A geological disaster full-spectrum remote sensing response simulation method coupling a surface process model and a three-dimensional radiation transfer model

Abstract

The application provides a geological disaster full-spectrum remote sensing response simulation method coupling a surface process model and a three-dimensional radiation transfer model, and belongs to the technical field of remote sensing and disaster simulation. The method comprises the following steps: firstly, acquiring a digital elevation model, disaster elements and prior knowledge; secondly, constructing a terrain grid and registering, combining with basin analysis to calculate disaster frequency and erosion deposition, and generating a surface change DEM before and after disaster occurrence; then inputting the change DEM into a three-dimensional radiation transfer model, constructing a multi-temporal scene and inputting component spectral parameters, carrying out full-process simulation, and generating a remote sensing reflectivity image covering visible light to microwave full spectrum; finally, verifying the accuracy by using real data and outputting the result. The application realizes physical coupling of surface dynamics and radiation transfer processes, can effectively simulate the comprehensive remote sensing signals generated by the surface morphology and structure changes caused by disasters, and provides high-fidelity, multi-temporal data support for disaster mechanism research and monitoring and early warning.

Description

Technical Field

[0001] This invention relates to the field of remote sensing and disaster simulation technology, specifically a method for simulating the full spectrum of remote sensing response to geological disasters by coupling a surface process model with a three-dimensional radiative transfer model. Background Technology

[0002] Geological disasters, especially landslides, debris flows, and collapses, are common natural disasters worldwide. They occur frequently and are highly destructive, posing a serious threat to people's lives and property, infrastructure operation, and the ecological environment. Remote sensing technology, with its unique advantages of macroscopic, rapid, non-contact, and periodic observation, has become an indispensable core technology for geological disaster investigation, hazard identification, emergency monitoring, and post-disaster assessment. By analyzing the changes in spectral, texture, and morphological characteristics of remote sensing images before and after a disaster, key information such as the disaster's extent and impact can be quickly obtained. However, the acquisition of realistic remote sensing images is often limited by revisit cycles, weather conditions (such as cloud cover), and sensor performance, especially the near inability to obtain instantaneous image data during a disaster. Therefore, developing simulation technology for remote sensing responses to geological disasters and generating highly realistic simulation data has significant theoretical and practical value for deepening the understanding of disaster mechanisms, optimizing monitoring and early warning algorithms, and conducting emergency response drills.

[0003] Existing research on geological hazard remote sensing mainly focuses on hazard identification and monitoring from single data sources such as optical remote sensing, radar remote sensing, and lidar, as well as the independent application of surface process simulation in hydrology, erosion, and landslide evolution. Surface process models can accurately describe topographic relief changes, material erosion and deposition processes, and reveal the evolution of surface morphology and structure before and after a hazard. Three-dimensional radiative transfer models can simulate the electromagnetic wave reflection and scattering characteristics of different surface material components and geometries under multispectral and multi-angle conditions, providing an important tool for the physical inversion of remote sensing images. However, existing technologies generally have the following shortcomings: (1) Model fragmentation: Surface process models and three-dimensional radiation transfer models are mostly operated independently, lacking effective coupling in terms of time, space and physical parameters, making it difficult to comprehensively reflect the entire process of geological disasters from morphological changes to radiation characteristic changes; (2) Limited spectral range: Most radiation transfer simulations are concentrated in the visible light or near-infrared bands, lacking the unified simulation capability covering the entire spectrum from visible light to near-infrared to microwave, which limits the effect of multi-sensor and multi-band fusion analysis; (3) Insufficient dynamic process: Existing methods are mostly based on single-phase data or static scenes for radiation transfer simulation, which cannot truly reproduce the dynamic changes before and after the disaster; (4) Lack of systematic accuracy verification: Most simulation results lack quantitative comparison and accuracy evaluation with real multi-source remote sensing data, resulting in insufficient reliability of the model in disaster monitoring and prediction. Therefore, there is an urgent need to propose a method that can organically couple surface processes with three-dimensional radiative transfer at the physical mechanism level, supporting full-spectrum, multi-temporal remote sensing response simulation of geological disasters. This method can not only depict the topographic and material changes during the disaster occurrence process, but also simulate its electromagnetic wave response characteristics in different bands, thereby providing more accurate and systematic technical support for disaster mechanism research, monitoring and early warning, and emergency assessment. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method for simulating the full-spectrum remote sensing response of geological hazards by coupling a surface process model with a three-dimensional radiative transfer model, comprising the following steps:

[0005] (1) Obtain digital elevation model (DEM) data of the monitoring area;

[0006] (2) Collect the constituent elements of the disaster scene, including the types and spatial distribution of elements such as soil, bedrock, vegetation, water bodies, and buildings in the disaster area;

[0007] (3) Collect prior knowledge about disasters, including the following: the spatial location, scope, type, and extent of the disaster, historical observation records, and the reflectance and texture features of the spectrum and remote sensing images of similar disasters that have been collected;

[0008] (4) Using DEM data as the main body for terrain calculation, construct a terrain grid;

[0009] (5) Georegister the terrain grid and define the resolution of spatial reference and resampling;

[0010] (6) Based on the surface process model, watershed analysis and visualization are carried out on a regular terrain grid to explicitly simulate the surface processes and geomorphological evolution caused by disasters such as landslides and debris flows, including changes in bedrock height and sediment thickness. Among them, the surface process calculation follows the mass balance equation, and the surface elevation is regarded as the sum of the bedrock surface elevation and the thickness of the movable loose layer. In each iteration, the grid unit elevation is updated based on the sedimentary rock conservation equation. The river erosion, sediment transport and deposition processes are solved using the SPACE model, combined with the calculation formulas of bedrock erosion rate, sediment entrainment rate and deposition flux. The soil formation process is simulated based on the exponential decay function of local thickness, and soil creep is modeled using nonlinear thickness-related transport law.

[0011] (7) Calculate the frequency of disaster occurrence—calculate the temporal and spatial probabilities based on the HyLands model, and combine them to obtain the instability probability of each grid;

[0012] (8) Calculate the disaster area distribution of landslides—determined by random number analysis and slope angle threshold conditions;

[0013] (9) Calculate the erosion and deposition amounts. The erosion and deposition amounts are calculated by the bedrock erosion and sediment entrainment rate formulas after the landslide is triggered, respectively. The sediment redistribution process is simulated based on the multi-flow direction or D8 algorithm, taking into account special cases such as landslide dam lake deposition.

[0014] (10) The key output of the surface process simulation is the surface change DEM before and after the disaster.

[0015] (11) The above-mentioned multi-temporal surface change DEM inputs a three-dimensional radiative transfer model (such as LESS or RAPID) to construct a multi-temporal three-dimensional scene. The scene includes several surface components such as soil, bedrock, vegetation, water bodies, buildings and disaster deposits, and retains spatial distribution information.

[0016] (12) Input the reflectance, scattering and other spectral parameters of each component in the visible light, near infrared and microwave bands.

[0017] (13) Combine sensor characteristics, observation geometry, atmospheric conditions and topographic parameters to carry out multi-temporal, full-spectrum radiative transfer simulation.

[0018] (14) The key output of the three-dimensional radiation transfer simulation is the remote sensing reflectance simulation image before and after the disaster.

[0019] (15) Using historical observation data or the reflectance of real remote sensing images, the accuracy of the simulation image is verified at multiple points, and quantitative indicators such as root mean square error (RMSE) and correlation coefficient (R) are calculated. The simulation results of the full spectrum of remote sensing response to geological disasters are then output.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0021] (1) At the physical mechanism level, the surface process model and the three-dimensional radiation transfer model are organically coupled, taking into account the changes in topography and material and the changes in electromagnetic radiation characteristics.

[0022] (2) Supports multi-temporal simulation covering the entire spectrum from visible light to near-infrared to microwave, meeting the needs of multi-source remote sensing data fusion and comparative analysis;

[0023] (3) It can fully reproduce the dynamic changes before and after the occurrence of geological disasters, providing high-precision and systematic technical support and data foundation for disaster monitoring, mechanism research, risk assessment and emergency decision-making. Attached Figure Description

[0024] To more clearly illustrate the technical advantages of the present invention, the following is a brief introduction to the present invention and its application examples with accompanying drawings.

[0025] Figure 1 This is a schematic diagram of the process of the present invention;

[0026] Figure 2 Example 1: High-resolution Google imagery of the Jinsha River high-altitude landslide;

[0027] Figure 3 The terrain grid after georegistration and resampling;

[0028] Figure 4 This is a watershed analysis diagram based on the Quinn algorithm;

[0029] Figure 5 The difference map of DEM changes caused by the landslide shows the erosion and subsidence of the landslide;

[0030] Figure 6 Visualization results of east-west and north-south elevation profiles before and after the landslide;

[0031] Figure 7 Importing the post-landslide DEM into a LESS model for 3D display;

[0032] Figure 8 Image of Sentinel 2 before the landslide;

[0033] Figure 9 The simulated reflectance image of LESS before the landslide;

[0034] Figure 10 Images of Sentinel 2 after the landslide;

[0035] Figure 11 Simulated reflectance image of LESS after landslide;

[0036] Figure 12 The image is a false-color composite image (NIR, R, G) of Sentinel 2 after the landslide;

[0037] Figure 13 The image shows a simulated false-color composite image (NIR, R, G) of LESS after the landslide.

[0038] Figure 14 For the verification of the accuracy of the red light band after the landslide;

[0039] Figure 15 For the accuracy verification of the green light band after the landslide;

[0040] Figure 16 For accuracy verification of the blue light band after the landslide;

[0041] Figure 17 For verification of near-infrared accuracy after landslide;

[0042] Figure 18 Example 2: High-resolution Google imagery of a landslide in Zayu County, Nyingchi City, Tibet.

[0043] Figure 19 The terrain grid after georegistration and resampling in Example 2;

[0044] Figure 20 This is a difference map of the DEM changes caused by the collapse in Example 2;

[0045] Figure 21 Visualization results of the east-west and north-south elevation profiles before and after the collapse in Example 2;

[0046] Figure 22 The true reflectance of the Sentinel-2 image in Example 2;

[0047] Figure 23 The LESS simulated reflectance image for Example 2;

[0048] Figure 24 Example 3: High-resolution Google imagery of the debris flow in Meilonggou, Banshanmen Town, Danba County, Sichuan Province;

[0049] Figure 25 The terrain grid after georegistration and resampling in Example 3;

[0050] Figure 26 This is a difference map of the DEM changes caused by debris flow in Example 3;

[0051] Figure 27Visualization results of east-west and north-south elevation profiles before and after the debris flow in Example 3;

[0052] Figure 28 The true reflectance of the Sentinel-2 image in Example 3;

[0053] Figure 29 The image shows the simulated reflectance of LESS for Example 3. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0055] Example 1

[0056] The specific implementation steps of the geological hazard full-spectrum remote sensing response simulation method of the present invention, which couples a surface process model with a three-dimensional radiative transfer model, in a landslide example are as follows:

[0057] (1) Obtain digital elevation model (DEM) data of the monitoring area from the OpenTopography open source interface, the product being SRTM GL1 (30m);

[0058] (2) Collect the constituent elements of the disaster scene, including the types and spatial distribution of elements such as soil, bedrock, vegetation, water body, and buildings in the disaster site. The surface of the Jinsha River high-level landslide disaster site is mostly severely weathered thin layer of mountain soil, underlain by the Proterozoic Erchang granite bedrock. The vegetation is mainly alpine shrubs and sparse grasslands. It is close to the Jinsha River water body. The buildings are mainly a small number of mountain rural houses.

[0059] (3) Collect prior knowledge about the disaster, including the following: The coordinates of the Jinsha River landslide are 98°42′29″E and 31°05′02″N. The landslide development area belongs to the high mountain landform of the Qinghai-Tibet Plateau. The exposed strata are Proterozoic Erchang granite, with northwest-trending fault structures. The geological conditions are complex, and the shallow surface is severely weathered, which is conducive to the development and formation of landslides. The top of the landslide is at a high altitude and has a large slope, which is a typical high-altitude landslide geological disaster. Satellite remote sensing data and textures are from the Sentinel official website and Google Earth. Spectral data are from the specchio open-source interface of the University of Zurich. High-resolution images of the landslide can be found in Google imagery, such as... Figure 2 As shown;

[0060] (4) Using DEM data as the main body for terrain calculation, construct a terrain grid;

[0061] (5) Georegister the terrain grid, define the spatial reference as the WGS84 coordinate system and the resampling resolution as 30m, and the visualization effect is as follows: Figure 3 As shown;

[0062] (6) Based on the surface process model, watershed analysis and visualization are carried out on regular terrain grids to explicitly simulate the surface processes and geomorphological evolution caused by disasters such as landslides and debris flows, including changes in bedrock height and sediment thickness;

[0063] (7) Calculate the frequency of disaster occurrence—calculate the temporal and spatial probabilities based on the HyLands model, and combine them to obtain the instability probability of each grid;

[0064] (8) Calculate the disaster area distribution of landslides—determined by random number analysis and slope angle threshold conditions;

[0065] (9) Calculate the erosion and deposition rates. The erosion and deposition rates are calculated using the formulas for bedrock erosion and sediment entrainment rates after landslide triggering, respectively. The sediment redistribution process is simulated based on the multi-flow direction Quinn or D8 algorithm, taking into account special cases such as landslide-dammed lake deposition. Among them, the idea of ​​the multi-flow direction algorithm Quinn is to distribute the flow rate of a unit to multiple directions according to the weight ratio related to the slope of adjacent units (e.g., Figure 4 The idea behind the D8 algorithm is that the flow direction of each grid cell points to the cell with the largest slope among its eight neighboring cells (Moore neighborhood).

[0066] The multi-flow algorithm MFD, Quinn, has the following formula:

[0067] Calculate the slope of each adjacent cell:

[0068] (18)

[0069] Weight calculation:

[0070] (19)

[0071] Normalization yields the flow ratio:

[0072] (20)

[0073] Traffic allocation:

[0074] (twenty one)

[0075] in: This represents the total flow rate of the unit. Is flowing to the first The ratio of neighbors satisfies ;for (Not downhill direction) .

[0076] The D8 (Deterministic 8) algorithm has the following formula:

[0077] Assume the current unit elevation is The elevation of adjacent units is The distance is (direct neighbors are) diagonal neighbors ):

[0078] (twenty two)

[0079] Select the direction of maximum gradient:

[0080] (twenty three)

[0081] Traffic allocation:

[0082] (twenty four)

[0083] in: This refers to the total inflow into this unit (including rainfall and upstream runoff).

[0084] (10) The key output of surface process simulation is the surface change DEM before and after the disaster (e.g., Figure 5 The differential DEM), in which the east-west and north-south landslide elevation profiles are the most intuitive landslide display results (e.g., Figure 6 ).

[0085] (11) The above-mentioned multi-temporal surface change DEM input three-dimensional radiative transfer model is as follows: Figure 7 A multi-temporal 3D scene is constructed, which includes several surface components such as soil, bedrock, vegetation, water bodies, buildings and disaster deposits, and retains spatial distribution information;

[0086] (12) Input the reflectance, scattering and other spectral parameters of each component in the visible light, near infrared, microwave and other bands, which include the spectra of the bedrock and soil complex before and after the landslide, and the spectra of vegetation leaves and branches.

[0087] (13) This example uses Sentinel-2 real remote sensing images as a reference, such as Figure 8 True-color composite of Sentinel 2 images before the landslide, Figure 10 True-color composite of Sentinel 2 images after the landslide. Figure 12 Based on the false-color composite images (NIR, R, G) of Sentinel 2 after the landslide, multi-temporal, full-spectrum radiative transfer simulations were carried out, taking into account sensor characteristics, observation geometry, atmospheric conditions, and topographic parameters.

[0088] (14) The key output of the three-dimensional radiative transfer simulation is the simulated remote sensing reflectance images before and after the disaster, such as Figure 9 True-color composite image of LESS before landslide, reflecting simulated reflectance. Figure 11 True-color composite image of LESS simulated reflectance after landslide. Figure 13 The simulated false-color composite image (NIR, R, G) of LESS after the landslide;

[0089] (15) Using the reflectance of each band of the Sentinel-2 real remote sensing image, multiple points were randomly generated to verify the accuracy of the reflectance of the LESS simulated image. Quantitative indicators such as root mean square error (RMSE) and correlation coefficient (R) were calculated, and the simulation results of the full-spectrum remote sensing response to geological disasters were output (e.g., Figure 14 For the accuracy verification of the red band after the landslide, Figure 15 For the accuracy verification of the green band after the landslide, Figure 16 For the accuracy verification of the blue light band after the landslide, Figure 17 This is for the verification of near-infrared accuracy after a landslide.

[0090] Example 2

[0091] The implementation results of the full-spectrum remote sensing response simulation method for geological hazards based on a coupled surface process model and a three-dimensional radiative transfer model of the present invention, in a landslide embodiment, are as follows:

[0092] (1) The disaster occurred in Zayu County, Nyingchi City, Tibet Autonomous Region (upper reaches of the Nujiang River), at 105°28'E, 26°30'N, with an altitude of 1902m. Digital elevation model (DEM) data of the monitored area was obtained from ALOS PALSAR products (12.5m resolution). Satellite remote sensing data and textures were obtained from the Sentinel official website and Google Earth. Spectral data were obtained from the specchio open-source interface of the University of Zurich. High-resolution images of the landslide can be found in Google imagery, such as... Figure 18 As shown;

[0093] (2) Using DEM data as the main body for terrain calculation, a terrain grid is constructed; the terrain grid is georeferenced, the spatial reference is defined as the WGS84 coordinate system and the resampling resolution is 12.5m, and the visualization effect is as follows: Figure 19 As shown;

[0094] (3) Based on the surface process model, watershed analysis and visualization are carried out on a regular terrain grid to explicitly simulate the surface processes and geomorphological evolution caused by disasters such as landslides, and to calculate erosion and deposition. The key output of the surface process simulation is the surface change DEM before and after the disaster (e.g., Figure 20The differential DEM), in which the east-west and north-south elevation profiles are relatively intuitive displays of the collapse results (e.g. Figure 21 );

[0095] (4) The above-mentioned multi-temporal surface change DEM input three-dimensional radiative transfer model, input the spectral parameters of each component, and combine sensor characteristics, observation geometry, atmospheric conditions and topographic parameters to carry out multi-temporal, full-spectrum radiative transfer simulation. This example uses Sentinel-2 real remote sensing image as a reference (e.g. Figure 22 ), and by comparison, it was found that the simulated images (such as Figure 23 The effect is good.

[0096] Example 3

[0097] The implementation results of the full-spectrum remote sensing response simulation method for geological hazards, which couples a surface process model with a three-dimensional radiative transfer model, in a debris flow example are as follows:

[0098] (1) The debris flow disaster occurred in Meilonggou, Banshanmen Town, Danba County, Sichuan Province, at 102°01'E, 30°97'N, with an altitude of 2047m. Digital elevation model (DEM) data of the monitored area was obtained from ALOS PALSAR products (12.5m resolution). Satellite remote sensing data and textures were obtained from the Sentinel official website and Google Earth. Spectral data were obtained from the specchio open-source interface of the University of Zurich. High-resolution images of the debris flow can be found in Google imagery, such as... Figure 24 As shown;

[0099] (2) Using DEM data as the main body for terrain calculation, a terrain grid is constructed; the terrain grid is georeferenced, the spatial reference is defined as the WGS84 coordinate system and the resampling resolution is 12.5m, and the visualization effect is as follows: Figure 25 As shown;

[0100] (3) Based on the surface process model, watershed analysis and visualization are carried out on a regular terrain grid to explicitly simulate the surface processes and geomorphological evolution caused by disasters such as debris flows, and to calculate erosion and deposition. The key output of the surface process simulation is the surface change DEM before and after the disaster (e.g., Figure 26 The differential DEM), in which the east-west and north-south elevation profiles are the most intuitive indicators of debris flow (e.g., Figure 27 );

[0101] (4) The above-mentioned multi-temporal surface change DEM input three-dimensional radiative transfer model, input the spectral parameters of each component, and combine sensor characteristics, observation geometry, atmospheric conditions and topographic parameters to carry out multi-temporal, full-spectrum radiative transfer simulation. This example uses Sentinel-2 real remote sensing image as a reference (e.g. Figure 28 ), and by comparison, it was found that the simulated images (such as Figure 29 The effect is good.

Claims

1. A method for simulating the full-spectrum remote sensing response of geological hazards by coupling a surface process model and a three-dimensional radiative transfer model, characterized in that, Includes the following steps: (1) Data collection steps: acquire digital elevation model (DEM), disaster scenario components and prior disaster knowledge; (2) Surface process simulation steps: Based on the data, a terrain grid is constructed and georegistered, watershed analysis and visualization are carried out, the frequency of disaster occurrence, the distribution of disaster area, and the amount of erosion and deposition are calculated, and a surface change DEM is generated before and after the disaster. The calculation of the frequency of disaster occurrence, the distribution of disaster area, the amount of erosion and deposition is as follows: a. Frequency of disaster occurrence Landslide triggering is used to predict the number of landslides within a given time span, given prior knowledge and known slope stability. It combines the temporal and spatial probabilities of events triggering landslide activity to assess the probability of landslide occurrence. In the HyLands model, the instability probability... Defined as: （1） This represents the probability of a landslide event occurring within a given time interval. This represents the probability of instability occurring at a specific location under conditions where a potential triggering event occurs. It calculates the probability that each grid cell in the model domain will transform into a critical sliding node; the area and volume of the landslide are not statically set, but are derived dynamically from the model. Regarding temporal probability, a Poisson model is used to predict the temporal probability of landslides occurring, with the goal of deriving the probability within a time interval. The probability of at least one landslide occurring, assuming the landslide is a random point event, on a conditionally unstable slope, within a time interval. Internal occurrence The probability formula for a landslide is as follows: （2） in The number of landslides. This represents the average occurrence rate of landslides. It is the average recurrence interval of the triggering event. The reciprocal of; obtained from equation (2), in time The probability that there is no landslide is: （3） The probability of at least one landslide occurring is: （4） In the numerical implementation of the model, equation (4) is used to calculate the probability of the triggered event occurring within each time step; Regarding spatial probability, according to Culmann's theory, the spatial probability of instability depends on the slope height. With maximum stable slope height The ratio: （5） in, This represents the probability of spatial instability. Local slope height represents the elevation difference between this grid cell and its highest neighbor. The maximum stable slope height is calculated using the following formula: （6） in, For cohesion, For rock density, It is the acceleration due to gravity. For local terrain slope angle, It is the internal friction angle of the material; Therefore, the comprehensive probability is calculated by substituting equations (4) and (5) into equation (1) to obtain the probability at a given position. time interval General formula for internal instability: （7） b. Distribution of disaster area The essence of disaster area distribution is to determine whether a landslide has occurred in the grid cell; in each model iteration, equation (1) will be applied to all terrain slope angles. greater than the material's critical friction angle For each candidate raster, the cell update generates a pseudo-random number uniformly distributed in the [0,1] range. ,like If a landslide occurs in a unit, the spatial range of that unit will be included in the landslide disaster area. Landslides cannot overlap. Within a time step, each node can only belong to a unique landslide event. c. Calculation of erosion and sedimentation Calculation of erosion volume – Sediment erosion caused by gravity sliding and bedrock erosion As a stochastic event process, the volume of material removed by the landslide when the triggering event occurs is calculated based on the Culmann stability model, assuming the material satisfies the Mohr-Coulomb condition, and its failure surface dip angle is... Take as: （8） When a landslide initiates at a critical node, the rupture surface propagates towards the surface until it intersects with the terrain. The propagation is simulated using a recursive algorithm: starting from the critical node, if the slope angle between any neighboring element and that node... greater than the tilt angle This reduces the elevation of the adjacent unit, making The element is then marked as a landslide node, and its neighbors are iterated until no element in a series of consecutive landslide nodes satisfies the condition. Up to this point; the landslide area and volume are obtained by summing the number of landslide-affected units and the erosion volume of each unit, respectively; Calculation of sediment volume—The redistribution of landslide sediments on the slope surface according to a nonlinear, nonlocal deposition scheme: （9） in, Deposition flux per unit area This represents the volumetric flow rate of landslide sediments. The transportation distance is calculated using the following formula: （10） In the formula, For slope, As a reference length scale, it indicates that a length is taken on the horizontal plane. , The critical slope gradient is taken as . Sediment redistribution uses a multi-flow-direction algorithm or a Deterministic 8 algorithm to distribute it to downstream units according to the slope ratio; once sediment flows into the dam area formed by the landslide, it will continue to deposit until the depression is filled or the sediment flow is exhausted. (3) Three-dimensional radiation transfer simulation steps: Input the surface change DEM into the three-dimensional radiation transfer model, construct a multi-temporal three-dimensional scene, input the surface component spectral parameters, carry out multi-temporal, full-spectrum radiation transfer simulation, and generate remote sensing reflectance simulation images before and after the disaster. (4) Accuracy verification and result output steps: The accuracy of the simulation image is verified by using historical observation data or real remote sensing image reflectance, and the simulation results of the full spectrum of remote sensing response to geological disasters are output.

2. The method according to claim 1, characterized in that, The elements constituting the disaster scenario in step (1) include: the soil, bedrock, vegetation, water bodies, and the type and spatial distribution of buildings in the disaster-stricken area.

3. The method according to claim 1, characterized in that, The prior knowledge of disasters in step (1) includes the following items: the spatial location, scope, type, disaster status, historical observation records, the spectrum of similar disasters that have been collected, and the reflectance and texture features of remote sensing images.

4. The method according to claim 1, characterized in that, In step (2), the construction of the terrain grid uses the DEM as the main body for terrain calculation, and georeferencing mainly defines the resolution of spatial reference and resampling.

5. The method according to claim 1, characterized in that, In step (2), watershed analysis and visualization are used to explicitly simulate surface processes of landslide geological hazards, erosion caused by geomorphic evolution, and changes in bedrock height and sediment thickness on a regular terrain grid: a. Mass balance The model represents terrain as a layer of weathered crust or colluvium covering bedrock, the thickness of which varies over time and space; surface elevation... Bedrock surface elevation and the thickness of the movable loose layer The temporal evolution follows the following mass balance equation, and in each model iteration, the elevation of all grid cells is updated according to the following sedimentary rock conservation equation: （11） （12） （13） This represents the rate of bedrock uplift relative to the reference surface. Surface sediment porosity, This represents the rate of bedrock erosion in rivers, while represents the volumetric flux per unit area. Indicates the sediment entrainment rate, Represents river deposition flux, total sediment flow. ,in For river transport flux, For soil peristaltic flux, This represents the maximum soil formation rate under zero soil thickness conditions. The soil attenuation depth and the bedrock erosion rate caused by the landslide are: The sediment entrainment rate is The deposition rate is ; b. River erosion, sediment transport and deposition processes The SPACE model was used to calculate the processes of river bedrock erosion, sediment transport, and deposition. The formula representing the sediment entrainment rate is as follows: （14） The formula for expressing the bedrock erosion rate is as follows: （15） and These are the erosibility coefficients of sediments and bedrock, respectively. = Water flow rate per unit width; Merging and middle, This refers to the catchment area index. For the river slope, Slope index, sediment thickness Compared with bedrock roughness scale The ratio determines the relative dominance of sediment entrainment and bedrock erosion; the formula for calculating sediment flux is as follows: （16） For sedimentation flux, The sediment volumetric flow rate, The volumetric flow rate of the water flow. The effective sediment settling velocity; in special cases, in closed topographic depressions formed by landslides where erosion does not occur, the settling velocity is taken as... The water flow path and dam area identification uses a priority filling depression flow direction algorithm; c. Soil formation and nonlinear soil creep Weathering of bedrock into a movable loose layer is represented by an exponential function of the local loose layer thickness; soil creep on the slope is simulated using nonlinear, thickness-dependent transport laws, and a truncated Taylor series is used to approximate the weathering when the slope approaches a critical value. The superlinear growth of time flux and the exponential soil creep velocity profile used to represent the effect of loose layer thickness on transport rate are thus combined to obtain the following soil flux formula: （17） in, The slope of the terrain To reduce soil peristalsis to The feature thickness at its maximum value, The diffusion coefficient is... This is the critical slope. Let be the number of terms in the Taylor series expansion.

6. The method according to claim 1, characterized in that, Step (3) The three-dimensional radiative transfer simulation steps are divided into constructing a multi-temporal three-dimensional scene, inputting the spectral parameters of the surface components, carrying out multi-temporal, full-spectrum radiative transfer simulation, using the three-dimensional radiative transfer models LESS and RAPID, and generating remote sensing reflectance simulation images of the pre-disaster, mid- and post-disaster stages: a. Constructing a three-dimensional scene of multi-temporal surface disasters In the stage of constructing a multi-temporal 3D scene, after inputting the DEM before and after the surface changes into the 3D radiation transfer model, the disaster scene components are considered, that is, the 3D scene is generated based on the type and spatial distribution of soil, bedrock, vegetation, water body and building elements in the disaster site. b. Input component spectrum Component spectra represent the reflection, absorption, transmission, and scattering characteristics of different objects in a scene across different wavelengths. In remote sensing image simulation, reflectance is the target. Component spectra represent the reflectance of different objects in a scene across different wavelengths, including the reflection and scattering characteristics of soil, bedrock, vegetation, water bodies, and buildings in the visible, near-infrared, and microwave wavelengths. When defining component spectra using LESS and RAPID three-dimensional radiative transfer models, the reflectance of different objects in the radiative transfer simulation wavelengths is defined. c. Simulation of disaster radiation transfer When conducting multi-temporal, full-spectrum radiative transfer simulations to generate remote sensing reflectance and images before, during, and after a disaster, the three-dimensional radiative transfer model used is based on the ray tracing method LESS and the radiometric method RAPID, which constrain the sensor, observation geometry, atmospheric conditions, and terrain.

7. The method according to claim 1, characterized in that, The accuracy verification in step (4) includes calculating the root mean square error of reflectance of the simulated image and the real remote sensing image, as well as the correlation coefficient R and the coefficient of determination R².

8. The method according to claim 1, characterized in that, The method is applicable to remote sensing response simulation of geological hazards such as landslides, collapses, and debris flows.

Images