Drainage basin sediment production estimation method based on dynamic evolution process of river blocking barrier lake

Abstract

The invention discloses a river basin sediment production estimation method based on a dynamic evolution process of a river-blocking barrier lake, and the method comprises the steps: carrying out the comparative analysis of key physicochemical characteristic parameters through an identification event layer in a residual barrier lake deposition sequence, and constructing a disaster chain extreme surface process river basin sediment production estimation method. Water level changes, hydrodynamic conditions and material source changes of the barrier lake are comprehensively analyzed, different sand production stages from the stable period, the decline period and the extinction period of the barrier lake are divided, the sediment change history from lacustrine facies suspended load to flood plain bed load is analyzed, and a complete barrier lake formation-evolution full life cycle evolution process mode is constructed. A layered triangular irregular network model is constructed to obtain layered accumulated sediment volume data, the layered volume and mass are calculated based on a dry density field of sedimentary stratification and spatial interpolation so as to calculate the total sediment yield, a physical information neural network model inverts a sedimentary thickness field, sediment yield estimation is achieved, and the non-stationary dynamic evolution characteristics of water and sediment of the barrier lake are quantified.

Description

Technical Field

[0001] This invention relates to the field of flood control and disaster reduction technology, specifically to a method for estimating watershed sediment yield based on the dynamic evolution of river dams and landslide-dammed lakes. Background Technology

[0002] Disaster chains, as extreme watershed-scale surface processes and geomorphological catastrophic events, record the complex upstream water and sediment flows during the formation and evolution of various dammed events such as landslides, rockfalls, and glacial lakes. These events result in outburst floods that cause severe river erosion and sediment deposition downstream. Different types of river-damming lakes record upstream runoff and sediment production processes, altering watershed sediment transport mechanisms and even amplifying the erosion and sediment deposition effects in downstream rivers, thus exacerbating the response of river-damming disasters to the river's water and sediment cycle. The sediment load generated by extreme disasters in mountainous major rivers far exceeds that under normal climatic conditions, often requiring decades or even centuries for a single watershed-damming lake to recover. In particular, the disaster chain effect of watershed erosion and deposition has a certain lag and long-term nature, creating flood control challenges for the planning and construction of major transportation projects (bridges, roadbeds), cascade hydropower projects, inter-basin water transfer projects, and important towns. Currently, short-term sediment hydrological observation data are scarce in high mountain and canyon areas. How to assess the river channel recovery cycle under extreme disaster chains and the evolution of sediment transport, erosion and deposition in the upper and lower reaches of the basin has become an urgent problem to be solved.

[0003] Currently, the main methods for calculating watershed sediment yield are river sampling and reservoir sedimentation measurement. River sampling involves setting up test sections at hydrological stations and using flow measurement and sediment sampling equipment to measure flow rate and sediment concentration. Reservoir sedimentation is measured using cross-sectional methods, topographic methods, or a combination of both, to determine sediment deposition volume. Then, combined with data such as sediment dry density and the reservoir's sediment retention ratio, the watershed sediment yield is calculated. However, both river sampling and reservoir sedimentation measurement methods heavily rely on hydrological station or reservoir sediment observations, which are rarely directly observable in remote mountainous areas with blocked rivers, landslide dams, or other data-scarce regions.

[0004] Another commonly used method for sediment estimation is the empirical relationship method or mathematical model method. Based on long-term measured sediment data from numerous hydrological stations within the region, the annual average sediment transport modulus of each station is calculated. Taking into account underlying surface conditions such as geology, geomorphology, and vegetation, a regional sediment transport modulus contour map is drawn using geographic interpolation methods (such as Kriging). Through multiple regression analysis, the statistical relationship between the annual average sediment transport or sediment transport modulus at the watershed outlet section and major influencing factors (such as watershed area, channel gradient, precipitation, vegetation cover, lithological factors, etc.) is established. General formulas are also used. S = k Aα Jβ Pγ C δ Xn Where: S: average annual sediment load (t) or sediment load modulus A: Watershed area (km²), the most commonly used predictor, usually exhibits a significant negative scaling effect with sediment transport modulus. J: Mean channel gradient or main channel slope. P: Factors characterizing rainfall erosivity (e.g., average annual precipitation, EI). 30 C: Vegetation cover and management factors (such as NDVI). K: Regional comprehensive coefficient, covering factors that are not clearly expressed, such as lithology and soil erodibility. : Regression indices for each factor. This is a spatialized empirical method with a simple structure and low data requirements, widely used in estimation at the regional scale and in areas without data. However, it is highly regional, its accuracy depends on the density and uniformity of station distribution, its parameter extrapolation is poor, and it cannot simulate the sediment process of a single event. It is suitable for macro-planning and trend judgment. Due to the limitations of sediment observation of blocked rivers and landslide dams, especially in the dynamic evolution of different types of blocked rivers under different geological environments, there is a lack of universal empirical methods and models for watershed sediment transport and production.

[0005] A spatiotemporal discretization simulation was performed using physical equations, and the sediment continuity equation and motion equation were jointly solved using a distributed hydrological model. The flow continuity equation and motion equation (Saint-Venant equations), and the sediment continuity equation (one-dimensional): ( AC ) / t + ( QC ) / x + ( Bzb ) / t =0. Where, For the water flow area, This represents the average sand content of the cross section. For traffic, For the river width, This represents the riverbed elevation. This equation describes the erosion and deposition of the riverbed. The dynamic balance between sediment yield and sediment transport capacity of water flow is crucial. This method relies on refined simulations with well-defined controlled physical equations. However, the parameters of models for sediment yield and transport in dammed lake basins, which affect the erosion process and mechanism of dams, are complex and require a huge amount of data. Conventional distributed hydrological models do not consider changes in sediment transport due to dam failures. The calculation and simulation of sediment yield in the basin are influenced by the dynamic evolution of controlled natural dammed lakes. Existing water-sediment coupling mechanisms are insufficient to achieve dynamic simulation of the entire basin process.

[0006] It is evident that conventional sediment estimation relies heavily on observational data from river hydrological stations and reservoirs. Most landslide-dammed lakes are located in remote mountainous areas, directly limiting sediment estimation in areas without available data. Similarly, empirical formulas require long-term data accumulation and cannot meet the requirements for sediment calculation during dynamic evolution of landslide-dammed lakes. Furthermore, the complex parameters of distributed physical models and the clear understanding of erosion and sediment transport mechanisms directly affect the calculation accuracy. The erosion mechanisms of different types of landslide dams in different regions are still unclear, and the inaccurate identification of the long-term dynamic evolution of landslide-dammed lakes also affects the estimation of sediment transport in the disaster chain. Summary of the Invention

[0007] To address the aforementioned shortcomings in existing technologies, this invention provides a watershed sediment yield estimation method based on the dynamic evolution of river dams and landslide dams, which solves the problem that existing technologies struggle to accurately estimate watershed sediment yield in areas lacking hydrological data.

[0008] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: A watershed sediment yield estimation method based on the dynamic evolution of landslide dammed lakes is provided, which includes the following steps: S1. Obtain basic data on the landslide dammed lake, including the lake shore elevation line, water level, topographic data, the inundation range of the landslide dammed lake before and after the river blockage, the thickness of sediment deposition, the core particle size of the landslide dammed lake, and parameters that indicate sediment dynamics and the intensity of sediment weathering. S2. Based on the inundation range of the landslide dammed lake, the distribution of lacustrine sediments, the water level change sequence, the grain size of the landslide dammed lake core, and parameters that indicate sediment dynamics and the intensity of source weathering, the life evolution stages of the landslide dammed lake are divided, including the stable period, the retreat period, and the disappearance period. S3. Based on the inundation range and topographic data of the landslide dammed lake, calculate the area and storage capacity of the landslide dammed lake, construct the relationship curve between lake water level and storage capacity, and extract the lake basin water depth and storage capacity determined by different life evolution stages of the landslide dammed lake. S4. Based on the lake basin water depth, reservoir capacity, residual dammed lake sediment particle size and sediment deposition thickness at different life evolution stages of the landslide dammed lake, TIN models for different evolution stages after deposition were constructed using the lake shore elevation line and lacustrine core stratification as elevation control points and the original lake basin elevation data. S5. Obtain the sediment volume by the volume difference between the post-deposition and original TIN models, plot the relationship curve between lake shore elevation and sediment accumulation capacity, and coordinate the analysis with the relationship curve between lake water level and capacity to correct the remaining effective capacity and water-sediment ratio of the landslide dammed lake, and couple the analysis of the evolution process of the water-sediment curve. S6. Based on the sediment depth or sediment properties of the landslide dammed lake at different evolution stages, the lacustrine core is divided into several horizontal layers for vertical stratification. The measured dry density data of sediment in the corresponding horizontal layer is interpolated within the layer using two-dimensional kriging space interpolation to obtain the dry density planar map of each horizontal layer. S7. Based on the cumulative sediment volume at the elevation section, calculate the volume of each horizontal layer, and combine it with the corresponding dry density plane map to calculate the mass of each layer, and then obtain the total sediment volume. S8. The reservoir capacity of the landslide dammed lake, the water-sediment curve, the cumulative sediment volume at the elevation section, the dry density plane map of the horizontal layer, and the water depth and velocity field simulated by the hydrodynamic model are used together as the external boundary conditions and geometric-physical constraints of the PINNs model. The hydrological parameters and spatiotemporal coordinates of the water-sediment ratio characteristics of the landslide dammed lake at different evolution stages are used as inputs, and the core layer thickness and total sediment amount are used as the supervision terms of the PINNs model for coordinated training, so as to obtain the PINNs model of the specific water-sediment coupling mechanism of the landslide dammed lake after deep training. S9. Based on the PINNs model trained by the dynamic evolution process, quantitatively invert the sediment deposition thickness and sediment yield of landslide dammed lakes in different times and spaces.

[0009] The beneficial effects of this invention are as follows: This invention assesses the sediment yield of a watershed from the sedimentary evolution process of a landslide dammed lake. First, it obtains the inundation range and water level fluctuations of the landslide dammed lake, identifying stages where water level fluctuations are greater than those of conventional seasonal floods (>3-5m). Second, it uses geophysical exploration and drilling to obtain the thickness or depth of the landslide dammed lake, and uses core scanning to obtain lacustrine sedimentary geophysical (grain size, etc.) and geochemical information. Based on the large water level fluctuations, coarser lacustrine grain size, and abrupt changes in other physicochemical information corresponding to each flood event, it identifies the history of sediment surges indicated by sand layers within the landslide dammed lake. Furthermore, based on the modern floodplain facies upstream and downstream, comparing the development of floodplain facies sand layers at the top of the landslide dammed lake indicates the beginning of the lake's disappearance, i.e., the re-establishment of fluvial facies. The development of fluvial gravel layers upstream and downstream signifies the complete disappearance of the landslide dammed lake and the stage of water-sediment balance. Based on the reservoir capacity, elevation, and stratified sediment parameters of the landslide dammed lake, the total sediment load is estimated by combining the volume difference of the stratified triangular network (TIN) with the Kriging interpolation method. The sediment deposition thickness is then inverted using a physical information neural network model (PINNs), enabling the spatiotemporal estimation and volume closure verification of the watershed sediment yield. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating the method. Figure 2 This is a flowchart illustrating the technical operation of this method. Detailed Implementation

[0011] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0012] like Figure 1 and Figure 2 As shown, the watershed sediment yield estimation method based on the dynamic evolution of the river dammed lake includes the following steps: S1. Obtain basic data on the landslide dammed lake, including the lake shore elevation line, water level, topographic data, the inundation range of the landslide dammed lake before and after the river blockage, the thickness of sediment deposition, the core particle size of the landslide dammed lake, and parameters that indicate sediment dynamics and the intensity of sediment weathering. S2. Based on the inundation range of the landslide dammed lake, the distribution of lacustrine sediments, the water level change sequence, the grain size of the landslide dammed lake core, and parameters that indicate sediment dynamics and the intensity of source weathering, the life evolution stages of the landslide dammed lake are divided, including the stable period, the retreat period, and the disappearance period. S3. Based on the inundation range and topographic data of the landslide dammed lake, calculate the area and storage capacity of the landslide dammed lake, construct the relationship curve between lake water level and storage capacity, and extract the lake basin water depth and storage capacity determined by different life evolution stages of the landslide dammed lake. S4. Based on the lake basin water depth, reservoir capacity, residual dammed lake sediment particle size and sediment deposition thickness at different life evolution stages of the landslide dammed lake, TIN models for different evolution stages after deposition were constructed using the lake shore elevation line and lacustrine core stratification as elevation control points and the original lake basin elevation data. S5. Obtain the sediment volume by the volume difference between the post-deposition and original TIN models, plot the relationship curve between lake shore elevation and sediment accumulation capacity, and coordinate the analysis with the relationship curve between lake water level and capacity to correct the remaining effective capacity and water-sediment ratio of the landslide dammed lake, and couple the analysis of the evolution process of the water-sediment curve. S6. Based on the sediment depth or sediment properties of the landslide dammed lake at different evolution stages, the lacustrine core is divided into several horizontal layers for vertical stratification. The measured dry density data of sediment in the corresponding horizontal layer is interpolated within the layer using two-dimensional kriging space interpolation to obtain the dry density planar map of each horizontal layer. S7. Based on the cumulative sediment volume at the elevation section, calculate the volume of each horizontal layer, and combine it with the corresponding dry density plane map to calculate the mass of each layer, and then obtain the total sediment volume. S8. The reservoir capacity of the landslide dammed lake, the water-sediment curve, the cumulative sediment volume at the elevation section, the dry density plane map of the horizontal layer, and the water depth and velocity field simulated by the hydrodynamic model are used together as the external boundary conditions and geometric-physical constraints of the PINNs model. The hydrological parameters and spatiotemporal coordinates of the water-sediment ratio characteristics of the landslide dammed lake at different evolution stages are used as inputs, and the core layer thickness and total sediment amount are used as the supervision terms of the PINNs model for coordinated training, so as to obtain the PINNs model of the specific water-sediment coupling mechanism of the landslide dammed lake after deep training. S9. Based on the PINNs model trained by the dynamic evolution process, quantitatively invert the sediment deposition thickness and sediment yield of landslide dammed lakes in different times and spaces.

[0013] Based on the above steps, this embodiment can be divided into five stages, namely: Phase 1: Multi-source data acquisition and testing analysis 1) Remote sensing imagery and UAV measurement: Extract the inundation range of the landslide dammed lake and the shoreline elevation line through the water surface line, obtain the time series of historical inundation range changes of the landslide dammed lake and the shoreline elevation (water level) fluctuation curve, and identify abnormal water level change signals, such as stages with water level fluctuations greater than the usual seasonal water level fluctuations (3-5m). 2) Field investigation and exploration (geophysical exploration, drilling): Obtain the distribution, water depth changes and lacustrine rock cores of the landslide dammed lake before and after the damming of the river, and draw a cross-sectional map of sediment deposition from the tail of the landslide dammed lake to the dam site; 3) Laboratory physicochemical testing and analysis: The particle size of the rock core from the landslide dammed lake was obtained using a laser particle size analyzer, including the median particle size (D50), average particle size (Mz), clay (<16μm), silt (16-63μm), and sand (>63μm) content; XRF analysis of the rock core geochemical elements and other indicators was used to calculate parameters such as Rb / Sr and Zr / Rb, which indicate sediment dynamics and source weathering intensity.

[0014] Phase Two: Lacustrine Cores and Evolutionary Stage Division: 4) Grain size threshold discrimination criteria and life evolution stage division of landslide dammed lakes based on sedimentary facies sequence: According to the watershed-scale sediment deposition dynamic process, it includes landslide dammed lake facies dominated by fine silt (<16μm), floodplain facies dominated by coarse silt and fine sand (>63μm), and fluvial facies dominated by sand and gravel. 5) Based on the abnormal water level changes and physicochemical index thresholds (D50, average grain size Mz, Rb / Sr, etc.), lacustrine core stratification (Li, i = 1 to N) was performed using a multi-index combination to divide the lacustrine lake into different stages of evolution: stable period, retreat period (with the emergence of floodplain facies), and disappearance period (with the emergence of fluvial facies). Among these, lacustrine facies sediments were formed during the stable period, while floodplain facies developed during the retreat period, and riverbed facies sediments marked the disappearance stage of the lacustrine lake. 6) Based on the evolution stages from the suspended sediment threshold (<16μm) of the landslide dammed lake to the bedload sediment threshold (>63μm), identify sediment surge depositional units in the evolution process, with a grain size threshold between 16-63μm, and determine the sediment thickness of the landslide dammed lake when "the grain size becomes significantly coarser" and "the element ratio changes abruptly".

[0015] In this stage, water level changes are identified through the time series of historical inundation range changes and water level fluctuation curves of the landslide dammed lake. The specific method for dividing the life evolution stages of the landslide dammed lake includes the following steps: S2-1. Determine whether the water level change of the landslide dammed lake exceeds the fluctuation threshold. If so, proceed to step S2-2; otherwise, proceed to step S2-4. S2-2. Determine whether the core grain size of the landslide dammed lake is between the first grain size threshold and the second grain size threshold. If so, proceed to step S2-3; otherwise, determine that the landslide dammed lake is a conventional lacustrine deposit. The first grain size threshold is less than the second grain size threshold. S2-3. Determine whether there are abrupt changes in parameters that indicate sediment dynamics and source weathering intensity. If so, determine that the landslide dammed lake is a sediment surge flood deposit layer; otherwise, determine that the landslide dammed lake is a conventional lacustrine deposit. S2-4. Determine whether the core grain size of the landslide dammed lake is smaller than the first grain size threshold. If so, determine that the landslide dammed lake is a conventional lacustrine deposit; otherwise, determine that the landslide dammed lake is a fluvial deposit.

[0016] The amplitude threshold ranges from 3 to 5 μm; the first particle size threshold and the second particle size threshold are 16 μm and 63 μm, respectively.

[0017] Phase Three: Calculation of the Reservoir Capacity and Sediment Volume of the Landslide Dam: 7) Based on the elevation changes of the landslide dammed lake shore (ΔH) W Using topographic data, the Spatial Analyst module in GIS software is used to calculate the inundation area of ​​the landslide dammed lake (A). w Input the area and topographic data of the landslide dammed lake into the 3D Analyst module of the GIS software to calculate the volume of the landslide dammed lake, i.e., its reservoir capacity (V). w Construct the lake water level (ΔH) W ) and storage capacity (V) w The relationship curve; it should be noted that the inundation range of a landslide dammed lake focuses on its geometric location and spatial distribution, and the area can only be obtained through calculation. The planar distribution of water and sediment defines the range, and the water and sediment area is calculated from this range; 8) By integrating the original lake basin elevation data before the formation of the landslide dam ( Using the lake shore elevation line as a topographic benchmark, the topographic elevation line is introduced. Using a unified closed elevation constraint boundary, and combining sedimentary strata and thickness information obtained from lacustrine core drilling, the sedimentary thickness at different stages of the landslide dammed lake's life evolution was used as a constraint condition. Based on this, for each evolution stage, a triangulation algorithm was used, with the original lake basin elevation data (…) as the constraint condition. ) and the corresponding sedimentary thickness inversion elevation ( ) as the vertex, at the lake shore elevation line ( Under constraints, post-depositional triangular network (TIN) models for different time stages were constructed. The sediment volume was calculated using the volume difference method. For equidistant layered sections ( It can calculate the volume difference above and below the elevation, and obtain the cumulative sediment volume at that elevation section. The calculated cumulative sediment deposition volume ( Using as the x-axis, plot the curve showing the relationship between lake elevation and sediment accumulation capacity. ).

[0018] Based on the reduction of lake area and reservoir capacity during the evolution of the landslide dammed lake, and the continuous siltation and exposure of sediment, a basis can be provided for subsequent estimation of sediment reservoir capacity at different stages.

[0019] Phase Four: Estimate of Sediment Yield 9) Considering the vertical consolidation variations and planar spatial non-uniformity, calculate the total sediment yield based on vertical stratification and intra-layer interpolation data. Based on sediment depth or sediment properties, divide the entire sedimentary body into k horizontal layers for vertical stratification. Perform two-dimensional kriging spatial interpolation on the measured dry density data of sediment within the corresponding horizontal layers to obtain a stratified dry density planar map. The total sediment load is calculated using an overlay algorithm, based on the previously calculated (…). ) Calculate the volume of each floor ( ), and then estimate the mass of each layer. Finally, the total amount of sediment was calculated. .

[0020] Phase 5: Model Coupling and Validation Analysis Based on sediment parameters and hydrodynamic models, the sediment transport equations and hydrodynamic boundaries are "embedded" into a Physical Information Neural Network (PINNs) model. This enables the temporal-spatial inversion of the sediment thickness field that satisfies physical laws, achieving volumetric closure verification of the estimated sediment yield. The PINNs model takes time (t) and spatial coordinates (x, y) as inputs; the lake volume (V) is also considered. w ), the cumulative sediment volume at the sediment stratification section ( Spatial distribution map of dry density of each layer of sediment ( The water depth and velocity field simulated by the hydrodynamic model serve as both external boundary conditions and geometric-physical constraints; total sediment load ( M This term serves as a monitoring term to constrain overall mass closure. It is applied through the sediment continuity equation based on the principle of mass conservation. Calculate the residuals of the PINNs model and define the loss function. The PINNs model is trained, and during the training process, the solution that satisfies mass conservation is continuously approximated, thereby inverting the spatiotemporal distribution of sediment deposition thickness. ).in The loss function; The error between the predictions of the PINNs model and the actual observed data; For model residuals; For weights.

[0021] This invention overcomes the limitations of hydrological observation data, which are typically limited to only a few decades and where sediment yield observations are insufficient in most landslide-dammed lake basins. It provides a reliable and independent technical approach for estimating sediment yield in landslide-dammed lakes in data-scarce areas. This method can quantitatively assess the long-term evolution of erosion and sediment yield in post-earthquake landslide-dammed lake basins, the sediment yield in disaster chain basins such as glacial till dams and glacial lakes under climate change, and is also applicable to estimating extreme sediment yields in various disaster chains, including the sedimentary history of ancient landslide-dammed lakes.

[0022] In summary, this invention uses landslide-dammed lakes as a continuous monitoring medium for watershed sediment yield. By identifying event layers in the residual landslide-dammed lake sedimentary sequence and comparing and analyzing key physicochemical parameters, a method for estimating watershed sediment yield during extreme surface processes of the disaster chain is constructed. Through comprehensive analysis of water level changes (remote sensing-UAV measurement), hydrodynamic conditions (grain size index), and sediment source changes (geochemical element index), different sediment yield stages are delineated from the stable, retreating, and disappearing phases of the landslide-dammed lake. The history of sediment changes from lacustrine suspended sediment to floodplain bedload is analyzed, constructing a complete life-cycle evolution model of the landslide-dammed lake. The proportion of sediment in the lacustrine, floodplain, and riverbed facies during its dynamic evolution is assessed. A layered triangular irregular network (TIN) model is constructed to obtain layered cumulative sediment volume data. Based on sedimentary stratification and spatial interpolation of the dry density field, the layered volume and mass are calculated to estimate the total sediment yield. A physical information neural network model inverts the sedimentary thickness field and achieves sediment yield estimation, quantifying the non-stationary dynamic evolution characteristics of water and sediment in landslide-dammed lakes.

Claims

1. A watershed sediment yield estimation method based on the dynamic evolution of river dammed lakes, characterized in that, Includes the following steps: S1. Obtain basic data on the landslide dammed lake, including the lake shore elevation line, water level, topographic data, the inundation range of the landslide dammed lake before and after the river blockage, the thickness of sediment deposition, the core particle size of the landslide dammed lake, and parameters that indicate sediment dynamics and the intensity of sediment weathering. S2. Based on the inundation range of the landslide dammed lake, the distribution of lacustrine sediments, the water level change sequence, the grain size of the landslide dammed lake core, and parameters that indicate sediment dynamics and the intensity of source weathering, the life evolution stages of the landslide dammed lake are divided, including the stable period, the retreat period, and the disappearance period. S3. Based on the inundation range and topographic data of the landslide dammed lake, calculate the area and storage capacity of the landslide dammed lake, construct the relationship curve between lake water level and storage capacity, and extract the lake basin water depth and storage capacity determined by different life evolution stages of the landslide dammed lake. S4. Based on the lake basin water depth, reservoir capacity, residual dammed lake sediment particle size and sediment deposition thickness at different life evolution stages of the landslide dammed lake, TIN models for different evolution stages after deposition were constructed using the lake shore elevation line and lacustrine core stratification as elevation control points and the original lake basin elevation data. S5. Obtain the sediment volume by the volume difference between the post-deposition and original TIN models, plot the relationship curve between lake shore elevation and sediment accumulation capacity, and coordinate the analysis with the relationship curve between lake water level and capacity to correct the remaining effective capacity and water-sediment ratio of the landslide dammed lake, and couple the analysis of the evolution process of the water-sediment curve. S6. Based on the sediment depth or sediment properties of the landslide dammed lake at different evolution stages, the lacustrine core is divided into several horizontal layers for vertical stratification. The measured dry density data of sediment in the corresponding horizontal layer is interpolated within the layer using two-dimensional kriging space interpolation to obtain the dry density planar map of each horizontal layer. S7. Based on the cumulative sediment volume at the elevation section, calculate the volume of each horizontal layer, and combine it with the corresponding dry density plane map to calculate the mass of each layer, and then obtain the total sediment volume. S8. The reservoir capacity of the landslide dammed lake, the water-sediment curve, the cumulative sediment volume at the elevation section, the dry density plane map of the horizontal layer, and the water depth and velocity field simulated by the hydrodynamic model are used together as the external boundary conditions and geometric-physical constraints of the PINNs model. The hydrological parameters and spatiotemporal coordinates of the water-sediment ratio characteristics of the landslide dammed lake at different evolution stages are used as inputs, and the core layer thickness and total sediment amount are used as the supervision terms of the PINNs model for coordinated training, so as to obtain the PINNs model of the specific water-sediment coupling mechanism of the landslide dammed lake after deep training. S9. Based on the PINNs model trained by the dynamic evolution process, quantitatively invert the sediment deposition thickness and sediment yield of landslide dammed lakes in different times and spaces.

2. The watershed sediment yield estimation method based on the dynamic evolution of the river dammed lake as described in claim 1, characterized in that, In step S1, the lake shore elevation line, water level, and topographic data are obtained through remote sensing imagery and / or drone measurements; the inundation range and sediment thickness of the landslide dammed lake before and after the river blockage are obtained through field surveys and explorations; the core grain size of the landslide dammed lake and parameters indicating sediment dynamics and source weathering intensity are obtained through physicochemical testing and analysis, including the median core grain size; parameters indicating sediment dynamics and source weathering intensity include Rb / Sr and Zr / Rb.

3. The watershed sediment yield estimation method based on the dynamic evolution of the river dammed lake as described in claim 2, characterized in that, The water level changes in step S2 are identified by the time series of historical inundation range changes of the landslide dam and the water level fluctuation curve.

4. The watershed sediment yield estimation method based on the dynamic evolution of landslide dammed lakes according to claim 3, characterized in that, The specific method for dividing the life evolution stages of the landslide dammed lake in step S2 includes the following steps: S2-1. Determine whether the water level change of the landslide dammed lake exceeds the fluctuation threshold. If so, proceed to step S2-2; otherwise, proceed to step S2-4. S2-2. Determine whether the core grain size of the landslide dammed lake is between the first grain size threshold and the second grain size threshold. If so, proceed to step S2-3; otherwise, determine that the landslide dammed lake is a conventional lacustrine deposit. The first grain size threshold is less than the second grain size threshold. S2-3. Determine whether there are abrupt changes in parameters that indicate sediment dynamics and source weathering intensity. If so, determine that the landslide dammed lake is a sediment surge flood deposit layer; otherwise, determine that the landslide dammed lake is a conventional lacustrine deposit. S2-4. Determine whether the core grain size of the landslide dammed lake is smaller than the first grain size threshold. If so, determine that the landslide dammed lake is a conventional lacustrine deposit; otherwise, determine that the landslide dammed lake is a fluvial deposit.

5. The watershed sediment yield estimation method based on the dynamic evolution of the river dammed lake as described in claim 4, characterized in that, The amplitude threshold ranges from 3 to 5 m; the first and second particle size thresholds are 16 μm and 63 μm, respectively.

6. The watershed sediment yield estimation method based on the dynamic evolution of the river dammed lake as described in claim 4, characterized in that, In step S3, the inundation area of ​​the landslide dammed lake was calculated using the Spatial Analyst module of the GIS software; the reservoir capacity of the landslide dammed lake was calculated using the 3D Analyst module of the GIS software.

7. The watershed sediment yield estimation method based on the dynamic evolution of landslide dammed lakes according to claim 6, characterized in that, Specific methods for obtaining sediment volume by the volume difference between the post-depositional and original TIN models include: for equidistant stratified sections, obtaining the cumulative sediment volume at the elevation section by calculating the volume difference above and below the elevation.

8. The watershed sediment yield estimation method based on the dynamic evolution of landslide dammed lakes according to claim 7, characterized in that, In step S8, the mass of each layer is equal to the product of the volume of the corresponding horizontal layer and the dry density plane diagram.

9. The watershed sediment yield estimation method based on the dynamic evolution of river dammed lakes according to claim 1, characterized in that, The PINNs model embeds physical mechanisms representing the dynamic evolution process during training. It captures the water and sediment balance state of the landslide dammed lake in real time through the sediment continuity equation based on the principle of mass conservation and calculates the model residuals. Based on the error between the prediction results of the PINNs model and the actual observation data and the model residuals, a loss function is constructed to iteratively train the PINNs model.

10. The watershed sediment yield estimation method based on the dynamic evolution process of a landslide dammed lake as described in claim 9, characterized in that, The expression for the loss function is: ; in The loss function; The error between the predictions of the PINNs model and the actual observed data; For model residuals; For weights.

Images