A short-term strong earthquake occurrence time prediction method, device, medium and equipment

Abstract

The application discloses a short-term strong earthquake occurrence time prediction method and device, medium and equipment, and relates to the technical field of earthquake prediction. The method performs standardization preprocessing on atmospheric tide gravity signals in a preset time window, inputs a plurality of target days into a prediction model constructed in advance after detecting an atmospheric tide gravity anomaly event, and obtains the probability of strong earthquakes occurring in each target day after the atmospheric tide gravity anomaly event. The prediction model is trained according to the statistical correlation characteristics between the atmospheric tide gravity anomaly event occurrence time and the strong earthquake occurrence time. The strong earthquake refers to an earthquake with a magnitude greater than a preset threshold. According to the probability of strong earthquakes occurring in each target day after the anomaly, the strong earthquake occurrence time prediction result is determined. The method realizes quantitative probability prediction of the strong earthquake occurrence time, and has the advantages of being interpretable, calibratable and online updatable.

Description

Technical Field

[0001] This invention relates to the field of earthquake prediction technology, and in particular to a method, device, medium, and equipment for predicting the occurrence time of short-term strong earthquakes. Background Technology

[0002] Over the past few decades, scientists have conducted extensive research in the field of earthquake prediction. Although earthquake prediction has been found to be "far more difficult than imagined," a wealth of observational evidence suggests that earthquakes are not entirely unpredictable. Strong earthquakes often exhibit certain patterns and precursory phenomena, making earthquake prediction theoretically possible.

[0003] In terms of existing technologies, various techniques have been attempted both domestically and internationally for short-term prediction of strong earthquakes, but they are still mainly based on qualitative assessments. Traditional approaches rely heavily on empirical analysis and comprehensive consultation of multi-source precursors (such as seismic activity, crustal deformation, underground fluids, electromagnetic fields, and meteorological phenomena) to provide judgments on dangerous time windows or regions. However, the lack of unified quantitative standards results in insufficient accuracy in the judgment of strong earthquakes using existing technologies. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, device, medium, and equipment for predicting the occurrence time of short-term strong earthquakes, which can improve the accuracy of strong earthquake occurrence time prediction.

[0005] The present invention adopts the following technical solution:

[0006] This invention provides a method for predicting the occurrence time of short-term strong earthquakes, including:

[0007] Acquire the atmospheric tidal gravity signal sequence within a preset time window, and detect atmospheric tidal gravity anomaly events occurring within the preset time window based on the atmospheric tidal gravity signal sequence.

[0008] After an atmospheric tidal gravity anomaly event is detected, multiple target days are input into a pre-built prediction model to obtain the probability of a strong earthquake occurring on each target day after the atmospheric tidal gravity anomaly event. The prediction model is trained based on the statistical correlation characteristics between the occurrence time of the atmospheric tidal gravity anomaly event and the occurrence time of the strong earthquake. A strong earthquake is defined as an earthquake with a magnitude greater than 6.

[0009] Based on the probability of a strong earthquake occurring within each target day, the predicted time of occurrence of the strong earthquake is determined.

[0010] Optionally, the process of building the prediction model includes:

[0011] Acquire multiple atmospheric tidal and gravity anomaly events detected over a historical period;

[0012] For any given atmospheric tidal gravity anomaly event, search the earthquake catalog for the most recent strong earthquake that occurred after the event; the earthquake catalog includes the occurrence time of each earthquake.

[0013] Calculate the time difference between the occurrence time of the sample atmospheric tidal gravity anomaly event and the occurrence time of the corresponding strong earthquake;

[0014] A prediction model is constructed based on the time difference values ​​corresponding to atmospheric tidal gravity anomalies in all samples.

[0015] Optionally, a prediction model is constructed based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events, including:

[0016] Based on the time difference values ​​corresponding to atmospheric tidal gravity anomaly events in all samples, the frequency of occurrence of different time differences was statistically analyzed.

[0017] Based on the frequency of occurrence of different time differences, the conditional probability distribution of strong earthquakes relative to precursor anomalies is determined; the conditional probability distribution of strong earthquakes relative to precursor anomalies represents the probability of a strong earthquake occurring within the number of days corresponding to different time differences after the occurrence of a sample atmospheric tidal gravity anomaly event.

[0018] Based on the conditional probability distribution, the cumulative probability distribution is determined, and the cumulative probability distribution is used to fit and model the prediction model.

[0019] Optionally, a prediction model is obtained by fitting the model using the cumulative distribution probability, including:

[0020] A prediction model is obtained by fitting and modeling the cumulative distribution probability using a decision tree.

[0021] Optionally, a prediction model is obtained by fitting and modeling the cumulative distribution probability using a decision tree, including:

[0022] By using different time difference intervals as input features of the prediction model, and through iterative learning of the relationship between time and whether a strong earthquake event will occur after an atmospheric tidal gravity anomaly event, the prediction model is obtained.

[0023] Optionally, based on the probability of a strong earthquake occurring within each target day, the predicted time of occurrence of the strong earthquake is determined, including:

[0024] Based on the probability of a strong earthquake occurring within each target day, the probability of a strong earthquake occurring in multiple time intervals after an atmospheric tidal gravity anomaly event is determined.

[0025] The prediction of the occurrence time of a strong earthquake is determined based on the probability of a strong earthquake occurring in multiple time intervals.

[0026] Optionally, based on the atmospheric tidal gravity signal sequence, atmospheric tidal gravity anomaly events occurring within a preset time window are detected, including:

[0027] By removing errors caused by external factors from the atmospheric tidal gravity signal sequence, a stationary residual sequence is obtained.

[0028] Calculate the median absolute deviation of the stationary residual sequence, and based on the median absolute deviation, calculate the standardized residual corresponding to each time point within the preset time window;

[0029] If the standardized residuals at all times within a continuously preset first time period are greater than a preset threshold, then it is determined that there is an atmospheric tidal gravity anomaly event in that first time period.

[0030] If the interval between two atmospheric tidal gravity anomaly events is less than the second time period, then the two atmospheric tidal gravity anomaly events will be merged into a single atmospheric tidal gravity anomaly event.

[0031] This invention provides a device for predicting the occurrence time of short-term strong earthquakes, comprising:

[0032] The detection module is used to acquire the atmospheric tidal gravity signal sequence within a preset time window, and to detect atmospheric tidal gravity anomaly events occurring within the preset time window based on the atmospheric tidal gravity signal sequence.

[0033] The prediction module is used to input multiple target days into a pre-built prediction model after an atmospheric tidal gravity anomaly event is detected, and obtain the probability of a strong earthquake occurring on each target day after the atmospheric tidal gravity anomaly event. The prediction model is trained based on the statistical correlation features between the occurrence time of the atmospheric tidal gravity anomaly event and the occurrence time of the strong earthquake. A strong earthquake is defined as an earthquake with a magnitude greater than 6.

[0034] The determination module is used to determine the predicted time of occurrence of strong earthquakes based on the probability of strong earthquakes occurring within each target day.

[0035] The present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for predicting the occurrence time of short-term strong earthquakes.

[0036] The present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the above-mentioned method for predicting the occurrence time of short-term strong earthquakes.

[0037] The above-mentioned at least one technical solution adopted in this invention can achieve the following beneficial effects:

[0038] This invention constructs a predictive model based on the correlation between the timing of atmospheric tidal gravity anomalies and the timing of strong earthquakes. By monitoring atmospheric tidal gravity signals in real time, when an atmospheric tidal gravity anomaly is detected, the predictive model predicts the probability of a strong earthquake occurring within a certain number of days. This invention transforms the "complex and ambiguous problem" of traditional strong earthquake prediction into a "targeted, quantitative, and verifiable problem." Its core logic is an upgrade from finding anomalies to using anomalies to find temporal patterns. This not only conforms to the objective laws of geophysical phenomena but also follows the scientific logic of machine learning, thus effectively improving the accuracy of strong earthquake occurrence time prediction. Attached Figure Description

[0039] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0040] Figure 1 This is a schematic diagram of a method for predicting the occurrence time of short-term strong earthquakes provided by the present invention;

[0041] Figure 2 A schematic diagram showing the degree of agreement between the probability distribution predicted by the prediction model provided by this invention and the actual distribution;

[0042] Figure 3 A schematic diagram of a computer device for predicting the occurrence time of short-term strong earthquakes, provided by the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0044] Currently, countries worldwide monitor various types of earthquake precursor signals, including nearly a hundred methods across more than a dozen categories, such as seismic activity anomalies, crustal deformation, changes in groundwater levels and radon content, geoelectric and geomagnetic anomalies, stress-strain changes, meteorological and atmospheric phenomena, triggering factors, and macroscopic anomalies. Among these precursors, changes in the gravity field are considered a crucial physical quantity directly reflecting the mass migration and tectonic movements of the Earth's crust. Since the application of high-precision gravimeters to earthquake monitoring in the 1960s and 70s, studies have reported recording significant gravity anomalies before numerous strong earthquakes. With further research, researchers have gradually recognized that, in addition to gravity changes caused by traditional solid tides and crustal deformation, periodic disturbances in atmospheric mass distribution also have a measurable impact on surface gravity; this type of signal is called atmospheric tidal gravity. In geophysics, atmospheric tidal gravity refers to the micro-Gal-level signal formed by the spatiotemporal redistribution of atmospheric mass and periodic changes in surface air column pressure caused by atmospheric tides, acting on the surface gravity field through two pathways: direct gravity and elastic loading deformation. Its main energy is concentrated in the S1 (diurnal tide) and S2 (semi-diurnal tide) frequency bands, and it can be stably identified in superconducting gravimeter records. Existing research indicates that in certain regions and tectonic environments, celestial tidal forces and the resulting tidal stress modulation may push faults closer to their rupture critical state, thus statistically increasing the probability of earthquakes. Simultaneously, tidal stress and atmospheric parameter anomalies also show a certain correlation. For example, a study based on data from six earthquakes of magnitude 6 or higher in China during 2020-2021 found that during periods of significant tidal force influence, multiple parameters such as atmospheric temperature, gravitational potential height, and humidity on upper-level isobaric surfaces exhibited abnormal fluctuations. This suggests that celestial tidal forces may affect atmospheric parameters through Earth-atmosphere coupling processes during the pre-earthquake phase, indirectly reflecting adjustments in the tectonic stress field.

[0045] In recent years, with the advancement of high-precision gravity observation instruments, atmospheric tidal gravity observation has seen new developments. The Atmospheric Tidal Gravimeter (ATG) can cover a frequency band from static to 0.4 Hz with a resolution of 0.01 μGal, recording gravity disturbance signals with amplitudes of approximately 1-10 μGal and periods of 100-1000 s. It features low drift, high stability, and strong anti-interference capabilities, and is therefore a key instrument. Through matrix-deployed observation networks, researchers have successfully captured short-term temporary anomalies in the gravity field before several strong earthquakes. These anomalies reproduce the dynamic evolution of strong earthquakes from gestation to occurrence, and are considered to reflect the physical mechanism of "basic stability—locked-in energy storage—earthquake quiescence—energy release," confirming the potential reference value of atmospheric tidal gravity anomalies in short-term earthquake forecasting.

[0046] In summary, while short-term earthquake prediction faces numerous challenges, the accumulated achievements in precursor research and the development of new technologies provide a scientific foundation. Among these, atmospheric tidal gravity signals, as an emerging precursor carrier, have shown a close correlation with strong earthquake events in observations from 2021 to 2024. This lays the foundation for this application to further propose innovative methods and quantitatively characterize the relationship between precursor anomalies and the timing of major earthquakes.

[0047] To improve scientific rigor, researchers have introduced statistical methods, such as Molchan error plots, to examine the predictability of different signals for earthquake timing. However, these methods are primarily used for posterior evaluation and are difficult to directly translate into prospective probabilistic predictions. In recent years, machine learning and data mining methods have been widely applied to earthquake prediction research. Some models (such as decision trees, support vector machines, and long short-term memory networks) can identify pre-earthquake anomaly patterns or predict the probability of regional earthquakes to some extent. However, due to the scarcity of large earthquake samples and strong regional differences, these methods often face problems of insufficient generalization ability and lack of interpretability. Meanwhile, atmospheric, thermal anomalies, and ionospheric TEC (tectonics response) observational factors have also been attempted to be introduced into pre-earthquake analysis, but these largely remain at the qualitative identification level, and a direct quantitative relationship between them and earthquake timing has not yet been established. In summary, while existing methods have made some progress in precursor identification and statistical evaluation, they mostly remain at the level of qualitative judgment or post-event evaluation, and still lack quantitative models that can directly provide the probability distribution of strong earthquake occurrence time based on precursor anomalies. This gap is precisely the core problem that this invention aims to solve.

[0048] In view of the limitations of the prior art, this invention aims to solve the long-standing technical problem of predicting the short-term occurrence time of strong earthquakes. Specifically, this invention focuses on how to quantitatively estimate the probability of a strong earthquake occurring in different time periods after the detection of atmospheric tidal gravity anomalies. This problem has not been fully solved in previous studies—traditional methods mostly make qualitative or empirical judgments on "whether there is an earthquake risk in the near future," lacking a clear temporal probability model; while existing statistical models are usually based on historical earthquake sequences and fail to make full use of precursory physical information. Therefore, based on this, this application provides a method, device, medium, and equipment for predicting the short-term occurrence time of strong earthquakes. The purpose of this invention is to establish a prediction framework based on the statistical law of the time difference between precursor anomalies and earthquakes, providing a model that takes the occurrence time of precursor anomalies as input and outputs the probability of future strong earthquakes, helping earthquake forecasters and decision-makers to more rationally assess the risk of impending earthquakes; precursor anomalies represent atmospheric tidal gravity anomalies.

[0049] The core technical problem this invention aims to solve is how to accurately predict the temporal probability distribution of strong earthquakes using atmospheric tidal gravity precursor information (atmospheric tidal gravity anomaly events). This involves innovations at two levels: (1) extracting the statistical correlation characteristics between the occurrence time of atmospheric tidal gravity anomaly events and the occurrence time of strong earthquakes; and (2) constructing a model or algorithm that can use this correlation for forecasting. Solving this problem requires overcoming challenges such as difficulties in data matching, scarce samples, and ensuring the physical rationality and practical value of the probability prediction results. This invention will provide a new quantitative tool for short-term earthquake prediction, promoting the transformation of earthquake prediction methods from traditional empirical judgment to scientific prediction based on data and probability.

[0050] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0051] Figure 1 This is a schematic diagram of a method for predicting the occurrence time of short-term strong earthquakes according to the present invention, which specifically includes the following steps:

[0052] S101, acquire the atmospheric tidal gravity signal sequence within a preset time window, and detect atmospheric tidal gravity anomaly events occurring within the preset time window based on the atmospheric tidal gravity signal sequence.

[0053] The preset time window can be 14-30 days, and the atmospheric tidal gravity signal sequence within the preset time window can be atmospheric tidal gravity signals collected continuously for 14-30 days.

[0054] When it is necessary to predict the timing of a strong earthquake, atmospheric tidal gravity signals collected in real time from historical moments to the current moment are acquired; the time interval between historical moments and the current moment is a preset time window. These atmospheric tidal gravity signals can be collected in real time using a high-precision atmospheric tidal gravimeter.

[0055] In one embodiment, detecting atmospheric tidal gravity anomaly events occurring within a preset time window based on an atmospheric tidal gravity signal sequence includes: removing errors caused by external factors from the atmospheric tidal gravity signal sequence to obtain a stationary residual sequence; calculating the median absolute deviation of the stationary residual sequence, and calculating the standardized residual corresponding to each moment within the preset time window based on the median absolute deviation; if the standardized residual of all moments within a consecutive preset first time period is greater than a preset threshold, then it is determined that there is an atmospheric tidal gravity anomaly event in the first time period; if the interval between two atmospheric tidal gravity anomaly events is less than a second time period, then the two atmospheric tidal gravity anomaly events are merged into the same atmospheric tidal gravity anomaly event.

[0056] Specifically, errors caused by external factors in the atmospheric tidal gravity signal sequence are removed to obtain a stationary residual sequence. This includes sequentially performing de-mutation processing, detrending processing, bandpass filtering, pressure correction, and removal of major diurnal / semi-diurnal components on all atmospheric tidal gravity signals in the sequence to obtain a stationary residual sequence. The stationary residual sequence is obtained by removing tidal background and pressure effects from the original atmospheric tidal gravity signal sequence to obtain short-term transient disturbances.

[0057] The first time period can be 120 minutes, and the second time period can be 240 minutes. The median absolute deviation (MAD) of the stationary residual sequence is calculated, and based on the MAD, the standardized residual for each time step in the stationary residual sequence is calculated. If the standardized residual for all time steps within a consecutive 120 minutes is greater than a preset threshold, then an atmospheric tidal gravity anomaly event is determined to exist in that time period; if the interval between two anomalies is less than 240 minutes, they are merged into a single anomaly event.

[0058] Standardized residuals can measure the degree of deviation of observed values ​​from the background field. The preset threshold can be 3 times the normal fluctuation range.

[0059] S102, after detecting an atmospheric tidal gravity anomaly event, multiple target days are input into the pre-built prediction model to obtain the probability of a strong earthquake occurring within each target day after the atmospheric tidal gravity anomaly event. The prediction model is trained based on the statistical correlation characteristics between the occurrence time of the atmospheric tidal gravity anomaly event and the occurrence time of the strong earthquake. A strong earthquake refers to an earthquake with a magnitude greater than 6.

[0060] In one embodiment, the process of building a predictive model includes the following steps:

[0061] S201, acquires multiple atmospheric tidal gravity anomaly events detected over a historical period.

[0062] The atmospheric tidal gravity anomaly events in the sample carry a timestamp, which indicates the time when the atmospheric tidal gravity anomaly events occurred.

[0063] S202, for any sample atmospheric tidal gravity anomaly event, search the earthquake catalog for the most recent strong earthquake occurrence time after the sample atmospheric tidal gravity anomaly event occurred; the earthquake catalog includes the occurrence time of each earthquake.

[0064] The earthquake catalog provides information such as the time of occurrence, epicenter location, and magnitude of each earthquake (especially strong earthquakes of magnitude 6 or above). It is necessary to ensure that the time of occurrence of atmospheric tidal and gravity anomalies in the sample is consistent with the time base of the earthquake catalog, the time zone is unified, and each sample's atmospheric tidal and gravity anomaly is accurately associated with temporally adjacent earthquake events.

[0065] The sample atmospheric tidal gravity anomaly events can be scanned in chronological order: for each sample atmospheric tidal gravity anomaly event, the most recent strong earthquake that occurred after it is found in the earthquake catalog, and the occurrence time of the sample atmospheric tidal gravity anomaly event is matched with the occurrence time of the strong earthquake.

[0066] S203, calculate the time difference between the occurrence time of the sample atmospheric tidal gravity anomaly event and the occurrence time of the corresponding strong earthquake.

[0067] Based on the above process, data records of multiple sets of atmospheric tidal gravity anomaly events and corresponding time differences were obtained; among them, the time difference represents the time interval from the occurrence time of the sample atmospheric tidal gravity anomaly event to the occurrence of the strong earthquake, and the time difference is calculated in days.

[0068] S204. A prediction model is constructed based on the time difference values ​​corresponding to atmospheric tidal gravity anomaly events in all samples.

[0069] In one embodiment, a prediction model is constructed based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events, including: calculating the frequency of occurrence of different time differences based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events; determining the conditional probability distribution of strong earthquake occurrence relative to precursor anomalies based on the frequency of occurrence of different time differences; the conditional probability distribution of strong earthquake occurrence relative to precursor anomalies represents the probability of strong earthquake occurring within the number of days corresponding to different time differences after the occurrence of sample atmospheric tidal gravity anomaly events; determining the cumulative probability distribution based on the conditional probability distribution, and fitting and modeling using the cumulative probability distribution to obtain the prediction model.

[0070] Specifically, statistical analysis of different time differences The frequency of occurrence of each time difference is recorded, and a frequency distribution histogram is plotted. This is used to identify which time intervals are more frequent and whether there is a prominent peak or a specific range indicating a higher frequency of precursor anomalies. Next, the probability of occurrence corresponding to each time difference value is calculated: that is, the percentage of times the time difference equals a certain value (or falls within a certain interval) in the total observation sample. This yields the conditional probability distribution of strong earthquake occurrence relative to precursor anomalies. ( (days) = This abnormal precursor is in The probability of a strong earthquake occurring in the next few days. It is worth noting that this assumes each precursor anomaly will eventually correspond to a strong earthquake, only at different times. It should be noted that in this embodiment of the invention, atmospheric tidal gravity anomalies that fail to match a strong earthquake event within a set time window are defined as failed precursor anomalies. Although these anomalies do not trigger a strong earthquake, they still provide statistical constraint information and can be used as "negative samples" in probability estimation during model training. Specifically, when constructing the statistical model of the anomaly-earthquake relationship, all sample atmospheric tidal gravity anomalies are included in the modeling dataset; if a sample atmospheric tidal gravity anomaly matches a strong earthquake with a magnitude greater than a threshold within the time window, it is marked as a "positive sample," with a corresponding output label of 1; if a sample atmospheric tidal gravity anomaly does not correspond to a strong earthquake within the window, it is marked as a "negative sample," with a corresponding output label of 0. In this way, the model can simultaneously learn the dual statistical laws of "anomaly-strong earthquake" and "anomaly-no earthquake," thereby outputting the conditional probability distribution of strong earthquakes occurring at different time differences.

[0071] After obtaining the discrete conditional probabilities, the cumulative probability distribution is further calculated. This indicates that a strong earthquake occurred after abnormal precursors appeared. The probability of it occurring within a day. For example, if... (1 day) = 0.10 If (2 days) = 0.15, then the probability of accumulating to 2 days is... (2) = 0.25, indicating a 25% probability of a strong earthquake occurring within two days of the occurrence of a precursory anomaly. By summing up each possible time difference, a cumulative probability curve of earthquake occurrence over time can be plotted. This curve is generally expected to be non-decreasing. Follow (Increases monotonically), eventually approaching 1 over a long period (if all precursor anomalies eventually correspond to a strong earthquake). The shape of the curve reveals the temporal characteristics of the post-earthquake event: for example, the steeply rising portion of the curve represents a period of high probability density, while the flat portion represents a period of slow probability increase. Since the curve obtained from sample statistics may not be smooth enough in certain intervals, this invention introduces machine learning methods to fit and model it.

[0072] A predictive model is obtained by fitting and modeling the cumulative distribution probability, including: using a decision tree to fit and model the cumulative distribution probability. The reason for choosing a decision tree is that it is easy to handle nonlinear relationships and the results are interpretable.

[0073] In one embodiment, a prediction model is obtained by fitting and modeling the cumulative distribution probability using a decision tree, including: using different time difference intervals as input features of the prediction model, and iteratively learning the relationship between whether a strong earthquake event occurs after an atmospheric tidal gravity anomaly event and time through a decision tree, thereby obtaining the prediction model.

[0074] Specifically, the approach involves using different time difference intervals as input features for the decision tree model. Through training, the decision tree outputs corresponding cumulative probability values. Time can be divided into several interval nodes, and the decision tree learns the relationship between the time difference and whether a strong earthquake occurs after a precursor anomaly, thus approximating the cumulative probability curve. For example, the root node of the decision tree can correspond to determining whether a strong earthquake will occur after a precursor anomaly. The "occurrence" sub-node further refines the time threshold until the leaf node provides a probability estimate of a strong earthquake occurring within a certain time period. Cross-validation is used during training to avoid overfitting, and the tree depth and branches are optimized based on historical data.

[0075] After training, to verify the model's performance, a portion of the data can be used as a test set to compare the degree of agreement between the model's predicted probability distribution and the true distribution. For example... Figure 2 As shown, Figure 2 This diagram illustrates the comparison between the probability distribution predicted by the prediction model and the actual distribution.

[0076] The prediction model is input with a specific number of days. This will output the predicted probability. Therefore, after obtaining the prediction model, when an atmospheric tidal gravity anomaly event is detected, multiple target days can be input into the pre-built prediction model to obtain the probability of a strong earthquake occurring on each target day after the atmospheric tidal gravity anomaly event occurs.

[0077] It should be noted that multiple target days can be set according to actual needs. For example, multiple target days can be 1, 2, ..., 30 days. Days 1, 2, ..., 30 can be input into the prediction model to obtain the probability of a strong earthquake occurring within the corresponding number of days after an atmospheric tidal gravity anomaly event.

[0078] S103, based on the probability of a strong earthquake occurring within each target day, determine the predicted time of occurrence of the strong earthquake.

[0079] In one embodiment, the probability of a strong earthquake occurring on each target day after the occurrence of an atmospheric tidal gravity anomaly is directly output as the prediction result.

[0080] In another embodiment, the probability of a strong earthquake occurring within multiple time intervals after an atmospheric tidal gravity anomaly event can be determined based on the probability of a strong earthquake occurring within each target day. Based on the probability of a strong earthquake occurring within multiple time intervals, the predicted time of the strong earthquake can be determined. This allows for further division of the days into intervals, calculation of the probability of occurrence within each time interval, and the formation of prediction results for each time interval. The time intervals can be flexibly set according to application requirements, for example, using "daily" as the smallest unit, or merging multiple days into one interval. By accumulating the probability results for each day, a cumulative probability curve can be generated showing the occurrence of an atmospheric tidal gravity anomaly over time, and the cumulative probability of occurrence in different time periods can be determined accordingly, thus outputting a more intuitive and hierarchical time prediction result.

[0081] Alternatively, the cumulative probability of occurrence of atmospheric tidal gravity anomalies on a daily basis and within different intervals can be used as the prediction result for the occurrence time of strong earthquakes.

[0082] Through the above embodiments, the present invention has achieved the following technical effects in predicting the occurrence time of strong earthquakes:

[0083] 1. Quantitative Probability of Short-Term Earthquake Occurrence: Compared to previous approaches that only qualitatively described "there is a possibility of an earthquake in the near future" or a general time window, this invention can output a clear probability value. For example, it can answer questions such as "What is the probability of a strong earthquake occurring within n days after the anomaly occurs?" This quantitative information is more meaningful for decision-making departments and can be used for risk level assessment and emergency response preparation. In practical applications, if the model predicts that an anomaly has a greater than 50% chance of triggering a strong earthquake within the next two days, relevant departments can raise the alert level; conversely, if the probability is very low, excessive tension can be avoided.

[0084] 2. Fully Utilizing Precursor Physical Information: This invention focuses on atmospheric tidal gravity anomalies, a physical precursor anomaly. Compared to models solely based on earthquake statistical laws, it incorporates physical observations of the earthquake gestation process. Atmospheric tidal gravity signals directly reflect the redistribution of subsurface mass and stress changes; therefore, their appearance can be considered one of the indicators of pre-earthquake stress criticality. This method strengthens the reliability of the correlation between precursor anomalies and the final earthquake results by statistically analyzing a large number of such anomalies. This probabilistic prediction based on physical precursor anomalies is expected to improve the reliability of short-term earthquake forecasts.

[0085] 3. Simple and Interpretable Model: Decision tree models have an intuitive and clear structure, and the model's decision-making process can be explained through threshold conditions. For example, the model might state: "If no earthquake occurs within 3 days after the appearance of a precursor anomaly, the probability of an earthquake occurring on the 4th to 7th day increases significantly." Such rules are easy for forecasters to understand and verify. This avoids the problem of "black box" models being difficult to interpret, increasing the credibility of the results. Furthermore, the model is fast to compute, making it suitable for real-time updates of probability assessments during earthquake emergencies.

[0086] 4. Adaptability: As more precursory anomalies—earthquake samples—are accumulated through monitoring, the method of this invention can continuously improve its accuracy through retraining. When new data indicates a change in statistical regularity, the decision tree model will adjust its branch decisions accordingly, thus adapting to the latest situation. This rolling correction mechanism ensures the effectiveness and sophistication of the model.

[0087] When applying the short-term strong earthquake occurrence time prediction method provided by this invention, it is not necessary to rely on... Figure 1 The steps shown are executed in sequence. The specific execution order of each step can be determined as needed, and this invention does not impose any restrictions on it.

[0088] The above describes a method for predicting the occurrence time of short-term strong earthquakes according to one or more embodiments of the present invention. Based on the same idea, the present invention also provides a corresponding device for predicting the occurrence time of short-term strong earthquakes, which includes:

[0089] The detection module is used to acquire the atmospheric tidal gravity signal sequence within a preset time window, and to detect atmospheric tidal gravity anomaly events occurring within the preset time window based on the atmospheric tidal gravity signal sequence.

[0090] The prediction module is used to input multiple target days into a pre-built prediction model after an atmospheric tidal gravity anomaly event is detected, and obtain the probability of a strong earthquake occurring on each target day after the atmospheric tidal gravity anomaly event. The prediction model is trained based on the statistical correlation features between the occurrence time of the atmospheric tidal gravity anomaly event and the occurrence time of the strong earthquake. A strong earthquake is defined as an earthquake with a magnitude greater than 6.

[0091] The determination module is used to determine the predicted time of occurrence of strong earthquakes based on the probability of strong earthquakes occurring within each target day.

[0092] Specific limitations regarding the short-term strong earthquake occurrence time prediction device can be found in the limitations of the short-term strong earthquake occurrence time prediction method described above, and will not be repeated here. Each module in the aforementioned short-term strong earthquake occurrence time prediction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0093] The present invention also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 1 A method for predicting the occurrence time of short-term strong earthquakes is provided.

[0094] The present invention also provides Figure 3 The schematic diagram of the computer device shown is as follows: Figure 3 As shown, at the hardware level, this computer device includes a processor, internal bus, network interface, memory, and non-volatile memory, and may also include other hardware required for business operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then executes it to achieve the above. Figure 1 A method for predicting the occurrence time of short-term strong earthquakes is provided.

[0095] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0096] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this invention.

Claims

1. A method for predicting the occurrence time of short-term strong earthquakes, characterized in that, include: Acquire the atmospheric tidal gravity signal sequence within a preset time window, and detect atmospheric tidal gravity anomaly events occurring within the preset time window based on the atmospheric tidal gravity signal sequence. After an atmospheric tidal gravity anomaly event is detected, multiple target days are input into a pre-built prediction model to obtain the probability of a strong earthquake occurring on each target day after the atmospheric tidal gravity anomaly event. The prediction model is trained based on the statistical correlation characteristics between the occurrence time of the atmospheric tidal gravity anomaly event and the occurrence time of the strong earthquake. A strong earthquake is defined as an earthquake with a magnitude greater than a preset threshold of 6. Based on the probability of a strong earthquake occurring within each target day, the predicted time of occurrence of the strong earthquake is determined. The process of constructing the prediction model includes: acquiring multiple sample atmospheric tidal gravity anomaly events detected within a historical period; for any sample atmospheric tidal gravity anomaly event, searching the earthquake catalog for the most recent strong earthquake occurrence time after the occurrence of the sample atmospheric tidal gravity anomaly event; the earthquake catalog includes the occurrence time of each earthquake; calculating the time difference between the occurrence time of the sample atmospheric tidal gravity anomaly event and the corresponding strong earthquake occurrence time; and constructing a prediction model based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events. Based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events, a prediction model is constructed, including: calculating the frequency of occurrence of different time differences based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events; determining the conditional probability distribution of strong earthquake occurrence relative to precursor anomalies based on the frequency of occurrence of different time differences; the conditional probability distribution of strong earthquake occurrence relative to precursor anomalies represents the probability of strong earthquake occurring within the number of days corresponding to different time differences after the occurrence of sample atmospheric tidal gravity anomaly events; determining the cumulative probability distribution based on the conditional probability distribution, and fitting the model using the cumulative probability distribution to obtain the prediction model.

2. The method according to claim 1, characterized in that, By fitting and modeling the cumulative distribution probability, a prediction model is obtained, including: A prediction model is obtained by fitting and modeling the cumulative distribution probability using a decision tree.

3. The method according to claim 2, characterized in that, By fitting and modeling the cumulative distribution probability using a decision tree, a prediction model is obtained, including: By using different time difference intervals as input features of the prediction model, and through iterative learning of the relationship between time and whether a strong earthquake event will occur after an atmospheric tidal gravity anomaly event, the prediction model is obtained.

4. The method according to claim 1, characterized in that, Based on the probability of a strong earthquake occurring within each target day, the predicted time of occurrence of the strong earthquake is determined, including: Based on the probability of a strong earthquake occurring within each target day, the probability of a strong earthquake occurring in multiple time intervals after an atmospheric tidal gravity anomaly event is determined. The prediction of the occurrence time of a strong earthquake is determined based on the probability of a strong earthquake occurring in multiple time intervals.

5. The method according to claim 1, characterized in that, Based on the atmospheric tidal gravity signal sequence, detect atmospheric tidal gravity anomaly events occurring within a preset time window, including: By removing errors caused by external factors from the atmospheric tidal gravity signal sequence, a stationary residual sequence is obtained. Calculate the median absolute deviation of the stationary residual sequence, and based on the median absolute deviation, calculate the standardized residual corresponding to each time point within the preset time window; If the standardized residuals at all times within a continuously preset first time period are greater than a preset threshold, then it is determined that there is an atmospheric tidal gravity anomaly event in that first time period. If the interval between two atmospheric tidal gravity anomaly events is less than the second time period, then the two atmospheric tidal gravity anomaly events will be merged into a single atmospheric tidal gravity anomaly event.

6. A device for predicting the occurrence time of short-term strong earthquakes, characterized in that, include: The detection module is used to acquire the atmospheric tidal gravity signal sequence within a preset time window, and to detect atmospheric tidal gravity anomaly events occurring within the preset time window based on the atmospheric tidal gravity signal sequence. The prediction module is used to input multiple target days into a pre-built prediction model after an atmospheric tidal gravity anomaly event is detected, and obtain the probability of a strong earthquake occurring on each target day after the atmospheric tidal gravity anomaly event. The prediction model is trained based on the statistical correlation features between the occurrence time of the atmospheric tidal gravity anomaly event and the occurrence time of the strong earthquake. A strong earthquake is defined as an earthquake with a magnitude greater than 6. The determination module is used to determine the predicted time of occurrence of strong earthquakes based on the probability of strong earthquakes occurring within each target day; The process of constructing the prediction model includes: acquiring multiple sample atmospheric tidal gravity anomaly events detected within a historical period; for any sample atmospheric tidal gravity anomaly event, searching the earthquake catalog for the most recent strong earthquake occurrence time after the occurrence of the sample atmospheric tidal gravity anomaly event; the earthquake catalog includes the occurrence time of each earthquake; calculating the time difference between the occurrence time of the sample atmospheric tidal gravity anomaly event and the corresponding strong earthquake occurrence time; and constructing a prediction model based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events. Based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events, a prediction model is constructed, including: calculating the frequency of occurrence of different time differences based on the time differences corresponding to all sample atmospheric tidal gravity anomaly events; determining the conditional probability distribution of strong earthquake occurrence relative to precursor anomalies based on the frequency of occurrence of different time differences; the conditional probability distribution of strong earthquake occurrence relative to precursor anomalies represents the probability of strong earthquake occurring within the number of days corresponding to different time differences after the occurrence of sample atmospheric tidal gravity anomaly events; determining the cumulative probability distribution based on the conditional probability distribution, and fitting the model using the cumulative probability distribution to obtain the prediction model.

7. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method as described in any one of claims 1 to 5.

8. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1 to 5.

Images