuracy of model calibration, this application provides reference ranges for floodplain water storage coefficients for different typical floodplain types, as shown in Table 2: Table 2 Reference values ​​of floodplain water storage coefficient for different typical floodplain types Floodplain type Vegetation canopy height Micro-topographic relief Reference value of floodplain water storage coefficient bare land flat floodplain <0.5m <0.1m 0.90-0.95 Sparse herbaceous vegetation flooded the beach 0.5-1.5m 0.1-0.3m 0.78-0.88 shrubland 1.5-5m 0.2-0.5m 0.65-0.78 Reeds / emergent plants floodplain 1-4m 0.1-0.3m 0.60-0.72 woodland floodplain >5m >0.3m 0.50-0.65 Considering that existing water storage modeling methods typically assume the water body to be a perfect geometric shape when dealing with seasonal high-water-level floodplains, the quantitative consideration of the interception effect of vegetation canopy and the influence of micro-topography in the floodplain water storage coefficient is completely lacking. This leads to a serious overestimation of the result by directly multiplying the theoretical floodplain area by the floodplain water depth. Therefore, this invention innovatively introduces a floodplain water storage coefficient based on the joint parameterization of vegetation canopy height and micro-topographic undulation in the floodplain stage sub-model. This can accurately restore the real physical water storage state under complex floodplain surface environments, effectively deduct the space volume occupied by non-water bodies, and greatly improve the estimation accuracy and physical rationality of seasonal river and lake water storage during the high-water season.

[0121] The remote sensing estimation method for seasonal river and lake water storage provided by this invention adaptively selects a sub-model for the flow interruption stage, the main channel water passage stage, or the floodplain stage to calculate water storage based on the numerical relationship between the target area water level and the water level at the start of the flow interruption and the water level at the start of the floodplain. In the floodplain stage sub-model, a floodplain water storage coefficient determined based on the vegetation canopy height and micro-topographic undulation is introduced. This method can accurately match the real water storage physical mechanism of different hydrological stages of seasonal rivers and lakes, and improve the physical rationality and calculation accuracy of dynamic monitoring of water storage throughout the entire process.

[0122] Figure 5 This is a schematic diagram of the process for obtaining elevation observation data of a target area provided by the present invention, as shown below. Figure 5 As shown, as another optional embodiment provided by the present invention, obtaining elevation observation data of the target area includes, but is not limited to, the following steps: Step 510: Obtain the initial elevation observation data of the target area and the multispectral remote sensing image that matches the initial elevation observation data in terms of acquisition time.

[0123] Initial elevation observation data refers to the raw satellite radar altimetry products that have not undergone advanced anomaly removal through cross-modal feature fusion or observation surface data that has only been filtered by basic quality flags. For example, initial elevation observation data can be the set of water surface elevation observation pixels contained in the secondary grating elevation products of surface water and ocean topography satellites.

[0124] Matching the acquisition time means that the transit time of data acquired by two different satellite sensors is within the same small time window, so as to ensure that the observed water body morphology and water level status are basically consistent. For example, clear sky images with cloud cover of less than 10% can be selected and time-matched with images within three days before and after the transit time window of satellite radar elevation observation data.

[0125] Multispectral remote sensing imagery refers to Earth observation satellite images that contain information from multiple electromagnetic bands, including visible light, near-infrared, and short-wave infrared. For example, multispectral remote sensing imagery could be Sentinel-2 high-resolution surface reflectance imagery products.

[0126] Step 520: Based on the near-infrared band of the multispectral remote sensing image, determine the near-infrared gradient characteristics of the location where the initial elevation observation data is located.

[0127] The near-infrared gradient feature of the location of the initial elevation observation data refers to the degree of change of spatial pixel value or the normalized feature of spatial gradient amplitude of the elevation observation pixel in the corresponding optical image near-infrared band. For example, the near-infrared gradient feature can be the gradient amplitude calculated with the elevation observation point as the center within the corresponding pixel window of the optical image, using the significant difference in reflectivity between water and land in the near-infrared band, and then normalized by quantile value.

[0128] Specifically, by extracting the near-infrared band of multispectral remote sensing images, and using an edge detection operator to calculate the gradient amplitude within a set pixel window with each initial elevation observation point as the center, the gradient amplitude can be normalized using the specific quantile value of the near-infrared gradient of the water boundary region in the current image, thereby determining the near-infrared gradient features that reflect the possibility that the point is in a water-land mixed pixel region.

[0129] Step 530: Based on near-infrared gradient features, hydraulic model deviation features, boundary distance features, and water body proportion features of the initial elevation observation data, determine the comprehensive anomaly score of the initial elevation observation data.

[0130] Hydraulic model deviation characteristics refer to the normalized deviation between the actual elevation observation value and the elevation value predicted according to the hydraulic dynamic physical model. For example, hydraulic model deviation characteristics can be the ratio obtained by dividing the absolute difference between the initial elevation observation value and the hydraulic distance elevation propagation prediction value by the propagation uncertainty. It is highly sensitive to abnormal elevations that are physically unreasonable.

[0131] Boundary distance features refer to the spatial risk measure of the distance between the location of an elevation observation point and the nearest known physical boundary of a continuous water body. For example, the boundary distance feature can be a feature value obtained by converting the Euclidean distance from the observation point to the nearest optical remote sensing water body boundary through a negative exponential function. The larger the boundary distance feature value, the closer it is to the boundary, and the higher the prior probability of being affected by mixed pixels.

[0132] Water body proportion feature refers to the proportion of the area actually covered by water within an elevation observation pixel to the total area of ​​that pixel. For example, the water body proportion feature can be the land proportion value calculated using the water body proportion field provided within the satellite elevation product. The larger the value, the larger the land area within the pixel and the more severe the mixed pixel effect.

[0133] The comprehensive anomaly score refers to a comprehensive evaluation value used to quantify the anomaly probability of a single elevation observation point by weighted fusion of feature indicators from multiple modalities and physical sources through a machine learning model. For example, it can be obtained by multiplying the optimal weight coefficients learned by the support vector machine model trained with manually labeled or historical data by the corresponding elevation deviation feature, boundary distance feature, near-infrared gradient feature, and water body proportion feature, and then summing the four product results.

[0134] As an optional embodiment, the comprehensive anomaly score is obtained by weighted summation of four types of features, specifically including hydraulic model deviation features, boundary distance features, near-infrared gradient features, and water body proportion features, which can be calculated using the following formulas: f1=|H SWOT -H hydraulic | / σ hydraulic ; Where f1 is the hydraulic model deviation characteristic, H SWOT H is the initial elevation observation value. hydraulic The predicted elevation for hydraulic distance elevation propagation. σ hydraulic To propagate uncertainty.

[0135] f2=exp(-d boundary / d ref ); Where f2 is the boundary distance feature, d boundary Let d be the Euclidean distance from the observation point to the nearest water body boundary. ref This is the reference distance threshold.

[0136] f3=G NIR / G NIR ,max; Where f3 is the near-infrared gradient feature, G NIR G represents the gradient magnitude within a near-infrared pixel window centered on the observation point. NIR ,max, represents a specific quantile of the near-infrared gradient in the water boundary region of the current image. This near-infrared gradient feature incorporates the near-infrared band gradient of the optical image into the radar measurement quality assessment, effectively identifying mixed land and water pixels.

[0137] f4=1-water frac ; Where f4 represents the water body proportion characteristic, water frac This represents the percentage of water within a given pixel in an elevation observation product.

[0138] It should be noted that when removing anomalies based on the comprehensive anomaly score, an adaptive threshold can be used to identify first-level suspected anomalies firstly, and then the Local Outlier Factor (LOF) algorithm can be used for secondary verification: calculate the local reachability density of the first-level suspected anomalies. If the LOF value is greater than the set upper limit threshold (e.g., 2.0), it is finally confirmed as an anomaly and removed. If the LOF value is in the middle range (e.g., between 1.5 and 2.0), the point is retained but its weight in subsequent calculations is reduced.

[0139] Step 540: Based on the comprehensive anomaly score, anomalies are removed from the initial elevation observation data to obtain the elevation observation data.

[0140] Anomaly removal refers to the process of removing or reducing the weight of invalid elevation points that have excessively high comprehensive anomaly scores and are judged to be affected by mixed water and land pixel pollution or systematic bias from the observation dataset according to the set discrimination rules and verification mechanisms. For example, it can be achieved by combining a dynamic threshold that is adaptively adjusted according to the river width for preliminary screening and using the local outlier factor algorithm for secondary independent spatial density verification.

[0141] Specifically, the comprehensive anomaly score of each initial elevation observation point can be compared with the adaptive threshold corresponding to the width of its river segment. Points with scores greater than the threshold are marked as first-level suspected anomalies. Then, the local outlier algorithm is used to calculate the nearest neighbor density for these suspected anomalies. Points with local outliers greater than the set upper limit are confirmed as real anomalies and removed. At the same time, the time series consistency test can be combined to permanently remove points that have been marked as suspected anomalies for multiple consecutive periods. Finally, what is retained is the cleaned high-precision elevation observation data.

[0142] The remote sensing estimation method for seasonal river and lake water storage provided by this invention extracts near-infrared gradient features and combines them with hydraulic model deviation features, boundary distance features, and water body proportion features to comprehensively calculate anomaly scores for anomaly removal. This method can effectively perform cross-modal collaborative fusion of spatial spectral information from optical images and quality assessment of radar altimetry data, accurately distinguishing between reasonable elevation changes caused by real terrain undulations and systematic elevation deviations caused by mixed pixels at the land-water boundary, thereby improving the accuracy of identifying anomalous elevation points.

[0143] In another embodiment of the present invention, the method of determining the water storage of the target area using the water storage mapping model further includes: if water mask data of the target area at the time step to be measured is obtained, the water mask data is input into the water storage mapping model to determine the water storage at the time step to be measured; if water mask data of the target area at the time step to be measured is not obtained, the precipitation, water surface evaporation and outflow of the target area are obtained.

[0144] The time step to be measured refers to the time interval benchmark used for continuous dynamic water storage measurement and hydrological status update. For example, the time step to be measured can be a daily time window that needs to fill in blank observation data due to limitations of satellite revisit cycle or cloud and rain weather.

[0145] Precipitation, evaporation, and outflow in the target area refer to key meteorological and hydraulic flux parameters of the water cycle that drive the dynamic changes in the watershed's hydrology and determine the increase or decrease in water volume. For example, precipitation can be the areal precipitation rate obtained using global precipitation measurement satellite products, evaporation can be the regional water surface evaporation consumption estimated based on the Penman-Monteith formula, and outflow can be the amount of water flowing out of the downstream section of the river segment estimated based on the Manning formula.

[0146] Based on precipitation, water surface evaporation, outflow, and water storage at the previous time step, a water balance equation is constructed. By numerically integrating the water balance equation, the water storage at the time step to be measured is determined.

[0147] The water balance equation is an ordinary differential equation model that describes the quantitative relationship between the rate of change of water storage over time and various input and output hydrological elements within a specific spatial region, based on the law of conservation of mass. For example, the water balance equation can be specifically expressed as: dV / dt = P(t) × A_catch × α -E(t)×A_water(t)-Q_out(t); Where dV / dt is the rate of change of water storage over time, P(t) is the precipitation, A_catch is the catchment area of ​​the watershed, E(t) is the evaporation from the water surface, and A_water(t) is the water body coverage area. α Here, Q_out(t) represents the runoff coefficient, and Q_out(t) represents the outflow rate.

[0148] Specifically, when cloud cover over the target area prevents the extraction of effective water body mask areas from multispectral remote sensing images, the system automatically activates a water balance model to continuously fill in the data sequence. For example, the fourth-order Runge-Kutta method can be used to perform daily-step numerical integration of the constructed water balance ordinary differential equation. The initial water storage known at the previous time step is added to the volume change generated by the integral of various hydrological fluxes within this time step. Thus, during the gap period lacking effective optical remote sensing observations, the dynamic water storage of the target area at the time step to be measured can be continuously deduced and output, achieving high temporal resolution three-dimensional dynamic reconstruction of water storage under the limitation of satellite revisit cycle.

[0149] As an optional embodiment, after determining the water storage volume at the time step to be measured by numerical integration of the water balance equation, the system may further include updating the water storage volume using Kalman filtering. Specifically, whenever a new set of valid elevation observation data or cloud-free multispectral remote sensing imagery is acquired, the system uses the new observation data as the measurement value and updates the water balance equation using Kalman filtering.

[0150] For example, by calculating the Kalman gain, which is jointly determined by the integral error covariance and the observation error covariance, the predicted water storage can be corrected and compensated by utilizing the deviation between the actual observed water storage and the water storage predicted by the integral. This effectively eliminates the cumulative error caused by daily step integration and achieves a more robust and accurate dynamic reconstruction of daily-scale water storage.

[0151] The remote sensing estimation method for seasonal river and lake water storage provided by this invention estimates water storage by constructing a water balance equation and combining it with the water storage of the previous time step for numerical integration. This method can effectively overcome the objective observation limitations such as the susceptibility of multispectral remote sensing images to cloud cover and the long satellite revisit cycle. It can accurately fill in the missing information on water volume during the blank period when remote sensing data is lacking, thereby realizing high temporal resolution continuous dynamic water storage reconstruction that is not constrained by weather and observation frequency, and ensuring the integrity of the seasonal river and lake water resource monitoring process in the time series.

[0152] It should be noted that, in order to comprehensively evaluate the reliability of seasonal river and lake water storage estimates, this invention can also employ Monte Carlo simulation to quantify the overall uncertainty of water storage calculation. Specifically, the system can comprehensively consider multiple error sources, including but not limited to elevation errors in elevation observation data, relatively uniform distribution errors in water area extraction from multispectral remote sensing images, elevation errors in digital elevation models, and the uncertainty of the covariance matrix of the nonlinear least squares parameters of the water storage mapping model. By performing multiple Monte Carlo samplings on the above-mentioned multi-source errors (e.g., running 1000 times), and for each sampling result, performing the entire process of hydraulic distance elevation propagation, joint optimization and reconstruction of the elevation field, and piecewise nonlinear modeling, a 95% confidence interval for the water storage estimation result can be output. This uncertainty quantification mechanism can provide a highly valuable and reliable reference for dynamic water resource management and flood early warning in data-scarce areas.

[0153] It should be noted that, in order to verify the synergistic effect and independent contribution of each technical module of the present invention, a systematic ablation experiment was also conducted. Compared with the unconstrained baseline method that only uses the original elevation data and water area (whose relative root mean square error of water storage is as high as 35%-45%), the relative root mean square error of water storage of the complete method of the present invention can be reduced to 10%-15%.

[0154] Specifically, removing the anomaly removal module based on near-infrared gradients leads to a significant increase in elevation field error due to anomaly contamination; removing the hydraulic distance propagation module based on flow interruption sensing and replacing it with conventional spatial interpolation results in meaningless interpolation at the flow interruption point, causing the water storage error to rise to 28%-35%; removing the longitudinal and transverse joint constraint module leads to a deterioration in the consistency of transverse elevation; and removing piecewise nonlinear modeling and replacing it with a single power function leads to a significant increase in errors during the floodplain period and seasonal transition period (errors rise to 28%-38%). This demonstrates that the multi-source anomaly removal, hydraulic distance propagation, joint physical constraints, and piecewise modeling modules in this invention constitute an inseparable organic whole, with significant synergistic effects among the modules.

[0155] Figure 6This is a schematic diagram of the structure of the remote sensing estimation device for seasonal river and lake water storage provided by the present invention, as shown below. Figure 6 As shown, it mainly includes, but is not limited to: The data acquisition module 610 is used to acquire water mask data, elevation observation data and digital elevation model of the target area.

[0156] The flow interruption identification module 620 is used to determine the central axis of the river and lake in the target area based on water body mask data, identify waterless sections based on the central axis of the river and lake, and determine the flow interruption point.

[0157] The topology building module 630 is used to construct a river and lake network topology map based on the central axis of the river and lake, the flow interruption point, and elevation observation data; and to determine the hydraulic distance based on the river and lake network topology map.

[0158] The elevation reconstruction module 640 is used to propagate elevation data based on hydraulic distance to obtain an initial water surface elevation field; and to apply water level constraints to the initial water surface elevation field based on a digital elevation model to obtain a target three-dimensional water surface elevation field.

[0159] The water storage estimation module 650 is used to construct a water storage mapping model based on the target three-dimensional water surface elevation field and water body mask data, and to determine the water storage of the target area using the water storage mapping model.

[0160] It should be noted that the seasonal river and lake water storage remote sensing estimation device provided by the present invention can execute the seasonal river and lake water storage remote sensing estimation method described in any of the above embodiments during specific operation, which will not be elaborated in this embodiment.

[0161] The seasonal river and lake water storage remote sensing estimation device provided by the present invention constructs a river and lake network topology map through the flow interruption point to determine the hydraulic distance that reflects the true hydraulic connectivity characteristics. The hydraulic distance drives the elevation propagation to obtain the target three-dimensional water surface elevation field, and then constructs a water storage mapping model to estimate the water storage, thereby solving the problem of elevation field distortion and realizing accurate monitoring of water storage.

[0162] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0163] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A remote sensing method for estimating seasonal river and lake water storage, characterized in that, include: Acquire water mask data, elevation observation data, and digital elevation model of the target area; Based on the water body mask data, the central axis of the river and lake in the target area is determined, and the waterless section is identified based on the central axis of the river and lake to determine the flow interruption point. A river and lake network topology map is constructed based on the central axis of the river and lake, the flow interruption point, and the elevation observation data. Based on the aforementioned river and lake network topology map, the hydraulic distance is determined; Based on the hydraulic distance, the elevation observation data is propagated to obtain an initial water surface elevation field; based on the digital elevation model, the initial water surface elevation field is constrained by water level to obtain a target three-dimensional water surface elevation field. Based on the target three-dimensional water surface elevation field and the water body mask data, a water storage mapping model is constructed, and the water storage volume of the target area is determined using the water storage mapping model.

2. The method for remote sensing estimation of seasonal river and lake water storage capacity according to claim 1, characterized in that, The determination of hydraulic distance based on the river and lake network topology map includes: Obtain the flow direction weights and tortuosity coefficients between adjacent nodes in the river and lake network topology diagram; The hydraulic distance between adjacent nodes is determined based on the flow direction weight, the curvature coefficient, and the length of the river / lake central axis between adjacent nodes. The hydraulic distance between adjacent nodes that cross the flow interruption point is set to infinity.

3. The method for remote sensing estimation of seasonal river and lake water storage capacity according to claim 1, characterized in that, The process of performing elevation propagation on the elevation observation data based on the hydraulic distance to obtain an initial water surface elevation field includes: Obtain the minimum hydraulic distance path from the node to be interpolated in the river and lake network topology map to each of the elevation observation points, wherein the node to be interpolated is a node in the river and lake network topology map that lacks elevation observation data; If the minimum hydraulic distance path includes the flow interruption point, then the elevation observation point closest to the downstream of the node to be interpolated is determined as the propagation starting point, and the hydraulic distance between the propagation starting point and the node to be interpolated is obtained. Elevation propagation is performed based on the hydraulic distance between the propagation starting point and the node to be interpolated to obtain the initial water surface elevation field.

4. The method for remote sensing estimation of seasonal river and lake water storage capacity according to claim 1, characterized in that, The step of applying water level constraints to the initial water surface elevation field based on the digital elevation model to obtain the target three-dimensional water surface elevation field includes: Based on the digital elevation model, the longitudinal water surface gradient along the central axis of the river and lake is determined. The longitudinal water level gradient constraint term is determined based on the aforementioned longitudinal water level gradient. Based on the longitudinal water surface gradient constraint, the initial water surface elevation field, and the cross-sectional water level consistency constraint, a joint optimization function is constructed. Solving the joint optimization function yields the target three-dimensional water surface elevation field.

5. The method for remote sensing estimation of seasonal river and lake water storage capacity according to claim 4, characterized in that, The joint optimization function is derived based on the following mathematical model: ； in, For the joint optimization function, For the first i The weight of each pixel's elevation observation value The first in the initial water surface elevation field i The elevation observation value of each pixel For the target three-dimensional water surface elevation field to be solved, the first... i The elevation value of each pixel. , and These are the weighting coefficients. For the longitudinal water surface gradient constraint term, This refers to the cross-sectional water level consistency constraint term. This is a spatial smoothing constraint term.

6. The method for remote sensing estimation of seasonal river and lake water storage according to claim 1, characterized in that, The construction of a water storage mapping model based on the target three-dimensional water surface elevation field and the water body mask data includes: Based on the target three-dimensional water surface elevation field and the water body mask data, the water level area change rate curve is determined; Obtain the initial water level of the flow interruption, and determine the water level corresponding to the local maximum value of the water level area change rate curve as the initial water level of the floodplain; Based on the initial water level of the flow interruption and the initial water level of the floodplain, the water storage mapping model is constructed. The water storage mapping model includes a sub-model of the flow interruption stage, a sub-model of the main channel water passage stage, and a sub-model of the floodplain stage.

7. The method for remote sensing estimation of seasonal river and lake water storage according to claim 6, characterized in that, Determining the water storage capacity of the target area using the water storage mapping model includes: If the water level in the target area is lower than the initial water level of the flow interruption, the water storage capacity of the target area is determined using the flow interruption stage sub-model. If the water level in the target area is greater than or equal to the initial water level of the flow interruption and less than the initial water level of the floodplain, then the water storage capacity of the target area is determined using the main channel water passage stage sub-model. If the water level in the target area is greater than or equal to the initial water level of the floodplain, the water storage capacity of the target area is determined using the floodplain stage sub-model. The floodplain stage sub-model includes a floodplain water storage coefficient, which is determined based on the vegetation canopy height and micro-topographic undulation of the target area.

8. The method for remote sensing estimation of seasonal river and lake water storage according to claim 1, characterized in that, The acquisition of elevation observation data for the target area includes: Acquire initial elevation observation data of the target area and multispectral remote sensing images that match the acquisition time of the initial elevation observation data; Based on the near-infrared band of the multispectral remote sensing image, the near-infrared gradient characteristics of the location of the initial elevation observation data are determined. Based on the near-infrared gradient features, the hydraulic model deviation features, boundary distance features, and water body proportion features of the initial elevation observation data, the comprehensive anomaly score of the initial elevation observation data is determined. The initial elevation observation data is then processed by removing anomalies based on the comprehensive anomaly score to obtain the elevation observation data.

9. The method for remote sensing estimation of seasonal river and lake water storage according to claim 1, characterized in that, The step of determining the water storage capacity of the target area using the water storage mapping model further includes: If water mask data of the target area at the time step to be measured is obtained, the water mask data is input into the water storage mapping model to determine the water storage at the time step to be measured. If the water mask data of the target area at the time step to be measured is not obtained, then the precipitation, water surface evaporation and outflow of the target area are obtained; Based on the precipitation, the water surface evaporation, the outflow, and the water storage at the previous time step, a water balance equation is constructed, and the water storage at the time step to be measured is determined by numerical integration of the water balance equation.

10. A remote sensing estimation device for seasonal river and lake water storage, characterized in that, include: The data acquisition module is used to acquire water mask data, elevation observation data, and digital elevation models of the target area; The flow interruption identification module is used to determine the central axis of the river and lake in the target area based on the water body mask data, identify waterless sections based on the central axis of the river and lake, and determine the flow interruption point. The topology construction module is used to construct a river and lake network topology map based on the central axis of the river and lake, the flow interruption point, and the elevation observation data. Based on the aforementioned river and lake network topology map, the hydraulic distance is determined; The elevation reconstruction module is used to perform elevation propagation on the elevation observation data based on the hydraulic distance to obtain an initial water surface elevation field; and to apply water level constraints to the initial water surface elevation field based on the digital elevation model to obtain a target three-dimensional water surface elevation field. The water storage estimation module is used to construct a water storage mapping model based on the target three-dimensional water surface elevation field and the water body mask data, and to estimate the water storage of the target area using the water storage mapping model.

Images