A method and system for predicting intensity of landslide activity in a strong earthquake area based on multi-source remote sensing

Abstract

This application relates to a method and system for predicting landslide activity intensity in strong earthquake zones based on multi-source remote sensing. The method includes: acquiring multi-period high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models of the strong earthquake zone; obtaining a multi-period orthorectified image dataset through orthorectification, topographic radiometric correction, and registration; establishing a three-dimensional interpretation environment and constructing interpretation markers for landslide activity; interpreting and classifying various types of landslides and vegetation restoration areas periodically to generate a landslide activity logging dataset; calculating the landslide activity intensity for each year to obtain a time series; fitting the decay relationship of landslide activity intensity over time to obtain a landslide activity intensity decay function; and inputting the difference between the target prediction year and the year of the earthquake into the landslide activity intensity decay function to obtain the predicted landslide activity intensity value. This method enables long-term quantitative prediction of landslide activity intensity in strong earthquake zones, providing a time-scale basis for post-disaster reconstruction.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing monitoring and prediction technology for geological disasters, and in particular relates to a method and system for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing. Background Technology

[0002] Following a strong earthquake, the affected area often experiences numerous coseismic collapses and landslides, which are forms of gravity erosion. These coseismic landslides remain active and evolve for years to decades after the earthquake, posing a long-term threat to the recovery and reconstruction of the disaster area. Accurately assessing and predicting the intensity of post-earthquake landslide activity and its changes over time is a crucial prerequisite for scientifically determining the reconstruction sequence and rationally deploying disaster prevention projects.

[0003] Currently, the assessment of the long-term activity of landslides in strong earthquake zones mainly relies on visual interpretation of multi-phase remote sensing images. The interpretation results typically only qualitatively determine whether landslides are "active" or "inactive" in a particular year, making it difficult to form a quantitative description of activity covering a long post-earthquake timeline. Some studies attempt to statistically analyze landslide areas in specific years, but often conflate co-seismic landslides, later-expanding landslides, and newly formed landslides, failing to distinguish between different types of landslides and thus unable to extract universally applicable patterns of activity intensity decay. Furthermore, existing methods lack a complete technical chain for continuously extracting the extent of active landslides from multi-phase remote sensing images, quantifying them into activity intensity indicators, and establishing the relationship between activity intensity and time. This makes it impossible to provide key time points for post-disaster reconstruction, such as how long the active landslide period will last and when it will enter a stable state. Summary of the Invention

[0004] Therefore, it is necessary to provide a method and system for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing, which can provide long-term quantitative prediction of the intensity of landslide activity in strong earthquake zones, in order to address the above-mentioned technical problems.

[0005] Firstly, this application provides a method for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing, including:

[0006] Multiple high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models were acquired for the strong earthquake zone. Based on the digital elevation model, orthorectification, topographic radiometric correction, and registration were performed on the multiple high-resolution optical remote sensing images and synthetic aperture radar images to obtain a multi-phase orthorectified image dataset.

[0007] Based on the digital elevation model and multi-period orthophoto dataset, a three-dimensional interpretation environment was established, and the activity interpretation markers of landslide bodies were constructed based on the three-dimensional interpretation environment.

[0008] In the three-dimensional interpretation environment, landslides in each year are interpreted according to the landslide activity interpretation markers. The landslides are divided into coseismic landslides, expanded landslides, newly added landslides and dormant landslides, and vegetation restoration zones are delineated to generate a landslide activity logging dataset.

[0009] Based on the total area of ​​coseismic landslides, the area of ​​expanded landslides, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones in the landslide activity logging dataset, the landslide activity intensity for each year is calculated to obtain the time series of landslide activity intensity.

[0010] Based on the time series of landslide activity intensity, the decay relationship of landslide activity intensity over time is fitted to obtain the landslide activity intensity decay function;

[0011] The difference between the target predicted year and the year of the earthquake is calculated. This difference is then input into the landslide activity intensity attenuation function to predict the activity intensity, thus obtaining the predicted value of the landslide activity intensity for the target predicted year.

[0012] Secondly, this application also provides a system for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing, the system comprising:

[0013] The data acquisition module is used to acquire multiple high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models of the strong earthquake zone. Based on the digital elevation model, the module performs orthorectification, topographic radiometric correction, and registration on the multiple high-resolution optical remote sensing images and synthetic aperture radar images to obtain a multi-phase orthorectified image dataset.

[0014] The interpretation marker construction module is used to establish a three-dimensional interpretation environment based on the digital elevation model and multi-period orthophoto dataset, and to construct interpretation markers of landslide activity based on the three-dimensional interpretation environment;

[0015] The landslide classification module is used to interpret landslides of different years in a 3D interpretation environment based on landslide activity interpretation markers. It classifies landslides into coseismic landslides, expanded landslides, newly added landslides, and dormant landslides, and delineates vegetation restoration zones, generating a landslide activity logging dataset.

[0016] The activity intensity calculation module is used to calculate the activity intensity of landslides in each year based on the total area of ​​coseismic landslides, the area of ​​expanded landslides in each year, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones in the landslide activity logging dataset, and to obtain the time series of landslide activity intensity.

[0017] The decay function construction module is used to fit the decay relationship of landslide activity intensity over time based on the landslide activity intensity time series, and obtain the landslide activity intensity decay function.

[0018] The activity intensity prediction module is used to calculate the year difference between the target prediction year and the year of the earthquake. The year difference is input into the landslide activity intensity attenuation function to perform activity intensity prediction and obtain the predicted value of landslide activity intensity for the target prediction year.

[0019] The aforementioned method and system for predicting landslide activity intensity in strong earthquake zones based on multi-source remote sensing acquires multiple high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models (DEMs) of the strong earthquake zone. Based on the DEM, orthorectification, topographic radiometric correction, and registration are performed on the multi-source images to generate a multi-phase orthorectified image dataset. This enables geometric and radiometric consistency correction of multi-source remote sensing data, unifying the spatial coordinate system and radiometric reference of the multi-phase images. It addresses the problems of limited information in traditional single-source remote sensing data, spatial misalignment of multi-phase images, and poor image comparability and low subsequent interpretation accuracy caused by geometric distortions due to topography. By establishing a three-dimensional interpretation environment based on digital elevation models and multi-period orthophoto datasets, and constructing interpretation markers for landslide activity based on this environment, an upgrade from two-dimensional planar interpretation to three-dimensional interpretation can be achieved. A standardized index system for judging landslide activity can be established, addressing the problems of traditional two-dimensional interpretation, such as a single perspective, inability to accurately identify landslide boundaries and morphologies in complex terrain areas, and the lack of unified standards for interpretation markers leading to strong subjectivity and poor consistency in interpretation results. Furthermore, by classifying landslides from different years according to their activity interpretation markers within the three-dimensional interpretation environment, the interpretation can be further refined. Different types of landslides and vegetation restoration zones are used to generate landslide activity logging datasets. This enables accurate classification of landslides at different evolutionary stages after strong earthquakes and a complete record of their spatiotemporal evolution. It comprehensively captures changes in landslide activity and vegetation restoration processes, solving the problems of traditional interpretation methods that can only identify the overall extent of landslides, cannot distinguish between coseismic, expanding, newly formed, and dormant landslides, and ignore the distorted activity intensity assessment caused by the inhibitory effect of vegetation restoration on activity. The activity intensity of landslides in each year is calculated using the landslide activity logging dataset, generating a time series of activity intensity and fitting a landslide activity intensity decay function. By inputting the year difference into the landslide activity intensity decay function, the predicted activity intensity value for the target prediction year is obtained. This enables quantitative characterization of landslide activity intensity and accurate fitting of long-term decay patterns after strong earthquakes, allowing for quantitative prediction of activity intensity in future years. This addresses the problems of traditional landslide activity assessments, which are mostly qualitative descriptions, unable to quantify the decay patterns of activity intensity over time, and lacking scientific quantitative prediction methods, resulting in highly subjective and inaccurate predictions. It enables long-term quantitative prediction of landslide activity intensity in strong earthquake zones, providing a scientific quantitative basis for long-term risk assessment and prevention and control decisions regarding landslide disasters in strong earthquake zones. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a method for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing, according to the present invention.

[0022] Figure 2 This is a flowchart illustrating the generation of a time series of landslide activity intensity in one optional embodiment of the present invention;

[0023] Figure 3 This is a structural diagram of a landslide activity intensity prediction system for strong earthquake zones based on multi-source remote sensing, according to the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0025] In one embodiment, such as Figure 1 As shown, a method for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing is provided. This embodiment illustrates the application of this method to a landslide activity intensity prediction terminal. It is understood that this method can also be applied to a landslide activity intensity prediction server, and further to a landslide activity intensity prediction system including both a landslide activity intensity prediction terminal and a landslide activity intensity prediction server, and is implemented through the interaction between the two. In this embodiment, the method includes the following steps:

[0026] Step S101: Acquire multiple high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models of the strong earthquake zone. Based on the digital elevation model, perform orthorectification, topographic radiometric correction, and registration on the multiple high-resolution optical remote sensing images and synthetic aperture radar images to obtain a multi-phase orthorectified image dataset.

[0027] Optionally, the landslide activity intensity prediction terminal can acquire multiple high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models (DEMs) covering the strong earthquake zone from the year of the earthquake to the current year from a remote sensing satellite data service platform and a geographic information database. The terminal can then perform orthorectification on the high-resolution optical remote sensing images based on the DEMs to eliminate geometric distortions caused by terrain undulations and sensor attitude, resulting in orthorectified high-resolution optical remote sensing images. Based on the DEMs, the terminal can calculate terrain slope, aspect, and solar incidence angle to generate terrain radiometric correction coefficients, and then perform terrain radiometric correction on the DEMs to eliminate radiometric errors caused by terrain shadows and slope differences, resulting in corrected DEMs. Finally, the terminal can register the orthorectified high-resolution optical remote sensing images and the corrected DEMs using the geographic coordinate system of the DEMs as a reference, obtaining a multi-period orthorectified image dataset.

[0028] Specifically, high-resolution optical remote sensing imagery can be visible-near-infrared band remote sensing imagery with a spatial resolution better than 1 meter. High-resolution optical remote sensing imagery can be used to clearly identify the texture, tone, and boundary features of surface features. Synthetic aperture radar (SAR) imagery can be microwave band remote sensing imagery acquired through a SAR system. SAR imagery can be used for all-weather imaging, supplementing the data loss of optical imagery in cloudy or rainy weather. Digital elevation models (DEMs) can be raster datasets describing surface elevation information. DEMs can be used for terrain correction and 3D terrain modeling. Orthorectification is the process of correcting geometric distortions in remote sensing imagery caused by terrain undulations, sensor tilt, and Earth curvature using DEMs to generate images with orthorectified projection characteristics. Topographic radiometric correction is the process of eliminating differences in solar radiation energy received by the surface due to terrain undulations and correcting image grayscale values. Registration is the process of aligning remote sensing imagery from different sources and periods to the same geographic coordinate system. Multi-period orthorectified image datasets can be a collection of multiple years of high-resolution optical remote sensing imagery and SAR imagery that have undergone orthorectification, topographic radiometric correction, and registration.

[0029] Step S102: Based on the digital elevation model and multi-period orthophoto dataset, a three-dimensional interpretation environment is established, and the activity interpretation markers of landslide bodies are constructed based on the three-dimensional interpretation environment.

[0030] Optionally, the landslide activity intensity prediction terminal can construct a three-dimensional terrain model based on a digital elevation model; the landslide activity intensity prediction terminal can overlay each orthophoto from a multi-period orthophoto dataset onto the three-dimensional terrain model to establish a three-dimensional interpretation environment; the landslide activity intensity prediction terminal can perform multi-view and multi-scale observation of landslides at different times in the multi-period orthophoto dataset in the three-dimensional interpretation environment, extract typical features of landslides in different activity states, establish the correspondence between landslide activity and orthophoto features, and construct interpretation markers for landslide activity.

[0031] Specifically, the three-dimensional interpretation environment can be a three-dimensional visualization platform built based on digital elevation models and multi-period orthophoto datasets. The three-dimensional interpretation environment can be used to intuitively display the topographic relief, land cover distribution and spatial relationships of the land surface. The landslide activity interpretation markers can be a set of image features used to determine the activity status of landslides. The landslide activity interpretation markers can include, but are not limited to, hue, texture, micro-topography and vegetation cover features.

[0032] Step S103: In the three-dimensional interpretation environment, landslides of each year are interpreted according to the landslide activity interpretation markers. The landslides are divided into coseismic landslides, expanded landslides, newly added landslides, and dormant landslides, and vegetation restoration zones are delineated to generate a landslide activity logging dataset.

[0033] Optionally, the landslide activity intensity prediction terminal can load orthophotos in the 3D interpretation environment in chronological order; the landslide activity intensity prediction terminal can identify and delineate the boundaries of different types of landslides in each period according to the landslide activity interpretation markers, classify and label landslides, and obtain coseismic landslides, expanded landslides, newly added landslides, and dormant landslides; the landslide activity intensity prediction terminal can identify and delineate the boundaries of vegetation restoration zones, and divide the vegetation restoration zones; the landslide activity intensity prediction terminal can organize the interpretation results by year and type to generate a landslide activity logging dataset.

[0034] Specifically, a coseismic landslide can be a collapse or landslide directly induced by the earthquake motion at the moment of the earthquake; an expanding landslide can be a landslide formed after the earthquake due to the instability of loose material, which expands outward along the original boundary; a newly formed landslide can be a collapse or landslide that newly occurs in an area where no landslides or landslides occurred before the earthquake; a dormant landslide can be a landslide that has ceased activity and no longer undergoes significant displacement or material transport; a vegetation restoration zone can be an area where vegetation has regrowed and covered the surface of a landslide, and the vegetation coverage has reached a certain level; and a landslide activity logging dataset can be a structured dataset that includes the spatial location, boundary range, area, and attribute information of different types of landslides and vegetation restoration zones in different years.

[0035] Step S104: Based on the total area of ​​coseismic landslides, the area of ​​expanded landslides, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones in the landslide activity logging dataset, calculate the landslide activity intensity for each year to obtain the landslide activity intensity time series.

[0036] Optionally, the landslide activity intensity prediction terminal can extract the total area of ​​coseismic landslides from the landslide activity logging dataset; the landslide activity intensity prediction terminal can extract the area of ​​expanded landslides, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones for each year from the landslide activity logging dataset; the landslide activity intensity prediction terminal can calculate the landslide activity intensity for each year based on the total area of ​​coseismic landslides, the area of ​​expanded landslides, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones for each year; the landslide activity intensity prediction terminal can arrange the landslide activity intensity for each year in chronological order to generate a landslide activity intensity time series.

[0037] Specifically, landslide activity intensity can be a dimensionless quantitative indicator that characterizes the intensity of landslide activity in a strong earthquake zone; landslide activity intensity time series can be an ordered data set composed of landslide activity intensity data for each year arranged in chronological order, and landslide activity intensity time series can be used to reflect the decay law of landslide activity over time after an earthquake.

[0038] Step S105: Based on the time series of landslide activity intensity, fit the decay relationship of landslide activity intensity over time to obtain the landslide activity intensity decay function.

[0039] Optionally, the landslide activity intensity prediction terminal can select a suitable mathematical model as the basic attenuation function based on the general attenuation law of landslide activity intensity after the earthquake; the landslide activity intensity prediction terminal can use the landslide activity intensity time series as sample data and solve for the unknown parameters in the basic attenuation function through numerical fitting method; the landslide activity intensity prediction terminal can substitute the solved parameters into the basic attenuation function to obtain the landslide activity intensity attenuation function.

[0040] Specifically, the decay relationship of landslide activity intensity over time can be a mathematical relationship describing the gradual decrease of landslide activity intensity over time; the landslide activity intensity decay function can be a mathematical function that quantitatively characterizes the decay relationship of landslide activity intensity, and the landslide activity intensity decay function can be used to predict the landslide activity intensity in any future year.

[0041] Step S106: Calculate the year difference between the target predicted year and the year of the earthquake, input the year difference into the landslide activity intensity attenuation function, perform activity intensity prediction, and obtain the predicted value of landslide activity intensity for the target predicted year.

[0042] Optionally, the landslide activity intensity prediction terminal can obtain the target prediction year, calculate the integer number difference between the target prediction year and the year of the earthquake, and obtain the year difference value; the landslide activity intensity prediction terminal can substitute the year difference value into the landslide activity intensity attenuation function to calculate the predicted value of landslide activity intensity for the target prediction year.

[0043] Specifically, the target prediction year can be a future year in which the intensity of landslide activity needs to be predicted; the year difference can be the time interval between the target prediction year and the year of the earthquake, in years; and the predicted value of landslide activity intensity can be a quantitative estimate of the landslide activity intensity in the target year calculated based on the landslide activity intensity attenuation function.

[0044] The aforementioned method for predicting landslide activity intensity in strong earthquake zones based on multi-source remote sensing acquires multiple high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models of the strong earthquake zone. After orthorectification, topographic radiometric correction, and registration, a multi-period orthorectified image dataset is obtained. A three-dimensional interpretation environment is established, and landslide activity interpretation markers are constructed. Various types of landslides and vegetation restoration areas are interpreted and classified periodically to generate a landslide activity logging dataset. The landslide activity intensity for each year is calculated to obtain a time series. The decay relationship of landslide activity intensity over time is fitted to obtain a landslide activity intensity decay function. The difference between the target prediction year and the year of the earthquake is input into the landslide activity intensity decay function to obtain the predicted landslide activity intensity value. This method enables long-term quantitative prediction of landslide activity intensity in strong earthquake zones, providing a time-scale basis for post-disaster reconstruction.

[0045] In one embodiment, a three-dimensional interpretation environment is established based on a digital elevation model and a multi-period orthophoto dataset, and interpretation markers of landslide activity are constructed based on this three-dimensional interpretation environment, which may include:

[0046] Step S201: Generate an irregular triangular network based on the digital elevation model, and construct a three-dimensional terrain base based on the irregular triangular network.

[0047] Optionally, the landslide activity intensity prediction terminal can use the Delaunay triangulation algorithm to convert the grid points of the digital elevation model into an irregular triangular network; the landslide activity intensity prediction terminal can then construct a continuous three-dimensional surface model based on the irregular triangular network to obtain a three-dimensional terrain base.

[0048] Specifically, an irregular triangular mesh can be a digital terrain model composed of a series of continuous and non-overlapping triangles. Irregular triangular meshes can be used to accurately represent complex terrain undulations with a small amount of data. A three-dimensional terrain base can be a basic three-dimensional terrain framework for constructing a three-dimensional interpretation environment. A three-dimensional terrain base can be used to reflect changes in terrain elevation and slope in strong earthquake zones.

[0049] Step S202: Each orthophoto in the multi-period orthophoto dataset is used as a texture and mapped onto the three-dimensional terrain base to obtain the three-dimensional interpretation environment.

[0050] Optionally, the landslide activity intensity prediction terminal can convert each orthophoto in a multi-period orthophoto dataset into a texture image; the landslide activity intensity prediction terminal can map the texture image to the corresponding position of the three-dimensional terrain base according to the geographic coordinate information of the orthophoto, to obtain the mapped three-dimensional terrain base; the landslide activity intensity prediction terminal can perform fusion optimization on the mapped three-dimensional terrain base to eliminate the tonal differences between different images and obtain a three-dimensional interpretation environment.

[0051] Step S203: In the three-dimensional interpretation environment, the texture, tone and micro-topography of the landslide body in each orthophoto are simultaneously compared and observed in three dimensions to obtain the bare expansion characteristics and vegetation restoration characteristics of the landslide body.

[0052] Optionally, the landslide activity intensity prediction terminal can load multiple orthophotos in a 3D interpretation environment to simultaneously compare the texture, tone, and micro-topography of the same landslide in different orthophotos, and observe and identify features such as slip marks, steep slopes, and cracks of the landslide in three dimensions to obtain the exposed expansion characteristics of the landslide; the landslide activity intensity prediction terminal can also analyze the changes in vegetation cover on the surface of the landslide to obtain the vegetation restoration characteristics of the landslide.

[0053] Specifically, texture can be the detailed features of the ground surface in each orthophoto, with landslides typically exhibiting rough and chaotic rock and soil textures; hue can be the grayscale or color characteristics of ground features in each orthophoto, with landslides typically appearing in light hues, and vegetation-covered landslides appearing in green hues; micro-topographic features can be the small topographic features on the surface of the landslide, which can include, but are not limited to, landslide walls, landslide steps, cracks, and accumulation fans; exposed expansion features can be the outward expansion of the exposed rock and soil area of ​​the landslide; and vegetation restoration features can be the gradual growth and coverage of vegetation on the surface of the landslide.

[0054] Step S204: Based on the characteristics of bare expansion and vegetation restoration, interpret the activity markers of landslide bodies.

[0055] Optionally, the landslide activity intensity prediction terminal can classify and organize the bare expansion characteristics and vegetation restoration characteristics, establish the correspondence between landslides and bare expansion characteristics and vegetation restoration characteristics in different activity states, and obtain the interpretation markers of landslide activity.

[0056] Specifically, the interpretation markers of landslide activity can be a standardized set of features used to determine the activity state of landslides. For example, active landslides have obvious bare expansion characteristics, light color, coarse texture, and low vegetation coverage; dormant landslides have no obvious expansion characteristics, the color gradually darkens, and sporadic vegetation begins to appear; vegetation restoration areas have high vegetation coverage, dark green color, and uniform texture.

[0057] In one embodiment, landslides from different years are interpreted, and the landslides are classified into coseismic landslides, expanded landslides, newly added landslides, and dormant landslides, and vegetation restoration zones are delineated to generate a landslide activity logging dataset, which may include:

[0058] Step S301: Extract the orthophoto of the year of the earthquake from the multi-period orthophoto dataset to obtain the coseismic orthophoto. Based on the interpretation markers of landslide activity, identify the boundaries of the coseismic landslide on the coseismic orthophoto, generate the coseismic landslide surface layer, and delineate the coseismic landslide.

[0059] Optionally, the landslide activity intensity prediction terminal can extract orthophotos of the year of the earthquake from a multi-period orthophoto dataset to obtain coseismic orthophotos; the landslide activity intensity prediction terminal can load coseismic orthophotos in a 3D interpretation environment, identify and delineate the boundaries of coseismic landslides based on landslide activity interpretation markers, and generate a coseismic landslide surface layer; the landslide activity intensity prediction terminal can mark the landslides corresponding to the coseismic landslide surface layer as coseismic landslides.

[0060] Specifically, coseismic orthophotos can be orthophotos acquired at the time of an earthquake or shortly after an earthquake, and can be used to reflect the distribution of coseismic landslides; the coseismic landslide surface layer can be a vector layer that includes the boundaries and attribute information of the coseismic landslides.

[0061] Step S302: Extract orthophotos of each year after the year of the earthquake from the multi-period orthophoto dataset to obtain post-earthquake orthophotos of each year. Overlay the coseismic landslide surface layer onto the post-earthquake orthophotos of each year. Based on the landslide activity interpretation markers, identify the outward expansion of the coseismic landslide boundary in the post-earthquake orthophotos of each year to generate the enlarged landslide surface layer for each year and delineate the enlarged landslide.

[0062] Optionally, the landslide activity intensity prediction terminal can extract orthophotos of each year after the year of the earthquake from a multi-period orthophoto dataset to obtain post-earthquake orthophotos for each year; the landslide activity intensity prediction terminal can overlay a coseismic landslide surface layer onto the post-earthquake orthophoto, identify the outward expansion of the coseismic landslide boundary based on landslide activity interpretation markers, delineate the boundary of the expansion portion, and generate an expanded landslide surface layer for each year; the landslide activity intensity prediction terminal can mark the landslides corresponding to the expanded landslide surface layer as expanded landslides.

[0063] Specifically, post-earthquake orthophotos can be orthophotos from each year after the earthquake, and can be used to monitor the evolution of post-earthquake landslides; the enlarged landslide surface layer can be a vector layer that includes the boundary and attribute information of the extended portion of the coseismic landslide in each year.

[0064] Step S303: Based on the interpretation markers of landslide activity, the areas in the post-earthquake orthophotos of each year other than the coseismic landslide surface layer and the expanded landslide surface layer are inspected to identify landslides that have not appeared before, generate new landslide surface layers, and delineate the new landslides.

[0065] Optionally, the landslide activity intensity prediction terminal can conduct a comprehensive survey of areas other than coseismic landslides and extended landslides in the post-earthquake orthophotos of each year, identify newly generated landslides based on landslide activity interpretation markers, delineate the boundaries of newly added landslides, and generate new landslide surface layers for each year; the landslide activity intensity prediction terminal can mark the landslides corresponding to the new landslide surface layers as newly added landslides.

[0066] Specifically, the newly added landslide surface layer can be a vector layer that includes the boundaries and attribute information of the newly added landslides for each year.

[0067] Step S304: Based on the interpretation markers of landslide activity, determine the activity status of the expanded landslide surface layer and the newly added landslide surface layer in the post-earthquake orthophotos of each year. Mark the expanded landslide surface layer and the newly added landslide surface layer that have stopped activity and have vegetation as dormant landslide surface layers, and delineate dormant landslide bodies.

[0068] Optionally, the landslide activity intensity prediction terminal can load the expanded landslide surface layer and the newly added landslide surface layer from the previous year; the landslide activity intensity prediction terminal can overlay the expanded landslide surface layer and the newly added landslide surface layer from the previous year onto the post-earthquake orthophoto of the current year, and determine the activity status of each landslide based on the landslide activity interpretation markers, marking the expanded landslide surface layer and the newly added landslide surface layer that have ceased activity and have vegetation cover as dormant landslide surface layers; the landslide activity intensity prediction terminal can mark the landslides corresponding to the dormant landslide surface layers as dormant landslides.

[0069] Specifically, the dormant landslide surface layer can be a vector layer that includes the boundaries and attribute information of inactive landslides for each year.

[0070] Step S305: Based on the activity interpretation markers of landslide bodies, the surface layer of dormant landslide bodies with vegetation exceeding the restoration threshold is marked as the surface layer of vegetation restoration area, and the vegetation restoration area is delineated.

[0071] Optionally, the landslide activity intensity prediction terminal can calculate the normalized vegetation index (NVI) corresponding to each dormant landslide surface layer based on the amount of vegetation in the dormant landslide; the landslide activity intensity prediction terminal can compare the NVI with the vegetation recovery threshold, and mark the dormant landslide surface layers with the NVI exceeding the recovery threshold as vegetation recovery area surface layers; the landslide activity intensity prediction terminal can mark the area corresponding to the vegetation recovery area surface layer as a vegetation recovery area.

[0072] Specifically, the vegetation restoration threshold can be the normalized vegetation index (NVI) threshold. When the NVI of the landslide surface exceeds the vegetation restoration threshold, the vegetation is considered to have been basically restored. The vegetation restoration area layer can be a vector layer that includes the boundaries and attribute information of the vegetation restoration areas for each year.

[0073] Step S306: Collect coseismic landslides, expanded landslides from each year, newly added landslides, dormant landslides, and vegetation restoration areas to obtain a landslide activity logging dataset.

[0074] Optionally, the landslide activity intensity prediction terminal can aggregate coseismic landslide surface layers, expanded landslide surface layers for all years, newly added landslide surface layers, dormant landslide surface layers, and vegetation restoration zone surface layers, and add attribute information such as year, type, and area to the patches in each surface layer to obtain a landslide activity logging dataset.

[0075] Specifically, the landslide activity logging dataset can be a comprehensive dataset that includes spatial vector data and attribute data. The landslide activity logging dataset can be used to record the distribution, extent and evolution of different types of landslides and vegetation restoration areas in strong earthquake zones in different years.

[0076] In one embodiment, such as Figure 2 As shown, based on the total area of ​​coseismic landslides, the area of ​​expanded landslides, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones in the landslide activity logging dataset, the landslide activity intensity for each year is calculated, resulting in a time series of landslide activity intensity that can include:

[0077] Step S401: Add the expanded landslide area and the newly added landslide area for each year to obtain the preliminary active landslide area for each year. Subtract the dormant landslide area and the vegetation restoration area from the preliminary active landslide area to obtain the active landslide area for each year.

[0078] Optionally, the landslide activity intensity prediction terminal can extract the expanded landslide area, newly added landslide area, dormant landslide area, and vegetation restoration area for each year from the landslide activity logging dataset; the landslide activity intensity prediction terminal can add the expanded landslide area and newly added landslide area for the same year to obtain the preliminary active landslide area for that year; the landslide activity intensity prediction terminal can subtract the dormant landslide area and vegetation restoration area for the same year from the preliminary active landslide area to obtain the active landslide area for that year.

[0079] Specifically, the preliminary active landslide area can be the total area of ​​newly generated active landslides in each year; the active landslide area can be the net area of ​​landslides that are actually active in the current year after deducting the area of ​​dormant landslides and the area of ​​vegetation restoration zones from the preliminary active landslide area.

[0080] Step S402: Calculate the activity intensity of landslides in each year based on the total area of ​​the landslides caused by the co-seismic event and the area of ​​active landslides in each year.

[0081] Optionally, the formula for calculating the activity intensity of landslides is:

[0082] ;

[0083] In the formula, For the first Annual landslide activity intensity For the first The area of ​​landslides and collapses in the year. This represents the total area of ​​the landslide body caused by the same earthquake.

[0084] Specifically, the total area of ​​coseismic landslides can be the sum of the areas of all coseismic landslides, and the total area of ​​coseismic landslides can be used to eliminate the influence of differences in earthquake magnitude and area of ​​the study area.

[0085] Step S403: Arrange the landslide activity intensity of each year in chronological order to generate a time series of landslide activity intensity.

[0086] Optionally, the landslide activity intensity prediction terminal can sort the landslide activity intensity of each year in ascending order of year, and add a corresponding year label to the landslide activity intensity of each year to obtain the landslide activity intensity time series.

[0087] Specifically, the landslide activity intensity time series can be an ordered data set consisting of time and corresponding landslide activity intensity. The landslide activity intensity time series can be used to analyze the variation law of landslide activity intensity over time and fit the decay function.

[0088] In one embodiment, fitting the decay relationship of landslide activity intensity over time based on the landslide activity intensity time series to obtain the landslide activity intensity decay function may include:

[0089] Step S501: Construct an exponential decay function with translation parameters to obtain the decay function of the activity intensity of the landslide body to be determined.

[0090] Optionally, the landslide activity intensity prediction terminal can select an exponential decay model as the basic function based on the general pattern that the landslide activity intensity in strong earthquake zones rapidly decays after the earthquake and tends to stabilize in the later stage; the landslide activity intensity prediction terminal can introduce translation parameters into the basic exponential decay function to obtain the landslide activity intensity decay function to be determined.

[0091] Optionally, the expression for the attenuation function of the activity intensity of the landslide body to be determined is:

[0092] ;

[0093] In the formula, The difference in the number of years between the current year and the year the earthquake occurred. This is the initial attenuation amplitude coefficient. The decay rate coefficient, For translation parameters, This represents an exponential function with the natural constant e as its base.

[0094] Specifically, the exponential decay function with translation parameters can be a function that can fit the decay process of post-earthquake landslide activity intensity well; the landslide activity intensity decay function to be determined can be an exponential decay function containing unknown parameters, and the parameters can be solved by fitting the landslide activity intensity time series.

[0095] Step S502, extract the first from the time series of landslide activity intensity. Annual landslide activity intensity And calculate the year The difference in the number of years between the year the earthquake occurred and the year the earthquake occurred. The constructed landslide activity intensity data is used for .

[0096] Optionally, the landslide activity intensity prediction terminal can extract the landslide activity intensity for each year from the landslide activity intensity time series; the landslide activity intensity prediction terminal can calculate the number of years between each year and the year in which the earthquake occurred; the landslide activity intensity prediction terminal can pair the number of years between each year and the corresponding landslide activity intensity to construct a set of landslide activity intensity data pairs.

[0097] Specifically, the landslide activity intensity data pair can be data points composed of the annual difference and the corresponding landslide activity intensity. The landslide activity intensity data pair can be used to fit the parameters of the landslide activity intensity decay function to be determined.

[0098] Step S503: The activity intensity data of the landslide body are analyzed using the nonlinear least squares method. By performing fitting, the parameters in the attenuation function of the landslide activity intensity are obtained. , and The value of is used to construct the attenuation function of the landslide activity intensity.

[0099] Optionally, the landslide activity intensity prediction terminal can use landslide activity intensity data as input data and solve for the parameters using the Levenberg-Marquardt algorithm. , and The optimal value; the landslide activity intensity prediction terminal can determine the parameter. , and Substituting the optimal value into the landslide activity intensity attenuation function, we obtain the landslide activity intensity attenuation function.

[0100] Specifically, the nonlinear least squares method can be an optimization method for solving nonlinear model parameters; the Levenberg-Marquardt algorithm can be a commonly used nonlinear least squares solution algorithm; the landslide activity intensity decay function can be an exponential decay function including definite parameters, and the landslide activity intensity decay function can be used to quantitatively predict the landslide activity intensity in any future year.

[0101] In one embodiment, the year difference between the target predicted year and the year of the earthquake is calculated, and this year difference is input into the landslide activity intensity attenuation function to predict the activity intensity, thereby obtaining the predicted landslide activity intensity value for the target predicted year. This may include:

[0102] Step S601: Obtain the target predicted year, calculate the difference between the target predicted year and the year of the earthquake, and obtain the year difference.

[0103] Optionally, the landslide activity intensity prediction terminal can obtain the target prediction year and read the earthquake occurrence year from the system parameters; the landslide activity intensity prediction terminal can calculate the difference between the target prediction year and the earthquake occurrence year to obtain the year difference.

[0104] Specifically, the target prediction year can be any future year in which the intensity of landslide activity needs to be predicted; the year difference can be an integer number of years between the target prediction year and the year the earthquake occurred.

[0105] Step S602: Input the year difference into the landslide activity intensity decay function to calculate the predicted value of landslide activity intensity for the target predicted year.

[0106] Optionally, the landslide activity intensity prediction terminal can substitute the year difference into the landslide activity intensity decay function and calculate according to the function expression to obtain the predicted value of landslide activity intensity for the target prediction year.

[0107] Specifically, the predicted value of landslide activity intensity can be an estimate of the landslide activity intensity for a target year, calculated based on a landslide activity intensity decay function fitted with historical data. The predicted value of landslide activity intensity can be used to reflect the overall activity level of landslides in that year.

[0108] In one embodiment, the method for predicting the intensity of landslide activity in a strong earthquake zone based on multi-source remote sensing may further include:

[0109] Step S701: Set the active period threshold and the stable period threshold.

[0110] Optionally, the landslide activity intensity prediction terminal can set active period thresholds and stable period thresholds based on relevant geological disaster prevention and control standards and the actual geological conditions of strong earthquake zones.

[0111] Specifically, the active period threshold can be a critical value of the activity intensity of a landslide body used to determine whether it is in an active period; the stable period threshold can be a critical value of the activity intensity of a landslide body used to determine whether it has entered a stable period.

[0112] Step S702: Based on the landslide activity intensity decay function, calculate the number of years corresponding to when the landslide activity intensity value is equal to the active period threshold, and obtain the difference in the number of years at the end of the active period.

[0113] Optionally, the landslide activity intensity prediction terminal can substitute the active period threshold into the landslide activity intensity decay function to obtain an equation about the number of years difference; the landslide activity intensity prediction terminal can solve this equation to obtain the number of years difference corresponding to when the landslide activity intensity equals the active period threshold, and use this number of years difference as the difference in the year of the end of the active period.

[0114] Specifically, the difference in the number of years from the year the earthquake occurred to the time required for the intensity of landslide activity to decrease to the threshold of the active period.

[0115] Step S703: Based on the landslide activity intensity decay function, calculate the number of years corresponding to when the landslide activity intensity value is equal to the stable period threshold, and obtain the difference in the starting year of the stable period.

[0116] Optionally, the landslide activity intensity prediction terminal can substitute the stable period threshold into the landslide activity intensity decay function to obtain an equation about the number of years difference; the landslide activity intensity prediction terminal can solve this equation to obtain the number of years difference corresponding to when the landslide activity intensity equals the stable period threshold, and use this number of years difference as the starting year difference of the stable period.

[0117] Specifically, the difference in the starting year of the stabilization period can be the number of years required from the year of the earthquake to the point where the intensity of landslide activity drops to the threshold of the stabilization period.

[0118] Step S704: Add the difference between the year the earthquake occurred and the year the active period ended to obtain the year the active period ended; add the difference between the year the earthquake occurred and the year the stable period began to obtain the year the stable period began.

[0119] Optionally, the landslide activity intensity prediction terminal can add the difference between the year of the earthquake and the year of the end of the active period to calculate the year of the end of the active period; the landslide activity intensity prediction terminal can add the difference between the year of the earthquake and the year of the start of the stable period to calculate the year of the start of the stable period.

[0120] Specifically, the year that marks the end of the active period can be the year in which the landslide activity transitions from the active period to the recovery period, and the year that marks the end of the active period can be the last year of the active period; the year that marks the beginning of the stable period can be the year in which the landslide activity transitions from the recovery period to the stable period, and the year that marks the beginning of the stable period can be the first year of the stable period.

[0121] Step S705: Based on the end year of the active period and the beginning year of the stable period, divide the landslide body evolution stages. The period from the year of the earthquake to the end year of the active period is divided into the active period, the period from the year following the end of the active period to the year preceding the beginning year of the stable period is divided into the recovery period, and the period after the beginning year of the stable period is divided into the stable period.

[0122] Optionally, the landslide activity intensity prediction terminal can divide the post-earthquake landslide evolution process into three consecutive stages based on the end year of the active period and the beginning year of the stable period; the landslide activity intensity prediction terminal can define the period from the year of the earthquake to the end year of the active period as the active period; the landslide activity intensity prediction terminal can define the period from the year following the end of the active period to the year before the beginning year of the stable period as the recovery period; and the landslide activity intensity prediction terminal can define all years after the beginning year of the stable period as the stable period.

[0123] Specifically, the active period can be the stage where the intensity of landslide activity is high and the disaster risk is high after the earthquake, and the active period can last for 3-5 years; the recovery period can be the stage where the intensity of landslide activity gradually decreases and the disaster risk gradually decreases, and the recovery period can last for 10-20 years; the stable period can be the stage where the intensity of landslide activity is low and tends to be stable, and the disaster risk is relatively small.

[0124] Step S706: Compare the target predicted year, the end year of the active period, and the start year of the stable period to obtain the landslide evolution stage to which the target predicted year belongs.

[0125] Optionally, the landslide activity intensity prediction terminal can compare the target prediction year, the end year of the active period, and the start year of the stable period to obtain the comparison results; the landslide activity intensity prediction terminal can determine whether the target prediction year belongs to the active period, the recovery period, or the stable period based on the comparison results, and obtain the landslide evolution stage to which the target prediction year belongs.

[0126] Specifically, the evolution stages of landslides can include an active period, a recovery period, and a stable period. These stages can provide a timescale basis for post-disaster reconstruction planning, site selection for construction, and the design of geological disaster prevention and control projects.

[0127] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0128] Based on the same inventive concept, this application also provides a system for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing, used to implement the aforementioned method for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing. The solution provided by this system is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the system for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing provided below can be found in the limitations of the method for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing described above, and will not be repeated here.

[0129] In one exemplary embodiment, such as Figure 3 As shown, a multi-source remote sensing-based landslide activity intensity prediction system 500 for strong earthquake zones is provided, comprising:

[0130] The data acquisition module 501 can be used to acquire multiple high-resolution optical remote sensing images, synthetic aperture radar images and digital elevation models of strong earthquake zones, and perform orthorectification, topographic radiometric correction and registration on the multiple high-resolution optical remote sensing images and synthetic aperture radar images based on the digital elevation model to obtain a multi-phase orthorectified image dataset.

[0131] The interpretation marker construction module 502 can be used to establish a three-dimensional interpretation environment based on the digital elevation model and multi-period orthophoto dataset, and to construct interpretation markers of landslide activity based on the three-dimensional interpretation environment;

[0132] The landslide body classification module 503 can be used in a three-dimensional interpretation environment to interpret landslide bodies of different years based on landslide body activity interpretation markers, classify landslide bodies into coseismic landslide bodies, expanded landslide bodies, newly added landslide bodies, and dormant landslide bodies, and delineate vegetation restoration zones to generate landslide body activity logging datasets.

[0133] The activity intensity calculation module 504 can be used to calculate the activity intensity of landslides in each year based on the total area of ​​coseismic landslides, the area of ​​expanded landslides in each year, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones in the landslide activity logging dataset, and obtain the time series of landslide activity intensity.

[0134] The decay function construction module 505 can be used to fit the decay relationship of landslide activity intensity over time based on the landslide activity intensity time series to obtain the landslide activity intensity decay function;

[0135] The activity intensity prediction module 506 can be used to calculate the year difference between the target prediction year and the year of the earthquake, input the year difference into the landslide activity intensity attenuation function, perform activity intensity prediction, and obtain the landslide activity intensity prediction value for the target prediction year.

[0136] The above embodiments merely illustrate several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the embodiments of this application, and these all fall within the protection scope of the embodiments of this application.

Claims

1. A method for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing, characterized in that the method... include: Multiple high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models of the strong earthquake zone are acquired. Based on the digital elevation model, orthorectification, topographic radiometric correction, and registration are performed on the multiple high-resolution optical remote sensing images and the synthetic aperture radar images to obtain a multi-phase orthorectified image dataset. Based on the digital elevation model and the multi-period orthophoto dataset, a three-dimensional interpretation environment is established, and based on the three-dimensional interpretation environment, interpretation markers of landslide activity are constructed. In the three-dimensional interpretation environment, landslides in each year are interpreted according to the landslide activity interpretation markers. The landslides are divided into coseismic landslides, expanded landslides, newly added landslides, and dormant landslides, and vegetation restoration zones are delineated to generate a landslide activity logging dataset. Based on the total area of ​​coseismic landslides, the area of ​​expanded landslides, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones in the landslide activity logging dataset, the landslide activity intensity for each year is calculated to obtain the landslide activity intensity time series. Based on the time series of the landslide activity intensity, the decay relationship of the landslide activity intensity over time is fitted to obtain the landslide activity intensity decay function; The difference between the target predicted year and the year of the earthquake is calculated, and the difference is input into the landslide activity intensity attenuation function to predict the activity intensity, thereby obtaining the predicted value of the landslide activity intensity for the target predicted year.

2. The method according to claim 1, characterized in that, The process involves establishing a three-dimensional interpretation environment based on the digital elevation model and the multi-period orthophoto dataset, and constructing interpretation markers for landslide activity based on this three-dimensional interpretation environment, including: An irregular triangular network is generated based on the digital elevation model, and a three-dimensional terrain base is constructed based on the irregular triangular network. Each orthophoto in the multi-period orthophoto dataset is used as a texture and mapped onto the three-dimensional terrain base to obtain the three-dimensional interpretation environment; In the three-dimensional interpretation environment, the texture, tone and micro-topography of the landslide body in each of the orthophotos are synchronously compared and observed in three dimensions to obtain the bare expansion characteristics and vegetation restoration characteristics of the landslide body. Based on the exposed expansion characteristics and the vegetation restoration characteristics, the activity interpretation markers of the landslide body are constructed.

3. The method according to claim 1, characterized in that, The process involves interpreting landslides from different years, classifying them into coseismic landslides, expanded landslides, newly added landslides, and dormant landslides, and delineating vegetation restoration zones to generate a landslide activity logging dataset, including: The orthophotos of the year of the earthquake are extracted from the multi-period orthophoto dataset to obtain coseismic orthophotos. Based on the interpretation markers of the landslide activity, the boundaries of the coseismic landslide are identified on the coseismic orthophotos, a coseismic landslide surface layer is generated, and the coseismic landslide is delineated. The orthophotos of each year after the year of the earthquake are extracted from the multi-period orthophoto dataset to obtain post-earthquake orthophotos of each year. The coseismic landslide surface layer is superimposed on the post-earthquake orthophotos of each year. Based on the landslide activity interpretation markers, the outward expansion of the coseismic landslide boundary in the post-earthquake orthophotos of each year is identified to generate the enlarged landslide surface layer of each year and delineate the enlarged landslide. Based on the activity interpretation markers of the landslide bodies, the areas in the post-earthquake orthophotos of each year other than the coseismic landslide body surface layer and the expanded landslide body surface layer are inspected to identify the landslide bodies that have not appeared before, generate new landslide body surface layers, and delineate the new landslide bodies. Based on the activity interpretation markers of the landslide bodies, the activity status of the expanded landslide body surface layer and the newly added landslide body surface layer in the previous year on the post-earthquake orthophotos of each year is determined. The expanded landslide body surface layer and the newly added landslide body surface layer that have stopped activity and have vegetation are marked as dormant landslide body surface layers, and the dormant landslide bodies are delineated. Based on the activity interpretation markers of the landslide body, the surface layer of the dormant landslide body where the vegetation quantity exceeds the restoration threshold is marked as the surface layer of the vegetation restoration zone, and the vegetation restoration zone is divided. The activity tracking dataset of landslides is obtained by combining the coseismic landslides, the expanded landslides from each year, the newly added landslides, the dormant landslides, and the vegetation restoration areas.

4. The method according to claim 1, characterized in that, The landslide activity intensity is calculated for each year based on the total area of ​​coseismic landslides, the area of ​​expanded landslides, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones in the landslide activity logging dataset, resulting in a time series of landslide activity intensity, including: The area of ​​the expanded landslide and the area of ​​the newly added landslide in each year are added together to obtain the preliminary active landslide area for each year. The area of ​​the dormant landslide and the area of ​​the vegetation restoration zone are then subtracted from the preliminary active landslide area to obtain the active landslide area for each year. Based on the total area of ​​the co-seismic landslides and the area of ​​the active landslides in each year, the activity intensity of the landslides in each year is calculated; wherein, the formula for calculating the activity intensity of the landslides is: ； In the formula, For the first The intensity of landslide activity in the year. For the first The area of ​​the landslides mentioned in the year. The total area of ​​the co-seismic landslide body; The landslide activity intensity of each year is arranged in chronological order to generate a time series of landslide activity intensity.

5. The method according to claim 1, characterized in that, The step of fitting the decay relationship of the landslide activity intensity over time based on the time series of the landslide activity intensity to obtain the landslide activity intensity decay function includes: An exponential decay function with translation parameters is constructed to obtain the decay function of the activity intensity of the landslide body to be determined; wherein, the expression of the decay function of the activity intensity of the landslide body to be determined is: ； In the formula, The difference in the number of years between the current year and the year the earthquake occurred. This is the initial attenuation amplitude coefficient. The decay rate coefficient, The translation parameters are... This represents an exponential function with the natural constant e as its base. The first time series of the landslide activity intensity was extracted from the time series of the landslide activity intensity. The intensity of landslide activity in the year And calculate the year The difference in the number of years between the year the earthquake occurred and the year the earthquake occurred. The constructed landslide activity intensity data is used for ; The activity intensity data of the landslide body were analyzed using the nonlinear least squares method. By fitting the data, the parameters in the attenuation function of the landslide activity intensity are obtained. , and The value of is used to construct the attenuation function of the activity intensity of the landslide body.

6. The method according to claim 5, characterized in that, The calculation of the year difference between the target predicted year and the year of the earthquake, inputting the year difference into the landslide activity intensity attenuation function, and performing activity intensity prediction to obtain the predicted landslide activity intensity value for the target predicted year includes: Obtain the target predicted year, calculate the difference between the target predicted year and the year of the earthquake, and obtain the year difference; The year difference is input into the landslide activity intensity decay function to calculate the predicted value of the landslide activity intensity for the target predicted year.

7. The method according to claim 5, characterized in that, The method further includes: Set active period thresholds and stable period thresholds; Based on the landslide activity intensity decay function, the difference in the number of years corresponding to when the landslide activity intensity value is equal to the active period threshold is calculated to obtain the difference in the year of the end of the active period; Based on the landslide activity intensity decay function, the difference in the number of years corresponding to when the landslide activity intensity value is equal to the stability period threshold is calculated to obtain the difference in the starting year of the stability period; The difference between the year the earthquake occurred and the year the active period ended is added together to obtain the year the active period ended; the difference between the year the earthquake occurred and the year the stable period began is added together to obtain the year the stable period began. Based on the end year of the active period and the start year of the stable period, the evolution stages of the landslide body are divided. The period from the year of the earthquake to the end year of the active period is defined as the active period, the period from the year following the end of the active period to the year preceding the start year of the stable period is defined as the recovery period, and the period after the start year of the stable period is defined as the stable period. By comparing the target predicted year, the end year of the active period, and the start year of the stable period, the evolution stage of the landslide to which the target predicted year belongs is obtained.

8. A system for predicting the intensity of landslide activity in strong earthquake zones based on multi-source remote sensing, characterized in that, The system includes: The data acquisition module is used to acquire multiple high-resolution optical remote sensing images, synthetic aperture radar images, and digital elevation models of the strong earthquake zone, and to perform orthorectification, topographic radiometric correction, and registration on the multiple high-resolution optical remote sensing images and synthetic aperture radar images based on the digital elevation model to obtain a multi-phase orthorectified image dataset. The interpretation marker construction module is used to establish a three-dimensional interpretation environment based on the digital elevation model and the multi-period orthophoto dataset, and to construct interpretation markers for landslide activity based on the three-dimensional interpretation environment. The landslide body segmentation module is used to interpret landslide bodies of each year in the three-dimensional interpretation environment according to the landslide body activity interpretation markers, and to segment the landslide bodies into coseismic landslide bodies, expanded landslide bodies, newly added landslide bodies, and dormant landslide bodies, and to delineate vegetation restoration zones, thereby generating a landslide body activity logging dataset. The activity intensity calculation module is used to calculate the activity intensity of landslides in each year based on the total area of ​​coseismic landslides, the area of ​​expanded landslides in each year, the area of ​​newly added landslides, the area of ​​dormant landslides, and the area of ​​vegetation restoration zones in the landslide activity recording dataset, and to obtain the time series of landslide activity intensity. The decay function construction module is used to fit the decay relationship of the landslide activity intensity over time based on the landslide activity intensity time series, and obtain the landslide activity intensity decay function; The activity intensity prediction module is used to calculate the year difference between the target prediction year and the year of the earthquake, input the year difference into the landslide activity intensity attenuation function, perform activity intensity prediction, and obtain the landslide activity intensity prediction value for the target prediction year.

Images