Earthquake forecasting method based on remote sensing interpretation earthquake data

Abstract

The invention discloses an earthquake forecasting method based on remote sensing interpretation earthquake data, and relates to the field of earthquake monitoring and evaluation. The method comprises the following steps: determining a to-be-monitored target area on a longitude and latitude map; configuring a space-time sampler for the target area; determining G abnormal intensities of pregnancy and earthquake based on a space-time sampler; constructing G space-time anomaly triples according to the G abnormal intensities of pregnancy and earthquake and the time window and the regular grid of the space-time sampler; performing density clustering on the G space-time anomaly triples based on the space-time positions represented by the space-time anchor points to obtain K high-risk anomalous bodies; constructing an anomalous body disaster grade map based on the K high-risk anomalous bodies; according to the method, the time continuity and the space influence range of the remote sensing interpretation parameter anomaly can be reflected at the same time, the high-priority anomaly position with the large coverage range and the long duration time can be identified, and an objective and quantifiable disaster assessment grading basis is provided for earthquake precursor forecasting.

Description

Technical Field

[0001] This invention relates to the field of earthquake prediction, specifically to an earthquake prediction method based on remote sensing interpretation of earthquake data. Background Technology

[0002] Current earthquake precursor monitoring based on remote sensing data commonly employs multi-parameter analysis, such as simultaneously using indicators like thermal infrared brightness temperature, crustal deformation rate, and vegetation stress index to identify anomalies. However, these multiple parameters are typically processed independently in their respective time and space, and anomaly determination relies on the statistical thresholds of each parameter, lacking a unified time window and spatial grid alignment mechanism.

[0003] Patent document CN119167050A discloses a method for extracting pre-earthquake anomaly information from multi-source remote sensing datasets for earthquake prediction; it solves the prediction instability problem caused by single-parameter analysis in traditional methods. Although similar existing technologies fuse anomaly results from multiple parameters, the intensity of the fused anomalies is still difficult to compare across regions and time periods because the input data is not standardized under a consistent spatiotemporal unit. More importantly, even if clustering algorithms group neighboring anomalies into clusters, the risk ranking still mainly relies on local statistics such as the number of anomalies, average intensity, or maximum value, failing to characterize the duration and spatial extent of an anomaly cluster. Therefore, it is difficult to establish an objective and quantitative earthquake disaster risk assessment system. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an earthquake prediction method based on remote sensing interpretation of earthquake data. By introducing spatiotemporal volume, it characterizes the combined scale of the temporal persistence and spatial influence range of high-risk anomalies, thereby solving the technical problems mentioned in the background.

[0005] To achieve the above objectives, the present invention provides the following technical solution: An earthquake prediction method based on remote sensing interpretation of earthquake data, the method comprising: S1. Determine the target area to be monitored on the latitude and longitude map; S2. Configure a spatiotemporal sampler for the target area; The spatiotemporal sampler includes: A time window that slides along the time axis, and a mesh generator that divides the target region into N×M regular meshes within the time window; S3. Determine the intensity of G earthquake-pregnancy anomalies based on the spatiotemporal sampler; S4. Based on the intensity of the G earthquake-induced anomalies, as well as the time window and regular grid of the spatiotemporal sampler, construct G spatiotemporal anomaly triples. Among them, the spatiotemporal anomaly triple includes: a spatiotemporal anchor point used to identify the unique spatiotemporal location of the regular grid within the time window; S5. Based on the spatiotemporal anchor points, perform density clustering on the G spatiotemporal anomaly triples to obtain K high-risk anomalies; S6. Based on K high-risk anomalies, construct an anomaly disaster level map.

[0006] In some embodiments, a spatiotemporal sampler is configured for the target region, including: S2-1. Set the length of the time window and the sliding step size on the time axis of the target area; S2-2. Set the spatial resolution of the grid generator on the latitude and longitude map of the target area; S2-3. Bind the set time window to the grid generator to form the spatiotemporal sampler.

[0007] In some of these embodiments, the intensity of G seismogenic anomalies is determined based on a spatiotemporal sampler, including: S3-1. Determine several remote sensing interpretation parameters to be monitored in the target area within any time window; S3-2. For each remote sensing interpretation parameter, the spatiotemporal sampler collects the parameter observation values ​​of the current time window and the historical same period window; S3-3. Calculate the remote sensing anomaly index within the regular grid based on the parameter observation values ​​of the current time window and the historical windows of the same period; S3-4. Traverse several remote sensing interpretation parameters, obtain the remote sensing anomaly index of each remote sensing interpretation parameter in the regular grid, and concatenate the corresponding remote sensing anomaly indexes into a remote sensing anomaly vector. S3-5. Calculate the Euclidean modulus of the remote sensing anomaly vector and use the Euclidean modulus as the intensity of the earthquake-induced anomaly within the current time window for the regular grid. S3-6. Slide J time windows and repeatedly construct N×M earthquake anomaly intensities within each time window until G earthquake anomaly intensities are determined; where G=N×M×J.

[0008] In some embodiments, G spatiotemporal anomaly triples are constructed based on G anomaly intensities, the time window of the spatiotemporal sampler, and a regular grid, including: S4-1. Based on the time window and the regular grid, construct J time indices and N×M coordinate indices respectively; S4-2. Select a target time index from the J time indices; S4-3. Using the target time index as the query, match its N×M coordinate indices within the time window to form the spatiotemporal anchor points of the regular grid on the target time index; S4-4. Using spatiotemporal anchor points as queries, match their seismic anomaly intensity in the regular grid to form spatiotemporal anomaly triples in the regular grid. S4-5. Traverse the J time indices and generate spatiotemporal anomaly triples in a loop until G spatiotemporal anomaly triples are obtained.

[0009] In some embodiments, spatiotemporal density clustering is performed on G spatiotemporal anomaly triples to obtain K high-risk anomalies, including: S5-1. Initialize the clustering parameters for spatiotemporal density clustering, including neighborhood radius and minimum number of triplets; S5-2. Based on the clustering parameters, mark the G spatiotemporal anomaly triples as core tuples to obtain a core tuple set. S5-3. In the core tuple set, spatiotemporal anomaly triples are aggregated through density reachability relations to obtain K high-risk anomalies.

[0010] In some embodiments, the G spatiotemporal anomaly triples are labeled with core tuples based on the clustering parameters to obtain a core tuple set, including: S5-2-1. Select an unprocessed spatiotemporal anomaly triplet from the G spatiotemporal anomaly triplets and mark it as the target triplet. S5-2-2 Calculate the spatiotemporal distance between the target triplet and the other G-1 spatiotemporal anomaly triplets; S5-2-3. Count the number of spatiotemporal anomalous triples whose spatiotemporal distance does not exceed the neighborhood radius, and mark them as the number of neighborhood triples of the target triple; S5-2-4. If the number of neighborhood triples is not less than the number of minimum triples, then mark the target triple as a core triple; otherwise, mark it as a non-core triple. S5-2-5. Traverse the G spatiotemporal anomaly triples to complete the labeling of all core tuples and obtain the core tuple set.

[0011] In some embodiments, based on the core tuple set, spatiotemporal anomalous triples are aggregated through density reachability relations to generate K high-risk anomalous entities, including: S5-3-1. Select a core tuple that has not yet been assigned to a high-risk anomaly from the core tuple set as the starting core tuple, and create a new high-risk anomaly. S5-3-2. Recursively search for all spatiotemporal anomalous triples whose density is reachable from the initial core tuple; Density reachability is defined as follows: there exists a path consisting of core tuples, and the spatiotemporal distance between any two adjacent spatiotemporal anomaly triples on the path does not exceed the neighborhood radius. S5-3-3: Assign the initial core tuple and all its density-reachable spatiotemporal anomaly triples to the newly created high-risk anomaly body; S5-3-4. Traverse the set of core tuples until all core tuples are assigned to the corresponding high-risk anomalies, and finally obtain K high-risk anomalies.

[0012] In some embodiments, an anomaly hazard level map is constructed based on K high-risk anomalies, including: S6-1. Calculate the spatiotemporal volume of K high-risk anomalies; S6-2. Arrange the K high-risk anomalies in descending order based on the spatiotemporal volume to generate an ordered anomaly list; S6-3. Calculate the time coverage length, spatial coverage area, high anomaly ratio and peak anomaly intensity for each high-risk anomaly in the ordered anomaly list. S6-4. Mark the extracted time coverage length, spatial coverage area, high anomaly ratio, and peak anomaly intensity on their respective high-risk anomalies in the ordered anomaly list to generate the anomaly disaster level map.

[0013] In some embodiments, the spatiotemporal volume of K high-risk anomalies is calculated, including: S6-1-1, Anchoring the spatiotemporal anomaly triplet in each high-risk anomaly; S6-1-2. Based on the anchored spatiotemporal anomaly triples, extract their temporal coverage length and spatial coverage area; S6-1-3. Multiply the time coverage length by the spatial coverage area to generate the spatiotemporal volume of the high-risk anomaly.

[0014] This invention provides an earthquake prediction method based on remote sensing interpretation of earthquake data, which has the following beneficial effects: This invention uses spatiotemporal density clustering to form high-risk anomalies that characterize the clustering features of earthquake-predicting anomalies. Furthermore, it uses spatiotemporal volume to represent the comprehensive coverage scale of each high-risk anomaly in the target area. This spatiotemporal volume is the product of temporal coverage length and spatial coverage area, thus simultaneously reflecting the temporal persistence and spatial influence range of remote sensing interpretation parameter anomalies, avoiding assessment biases caused by relying solely on local statistics such as the number of anomalies and average intensity. Moreover, by organizing high-risk anomalies in descending order using spatiotemporal volume as the main framework, it can identify high-priority anomaly locations with large coverage areas and long durations, providing an objective and quantifiable basis for disaster assessment and classification for earthquake precursor prediction. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating an earthquake prediction method based on remote sensing interpretation of earthquake data according to the present invention. Figure 2 This is a schematic diagram illustrating the process of constructing the intensity of earthquake-induced anomalies as described in this invention; Figure 3 This is a schematic diagram of the generation process of the spatiotemporal anomaly triple described in this invention; Figure 4 This is a schematic diagram of the aggregation process of the high-risk anomalies described in this invention; Figure 5 This is a schematic diagram illustrating the generation process of the anomaly disaster level map described in this invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] First, the prior art and related concepts involved in the embodiments of the present invention will be described: Remote sensing interpretation parameters refer to observable physical quantities extracted from satellite or airborne remote sensing data that are related to crustal stress-strain processes or changes in the surface environment. They are used to characterize abnormal precursor signals that may occur in the target area during the earthquake preparation stage. In this embodiment, the remote sensing interpretation parameters include thermal infrared brightness temperature (or thermal infrared index), crustal deformation rate, vegetation stress index, etc., which respectively reflect thermal anomalies, surface deformation anomalies, and ecological stress anomalies that may accompany the earthquake preparation process.

[0018] Example 1: Please refer to Figure 1 This invention provides an earthquake prediction method based on remote sensing interpretation of earthquake data, comprising the following steps: S1. Determine the target area to be monitored on the latitude and longitude map; S2. Configure a spatiotemporal sampler for the target area; The spatiotemporal sampler includes: A time window that slides along the time axis, and a mesh generator that divides the target region into N×M regular meshes within the time window; S3. Determine the intensity of G earthquake-pregnancy anomalies based on the spatiotemporal sampler; S4. Based on the intensity of the G earthquake-induced anomalies, as well as the time window and regular grid of the spatiotemporal sampler, construct G spatiotemporal anomaly triples. Among them, the spatiotemporal anomaly triple includes: a spatiotemporal anchor point used to identify the unique spatiotemporal location of the regular grid within the time window; S5. Based on the spatiotemporal anchor points, perform density clustering on the G spatiotemporal anomaly triples to obtain K high-risk anomalies; The high-risk anomaly is an anomaly cluster composed of multiple spatiotemporal anomaly triples that are adjacent and continuous in time and space, and is used to characterize spatiotemporal anomaly regions with potential seismic risk.

[0019] S6. Based on K high-risk anomalies, construct an anomaly disaster level map.

[0020] Therefore, this embodiment constructs a spatiotemporal sampler on the target area to sequentially generate earthquake-induced anomaly intensity, spatiotemporal anomaly triples, and high-risk anomalies, and finally forms an anomaly hazard level map, thereby realizing the complete process from remote sensing interpretation data to structured earthquake-induced anomaly risk characterization.

[0021] Example 2: See Figures 2 to 5 The technical solution of this embodiment 2 differs from that of embodiment 1 in that it discloses the specific implementation method of the above-mentioned execution steps.

[0022] Specifically, in this embodiment, step S2 includes: S2-1. Set the length of the time window and the sliding step size on the time axis of the target area; Wherein, the length of the time window is D days, D≥7; and the sliding step size is S days, S≤D.

[0023] S2-2. Set the spatial resolution of the grid generator on the latitude and longitude map of the target area; The spatial resolution is R meters × R meters, where R ∈ [100, 5000].

[0024] S2-3. Bind the set time window to the grid generator to form the spatiotemporal sampler.

[0025] Specifically, each time a time window is activated, the grid generator outputs N×M regular grids for the current time window.

[0026] This embodiment constructs a spatiotemporal sampler that slides on the time axis and is fixedly divided in the target area space by binding a time window and a grid generator, providing a unified spatiotemporal reference for the structured acquisition of remote sensing anomaly parameters.

[0027] Specifically, in this embodiment, step S3 includes: S3-1. Determine several remote sensing interpretation parameters to be monitored in the target area within any time window; In this embodiment, the remote sensing interpretation parameters include parameters such as thermal infrared index, crustal deformation rate, and vegetation stress index, which can reflect the multidimensional precursor response characteristics of the Earth system during the earthquake-pregnancy process.

[0028] S3-2. For each remote sensing interpretation parameter, the spatiotemporal sampler collects the parameter observation values ​​of the current time window and the historical same period window; S3-3. Calculate the remote sensing anomaly index within the regular grid based on the parameter observation values ​​of the current time window and the historical windows of the same period; The formula for calculating the remote sensing anomaly index is: ; in: Indicates remote sensing anomaly indicators; This represents the remote sensing interpretation parameters of a specific regular grid within the target area during the current time window; It is represented as the arithmetic mean of the remote sensing interpretation parameters of the regular grid over the same historical window of the past T years; This represents the standard deviation of the remote sensing interpretation parameters of the regular grid within the same historical window of the past T years; In this embodiment, T ≥ 5 is preferred; It should be noted that the remote sensing anomaly index in this embodiment is essentially a statistical anomaly intensity, rather than a simple numerical standardization.

[0029] Specifically, remote sensing anomaly indicators This directly reflects the degree of deviation of the currently observed remote sensing interpretation parameters from their long-term historical background parameters: When the absolute value of the remote sensing anomaly index is less than 2, it indicates that the remote sensing interpretation parameters of the target area are within the normal fluctuation range. When the absolute value of the remote sensing anomaly index is ≥2, it indicates that there are significant anomalies in the remote sensing interpretation parameters of the target area, which may be related to seismogenic activity. When the absolute value of the remote sensing anomaly index is greater than 3, it indicates that the remote sensing interpretation parameters of the target area have extremely strong anomalies and need to be closely monitored.

[0030] Furthermore, under a standard normal distribution, approximately 95% of the observations fall within the interval [-1.96, 1.96], therefore the remote sensing anomaly index... ≥2 can be considered a critical point of significant deviation from the normal state; while remote sensing anomaly indicators A value greater than 3 corresponds to an extremely low probability event (approximately 0.3%), which is usually considered a very strong anomaly and requires close monitoring. As can be seen, the remote sensing anomaly index in this embodiment not only achieves dimensionless numerical values, but also endows it with anomaly identification capabilities by introducing historical background contemporaneous window data.

[0031] S3-4. Traverse several remote sensing interpretation parameters, obtain the remote sensing anomaly index of each remote sensing interpretation parameter in the regular grid, and concatenate the corresponding remote sensing anomaly indexes into a remote sensing anomaly vector. S3-5. Calculate the Euclidean modulus of the remote sensing anomaly vector and use the Euclidean modulus as the intensity of the earthquake-induced anomaly within the current time window for the regular grid. Specifically, the Euclidean modulus is the L2 norm of the remote sensing anomaly direction. Therefore, the Euclidean modulus can comprehensively reflect the overall deviation level of multi-source anomalies in the current time window of the regular grid. The larger the modulus, the higher the probability that multiple precursor parameters will show significant anomalies at the same time. Therefore, it has clear physical significance and discrimination value as the intensity of earthquake-induced anomalies in this spatiotemporal unit.

[0032] S3-6. Slide J time windows and repeatedly construct N×M earthquake anomaly intensities within each time window until G earthquake anomaly intensities are determined; where G=N×M×J.

[0033] This embodiment is based on the spatiotemporal sampler to perform historical data statistics on multi-source remote sensing interpretation parameters within each time window, calculate the remote sensing anomaly index of the regular grid, splice them into anomaly vector, and use its Euclidean modulus as the intensity of the earthquake-induced anomaly. Finally, G earthquake-induced anomaly intensities covering all time windows and regular grids are generated, thereby providing a unified index reflecting the degree of multi-source remote sensing anomalies for each spatiotemporal unit.

[0034] Specifically, in this embodiment, step S4 includes: S4-1. Based on the time window and the regular grid, construct J time indices and N×M coordinate indices respectively; Specifically, the time index is used to uniquely identify the position of each time window on the time axis. It is constructed according to the activation order of the time windows, assigning a unique time sequence identifier to each time window, which is then used as the time index.

[0035] Meanwhile, the coordinate index is used to uniquely identify the spatial location of each regular grid in the target area. It is constructed according to the row and column position or geographic coordinates of the regular grid on the latitude and longitude map, and assigns a unique spatial identifier to each regular grid, thus serving as a spatial index.

[0036] S4-2. Select a target time index from the J time indices; S4-3. Using the target time index as the query, match its N×M coordinate indices within the time window to form the spatiotemporal anchor points of the regular grid on the target time index; S4-4. Using spatiotemporal anchor points as queries, match their seismic anomaly intensity in the regular grid to form spatiotemporal anomaly triples in the regular grid. S4-5. Traverse the J time indices and generate spatiotemporal anomaly triples in a loop until G spatiotemporal anomaly triples are obtained.

[0037] This embodiment constructs a spatiotemporal anchor point by combining time index and coordinate index, and associates the corresponding earthquake anomaly intensity with the anchor point to form G spatiotemporal anomaly triples containing spatiotemporal location and anomaly intensity, thereby giving each earthquake anomaly intensity a clear spatiotemporal association.

[0038] Specifically, in this embodiment, step S5 includes: S5-1. Initialize the clustering parameters for spatiotemporal density clustering, including neighborhood radius and minimum number of triplets; S5-2. Based on the clustering parameters, mark the G spatiotemporal anomaly triples as core tuples to obtain a core tuple set. S5-3. In the core tuple set, spatiotemporal anomaly triples are aggregated through density reachability relations to obtain K high-risk anomalies.

[0039] This embodiment obtains K high-risk anomalies through density clustering, thereby achieving preliminary identification of high-risk anomalies by clustering discrete spatiotemporal anomaly triples.

[0040] Furthermore, step S5-2 specifically includes: S5-2-1. Select an unprocessed spatiotemporal anomaly triplet from the G spatiotemporal anomaly triplets and mark it as the target triplet. S5-2-2 Calculate the spatiotemporal distance between the target triplet and the other G-1 spatiotemporal anomaly triplets; The formula for calculating the spatiotemporal distance is: ; in: It represents the spatiotemporal distance, used to measure the comprehensive deviation of the target triplet from the spatiotemporally anomalous triplet in time and space, and serves as the basis for density clustering judgment; This represents the time difference between the target triplet and the corresponding time window in the spatiotemporal anomaly triplet. The normalized standard constant representing the time difference takes a value equal to the length of the time window; therefore, The normalized value representing the time difference reflects the relative scale of the time difference with respect to the length of the time window; This represents the latitude and longitude coordinate distance between the target triplet and the corresponding grid in the spatiotemporal anomaly triplet. The normalized standard constant representing the coordinate distance takes a value equal to the standard side length of the grid; therefore, The normalized value representing the latitude and longitude coordinate distance reflects the relative scale of the latitude and longitude coordinate distance with respect to the standard side length of the grid. S5-2-3. Count the number of spatiotemporal anomalous triples whose spatiotemporal distance does not exceed the neighborhood radius, and mark them as the number of neighborhood triples of the target triple; S5-2-4. If the number of neighborhood triples is not less than the number of minimum triples, then mark the target triple as a core triple; otherwise, mark it as a non-core triple. S5-2-5. Traverse the G spatiotemporal anomaly triples to complete the labeling of all core tuples and obtain the core tuple set.

[0041] Therefore, this embodiment, based on spatiotemporal distance measurement, identifies spatiotemporal anomaly triples with sufficient local density as core tuples by comparing neighborhood counts with thresholds, thus providing a seed set of core tuples for the allocation of high-risk anomalies.

[0042] Furthermore, step S5-3 specifically includes: S5-3-1. Select a core tuple that has not yet been assigned to a high-risk anomaly from the core tuple set as the starting core tuple, and create a new high-risk anomaly. S5-3-2. Recursively search for all spatiotemporal anomalous triples whose density is reachable from the initial core tuple; Density reachability is defined as follows: there exists a path consisting of core tuples, and the spatiotemporal distance between any two adjacent spatiotemporal anomaly triples on the path does not exceed the neighborhood radius. S5-3-3: Assign the initial core tuple and all its density-reachable spatiotemporal anomaly triples to the newly created high-risk anomaly body; S5-3-4. Traverse the set of core tuples until all core tuples are assigned to their corresponding high-risk anomalies, and finally obtain K high-risk anomalies.

[0043] Therefore, this embodiment starts with the core tuple and, based on the density reachability relationship, groups spatially adjacent and temporally continuous spatiotemporal anomaly triples into connected regions, forming K high-risk anomaly entities with clear spatiotemporal boundaries. This clusters discrete spatiotemporal anomaly triples into entities with aggregation and continuity that can reflect earthquake-pregnancy anomalies.

[0044] Specifically, in this embodiment, step S6 includes: S6-1. Calculate the spatiotemporal volume of K high-risk anomalies; S6-2. Arrange the K high-risk anomalies in descending order based on the spatiotemporal volume to generate an ordered anomaly list; S6-3. Calculate the time coverage length, spatial coverage area, high anomaly ratio and peak anomaly intensity for each high-risk anomaly in the ordered anomaly list. For example: The time coverage length represents the duration of the high-risk anomaly on the time axis. It is calculated by subtracting the minimum value from the maximum value of the start time of the time window corresponding to all spatiotemporal anomaly triples within the anomaly. The spatial coverage area represents the area of ​​influence of the high-risk anomaly in the target region, and is statistically represented by the area of ​​the polygon enclosed by all regular grids within the anomaly. Furthermore, area calculations can treat all regular grids within a high-risk anomaly as unit cells, with the spatial coverage area equal to the number of all regular grids multiplied by the area of ​​a single grid cell.

[0045] The high anomaly ratio refers to the proportion of spatiotemporal anomaly triples with anomaly intensity exceeding a preset threshold (e.g., |Z|>3) in the high-risk anomaly body, which is used to indicate the degree of concentration of anomalies within it. The peak anomaly intensity represents the maximum value of the seismogenic anomaly intensity of all spatiotemporal anomaly ternaries in the high-risk anomaly body.

[0046] S6-4. Mark the extracted time coverage length, spatial coverage area, high anomaly ratio, and peak anomaly intensity on their respective high-risk anomalies in the ordered anomaly list to generate the anomaly disaster level map.

[0047] It should be noted that the aforementioned anomaly disaster level map is a structured risk characterization framework constructed using high-risk anomalies as the basic unit. The map uses spatiotemporal volume (i.e., the coverage scale of the anomaly in time and space) as the core sorting criterion, arranges all high-risk anomalies in descending order, and associates four characteristic parameters with each anomaly: its temporal coverage length, spatial coverage area, high anomaly proportion, and peak anomaly intensity. This forms a structured risk assessment map with coverage scale as the main body and multidimensional characteristics as branches.

[0048] Furthermore, step S56-1 specifically includes: S6-1-1, Anchoring the spatiotemporal anomaly triplet in each high-risk anomaly; S6-1-2. Based on the anchored spatiotemporal anomaly triples, extract their temporal coverage length and spatial coverage area; S6-1-3. Multiply the time coverage length by the spatial coverage area to generate the spatiotemporal volume of the high-risk anomaly.

[0049] Specifically, the spatiotemporal volume represents the coverage scale of the high-risk anomaly in time and space, and is numerically equal to the product of the temporal coverage length (duration) and the spatial coverage area (area of ​​influence), used to quantify the magnitude of its potential risk.

[0050] In summary, this invention identifies K high-risk anomalies that characterize the clustering of earthquake-predicting anomalies by performing spatiotemporal density clustering on G spatiotemporal anomaly triples. Based on the spatiotemporal anomaly triples contained in each high-risk anomaly, its temporal coverage length and spatial coverage area are extracted, and their product is calculated to obtain the spatiotemporal volume. This spatiotemporal volume serves as a quantitative indicator of the overall coverage scale, reflecting both the temporal persistence and spatial impact range of remote sensing interpretation parameter anomalies, effectively overcoming the risk assessment bias caused by relying solely on local statistics such as the number of anomalies and average intensity. Furthermore, the K high-risk anomalies are organized in descending order using spatiotemporal volume as the backbone to construct an anomaly hazard level map, thereby automatically identifying high-priority anomaly locations with large coverage areas and long durations, providing an objective and quantifiable basis for hazard assessment and classification for earthquake precursor prediction.

[0051] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means.

[0052] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An earthquake prediction method based on remote sensing interpretation of earthquake data, characterized in that, include: S1. Determine the target area to be monitored on the latitude and longitude map; S2. Configure a spatiotemporal sampler for the target area; The spatiotemporal sampler includes: A time window that slides along the time axis, and a mesh generator that divides the target region into N×M regular meshes within the time window; S3. Determine the intensity of G earthquake-pregnancy anomalies based on the spatiotemporal sampler; S4. Based on the intensity of the G earthquake-induced anomalies, as well as the time window and regular grid of the spatiotemporal sampler, construct G spatiotemporal anomaly triples. Among them, the spatiotemporal anomaly triple includes: a spatiotemporal anchor point used to identify the unique spatiotemporal location of the regular grid within the time window; S5. Based on the spatiotemporal anchor points, perform density clustering on the G spatiotemporal anomaly triples to obtain K high-risk anomalies; S6. Based on K high-risk anomalies, construct an anomaly disaster level map.

2. The earthquake prediction method based on remote sensing interpretation of earthquake data according to claim 1, characterized in that, Configure a spatiotemporal sampler for the target region, including: S2-1. Set the length of the time window and the sliding step size on the time axis of the target area; S2-2. Set the spatial resolution of the grid generator on the latitude and longitude map of the target area; S2-3. Bind the set time window to the grid generator to form the spatiotemporal sampler.

3. The earthquake prediction method based on remote sensing interpretation of earthquake data according to claim 2, characterized in that, The intensity of G seismogenic anomalies was determined based on a spatiotemporal sampler, including: S3-1. Determine several remote sensing interpretation parameters to be monitored in the target area within any time window; S3-2. For each remote sensing interpretation parameter, the spatiotemporal sampler collects the parameter observation values ​​of the current time window and the historical same period window; S3-3. Calculate the remote sensing anomaly index within the regular grid based on the parameter observation values ​​of the current time window and the historical windows of the same period; S3-4. Traverse several remote sensing interpretation parameters, obtain the remote sensing anomaly index of each remote sensing interpretation parameter in the regular grid, and concatenate the corresponding remote sensing anomaly indexes into a remote sensing anomaly vector. S3-5. Calculate the Euclidean modulus of the remote sensing anomaly vector and use the Euclidean modulus as the intensity of the earthquake-induced anomaly within the current time window for the regular grid. S3-6. Slide J time windows and repeatedly construct N×M earthquake anomaly intensities within each time window until G earthquake anomaly intensities are determined; where G=N×M×J.

4. The earthquake prediction method based on remote sensing interpretation of earthquake data according to claim 1, characterized in that, Based on the intensity of G seismogenic anomalies, and the time window and regular grid of the spatiotemporal sampler, G spatiotemporal anomaly triples are constructed, including: S4-1. Based on the time window and the regular grid, construct J time indices and N×M coordinate indices respectively; S4-2. Select a target time index from the J time indices; S4-3. Using the target time index as the query, match its N×M coordinate indices within the time window to form the spatiotemporal anchor points of the regular grid on the target time index; S4-4. Using spatiotemporal anchor points as queries, match their seismic anomaly intensity in the regular grid to form spatiotemporal anomaly triples in the regular grid. S4-5. Traverse the J time indices and generate spatiotemporal anomaly triples in a loop until G spatiotemporal anomaly triples are obtained.

5. The earthquake prediction method based on remote sensing interpretation of earthquake data according to claim 1, characterized in that, Spatiotemporal density clustering was performed on the G spatiotemporal anomaly triples to obtain K high-risk anomalies, including: S5-1. Initialize the clustering parameters for spatiotemporal density clustering, including neighborhood radius and minimum number of triplets; S5-2. Based on the clustering parameters, mark the G spatiotemporal anomaly triples as core tuples to obtain a core tuple set. S5-3. In the core tuple set, spatiotemporal anomaly triples are aggregated through density reachability relations to obtain K high-risk anomalies.

6. The earthquake prediction method based on remote sensing interpretation of earthquake data according to claim 5, characterized in that, Based on the clustering parameters, the G spatiotemporal anomaly triples are labeled with core tuples to obtain a core tuple set, including: S5-2-1. Select an unprocessed spatiotemporal anomaly triplet from the G spatiotemporal anomaly triplets and mark it as the target triplet. S5-2-2 Calculate the spatiotemporal distance between the target triplet and the other G-1 spatiotemporal anomaly triplets; S5-2-3. Count the number of spatiotemporal anomalous triples whose spatiotemporal distance does not exceed the neighborhood radius, and mark them as the number of neighborhood triples of the target triple; S5-2-4. If the number of neighborhood triples is not less than the number of minimum triples, then mark the target triple as a core triple; otherwise, mark it as a non-core triple. S5-2-5. Traverse the G spatiotemporal anomaly triples to complete the labeling of all core tuples and obtain the core tuple set.

7. The earthquake prediction method based on remote sensing interpretation of earthquake data according to claim 6, characterized in that, Based on the core tuple set, spatiotemporal anomaly triples are aggregated through density reachability relations to generate K high-risk anomaly entities, including: S5-3-1. Select a core tuple that has not yet been assigned to a high-risk anomaly from the core tuple set as the starting core tuple, and create a new high-risk anomaly. S5-3-2. Recursively search for all spatiotemporal anomalous triples whose density is reachable from the initial core tuple; Density reachability is defined as follows: there exists a path consisting of core tuples, and the spatiotemporal distance between any two adjacent spatiotemporal anomaly triples on the path does not exceed the neighborhood radius. S5-3-3: Assign the initial core tuple and all its density-reachable spatiotemporal anomaly triples to the newly created high-risk anomaly body; S5-3-4. Traverse the set of core tuples until all core tuples are assigned to the corresponding high-risk anomalies, and finally obtain K high-risk anomalies.

8. The earthquake prediction method based on remote sensing interpretation of earthquake data according to claim 1, characterized in that, Based on K high-risk anomalies, an anomaly hazard level map is constructed, including: S6-1. Calculate the spatiotemporal volume of K high-risk anomalies; S6-2. Arrange the K high-risk anomalies in descending order based on the spatiotemporal volume to generate an ordered anomaly list; S6-3. Calculate the time coverage length, spatial coverage area, high anomaly ratio and peak anomaly intensity for each high-risk anomaly in the ordered anomaly list. S6-4. Mark the extracted time coverage length, spatial coverage area, high anomaly ratio, and peak anomaly intensity on their respective high-risk anomalies in the ordered anomaly list to generate the anomaly disaster level map.

9. The earthquake prediction method based on remote sensing interpretation of earthquake data according to claim 8, characterized in that, Calculate the spatiotemporal volume of K high-risk anomalies, including: S6-1-1, Anchoring the spatiotemporal anomaly triplet in each high-risk anomaly; S6-1-2. Based on the anchored spatiotemporal anomaly triples, extract their temporal coverage length and spatial coverage area; S6-1-3. Multiply the time coverage length by the spatial coverage area to generate the spatiotemporal volume of the high-risk anomaly.

Images