A lightning signal reconstruction method and system considering satellite observation geometry

Abstract

The application discloses a lightning signal reconstruction method and system considering satellite observation geometry, wherein the method comprises the following steps: calculating the median and quantile range through original lightning event data, performing scaling processing on the space-time characteristics, and obtaining scaled characteristic values of lightning events; based on the scaled characteristic values of lightning events, the geocentric angle is calculated by using a spherical geometry formula, and the satellite nadir angle of each lightning event is determined by combining the earth radius and the satellite height; the pixel size effect factor and the atmospheric path length factor are calculated by using the satellite nadir angle; the space scale factor is obtained through nonlinear transformation; based on the space scale factor, the space distance between lightning events is measured by using the spherical distance, and the integrated space-time distance is generated by combining the time distance obtained by using the Huber loss function; and the reconstructed lightning is obtained by adaptive clustering based on the integrated space-time distance. Therefore, the accuracy of lightning signal reconstruction is improved.

Description

Technical Field

[0001] This invention relates to the field of meteorological satellite remote sensing technology, and in particular to a method and system for reconstructing lightning signals that takes into account satellite observation geometry. Background Technology

[0002] Lightning, as a powerful discharge phenomenon over extremely long distances in the atmosphere, not only frequently accompanies disastrous weather events such as rainstorms, hail, and strong winds, but also plays an irreplaceable role in maintaining the electrical balance between the atmospheric ionosphere and the Earth, as well as in the generation of nitrogen oxides in the global nitrogen cycle.

[0003] Therefore, achieving accurate monitoring of lightning signals is of great scientific value and practical significance for real-time early warning of severe convective weather, disaster risk prevention and control, and in-depth exploration of the long-term impact of lightning on atmospheric chemistry and climate systems.

[0004] Geostationary satellites, with their optical imaging capabilities, can continuously observe lightning activity at high temporal resolution, covering a wide area and unaffected by ground-based observation conditions, thus effectively capturing the complete spatiotemporal evolution of lightning events. Lightning signals detected by satellites typically appear as discrete points of light in time-series images where the light radiation energy is significantly higher than the background noise—that is, lightning events. By analyzing the correlation patterns of these events in the spatial and temporal dimensions, the complete lightning process can be further reconstructed.

[0005] However, geostationary satellites observing from high orbits experience significant variations in spatial resolution due to the Earth's curvature, resulting in variations across different geographical locations. Near the satellite's nadir, the near-vertical observation angle leads to high spatial resolution and small pixel sizes. Multiple adjacent lightning events, due to their spatial density, are easily misclassified by clustering algorithms as belonging to a single lightning event, causing a systematic underestimation of lightning density and an abnormally large spatiotemporal extent. Conversely, in the peripheral regions far from the nadir, the tilted observation angle increases the pixel projection size on the Earth's surface, decreasing spatial resolution. Simultaneously, the increased atmospheric transmission path length causes light signal attenuation and scattering effects, making it easy for a single lightning event to be incorrectly fragmented into multiple lightning events, leading to an overestimation of lightning density and a compressed spatiotemporal extent.

[0006] Existing technologies generally employ a fixed spatial clustering radius processing strategy, which cannot adapt to dynamic spatial scale differences caused by geometric variations in satellite observations. This results in a persistent and systematic deviation between the reconstructed lightning distribution and the actual physical process, severely limiting the accuracy of monitoring data. This deviation not only affects the quantitative analysis of lightning activity but also weakens its reliability in weather forecasting and climate research.

[0007] Therefore, there is an urgent need for a technical solution that can dynamically adjust the spatial scale determination mechanism based on satellite observation geometry in order to eliminate reconstruction errors caused by resolution changes. Summary of the Invention

[0008] This invention provides a lightning signal reconstruction method and system that takes into account satellite observation geometry, which improves the accuracy of lightning signal reconstruction.

[0009] To achieve the above objectives, in a first aspect, this invention provides a lightning signal reconstruction method considering satellite observation geometry, comprising: calculating the median and quantile ranges from raw lightning event data, scaling the spatiotemporal features to obtain scaled feature values ​​of the lightning events; calculating the geocentric angle using spherical geometry formulas based on the scaled feature values ​​of the lightning events, and determining the satellite nadir angle for each lightning event by combining the Earth's radius and satellite altitude; calculating the pixel size effect factor and atmospheric path length factor using the satellite nadir angle, and obtaining a spatial scale scaling factor through nonlinear transformation; using the spatial scale scaling factor, measuring the spatial distance between lightning events using spherical distance, and combining it with the temporal distance obtained using the Huber loss function to generate a comprehensive spatiotemporal distance; introducing balanced weights and performing normalization based on the comprehensive spatiotemporal distance to calculate a balanced spatiotemporal distance for clustering, and constructing an objective function based on the balanced spatiotemporal distance; using principal component analysis to split the clusters along the principal direction, merging neighboring clusters, updating the cluster centers, until the objective function converges to obtain the reconstructed lightning signal.

[0010] Secondly, this invention provides a lightning signal reconstruction system considering satellite observation geometry, comprising: a feature value acquisition module, a satellite nadir angle determination module, a scaling factor acquisition module, a spatiotemporal distance generation module, an objective function construction module, and a lightning signal acquisition module. The feature value acquisition module calculates the median and quantile range from the original lightning event data, scales the spatiotemporal features, and obtains scaled feature values ​​for the lightning events. The satellite nadir angle determination module calculates the geocentric angle using spherical geometry formulas based on the scaled feature values ​​of the lightning events, and determines the satellite nadir angle for each lightning event by combining the Earth's radius and satellite altitude. The scaling factor acquisition module calculates the pixel size effect factor and atmospheric path length factor using the satellite nadir angle, and obtains a spatial scale scaling factor through nonlinear transformation. The spatiotemporal distance generation module uses the spatial scale scaling factor, measures the spatial distance between lightning events using spherical distance, and combines this with the temporal distance obtained using the Huber loss function to generate a comprehensive spatiotemporal distance. The objective function construction module is used to calculate the balanced spatiotemporal distance for clustering by introducing balanced weights and normalizing the calculated spatiotemporal distance, and then constructing the objective function based on the balanced spatiotemporal distance. The lightning signal acquisition module is used to split the clusters along the principal direction using principal component analysis, merge neighboring clusters, update the cluster centers, and continue until the objective function converges to obtain the reconstructed lightning signal.

[0011] Thirdly, the present invention provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a lightning signal reconstruction method considering satellite observation geometry as described above.

[0012] Fourthly, the present invention provides a computer-readable storage medium including a computer program and instructions, which, when the computer program or the instructions are executed on a computer, cause the computer to perform a lightning signal reconstruction method taking into account satellite observation geometry as described above.

[0013] Compared with existing technologies, the lightning signal reconstruction method and system of the present invention, which considers satellite observation geometry, dynamically adjusts the spatial clustering radius according to the satellite observation angle, measures the spatiotemporal distance of lightning events in combination with the physical process of lightning occurrence, and determines the optimal number of lightning events through an adaptive mechanism, thereby eliminating reconstruction errors and improving the accuracy of lightning signal reconstruction. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating a lightning signal reconstruction method considering satellite observation geometry according to Embodiment 1 of the present invention.

[0015] Figure 2 This is a schematic diagram of a lightning signal reconstruction system considering satellite observation geometry, as shown in Embodiment 2 of the present invention.

[0016] Figure 3 This is a schematic diagram of the structure of an electronic device according to Embodiment 3 of the present invention;

[0017] Figure 4 This is a schematic diagram of the logic flow of a lightning signal reconstruction method considering satellite observation geometry in a specific embodiment of the present invention;

[0018] Figure 5 This is a schematic diagram comparing the original LMI lightning with LIS in a specific embodiment of the present invention;

[0019] Figure 6 This is a schematic diagram illustrating the reconstruction of lightning and LIS using the method described in a specific embodiment of the present invention;

[0020] Figure 7 This is a schematic diagram comparing the original LMI lightning with LIS in a specific embodiment of the present invention;

[0021] Figure 8 This is a schematic diagram illustrating the reconstruction of lightning and LIS using the method described in a specific embodiment of the present invention. Detailed Implementation

[0022] The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and not intended to limit the scope of the invention. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the embodiments of the present invention, and not all structures.

[0023] To facilitate understanding, the main implementation concepts of the various embodiments of the present invention will be briefly described first.

[0024] In traditional geostationary satellite lightning detection technology, the spatial resolution of different regions decreases as the satellite's nadir angle increases due to the Earth's curvature. Near the satellite's nadir point, densely distributed lightning events with high spatial resolution are easily misreconstructed by clustering algorithms as belonging to a single lightning event, leading to an underestimation of lightning density and an overestimation of the spatiotemporal range. Conversely, in the peripheral regions far from the nadir point, spatial resolution decreases and signal attenuation occurs due to atmospheric path lengthening, making it easy for a single lightning event to be incorrectly split into multiple lightning events, resulting in an overestimation of lightning density and a compressed spatiotemporal range. Current fixed spatial clustering radius processing methods cannot dynamically adapt to these spatial scale changes caused by the observation angle, leading to a systematic deviation between the reconstructed lightning distribution and the actual situation, thus affecting the accuracy of lightning activity monitoring.

[0025] For example, when monitoring transoceanic severe convective weather systems, satellite observations of ocean areas (near the nadir) show high-density lightning activity. However, due to the high spatial resolution, multiple adjacent lightning events are clustered into a small number of lightning events, thus underestimating the actual lightning frequency. Simultaneously, in observations of high-latitude inland areas (far from the nadir), due to reduced spatial resolution and atmospheric attenuation, a single lightning event is fragmented into multiple lightning events, leading to an overestimation of lightning density. This bias causes severe convective weather warning models based on reconstructed data to misjudge the spatiotemporal evolution of lightning activity, thereby affecting the reliability of weather forecasts.

[0026] If the above problems are not addressed, the systematic bias in lightning detection data will reduce the meteorological warning system's ability to identify severe convective weather, potentially leading to false or missed warnings. At the same time, in atmospheric environmental research, erroneous lightning distribution data will interfere with the analysis of the global nitrogen cycle and ionospheric balance, reducing the credibility of related scientific research.

[0027] Furthermore, to address the resolution variation caused by satellite observation geometry, the spatial scale determination method needs to be dynamically adjusted to eliminate reconstruction errors. Specifically, by calculating the satellite nadir angle and introducing a spatial scale scaling factor, it is possible to adapt to observation conditions in different regions. The generation of the spatial scale scaling factor requires combining a pixel size effect factor and an atmospheric path length factor, achieved through nonlinear transformation. Therefore, the fusion of spatial and temporal distances calculated based on this scaling factor can construct a more accurate clustering basis. As a preferred implementation method, principal component analysis is used to split clusters along the principal direction and merge neighboring clusters to optimize the cluster structure. For example, introducing balanced weights and normalization during the clustering process can reduce the impact of the number of events on distance calculation. Finally, by iteratively updating the cluster centers until the objective function converges, accurate reconstruction of lightning signals is achieved.

[0028] Example 1, Figure 1 This is a flowchart illustrating a lightning signal reconstruction method considering satellite observation geometry according to Embodiment 1 of the present invention, as shown below. Figure 1 As shown, Embodiment 1 provides a lightning signal reconstruction method considering satellite observation geometry, including:

[0029] Y100 calculates the median and quantile ranges from the raw lightning event data, scales the spatiotemporal features, and obtains scaled feature values ​​of the lightning events.

[0030] In one implementation, step Y100 may further include:

[0031] Y110, acquire raw lightning event data, and construct a multi-dimensional spatiotemporal feature matrix based on the raw lightning event data;

[0032] Y120, calculates the median and dispersion benchmark of the statistical distribution for the multidimensional spatiotemporal feature matrix;

[0033] Y130, using the median and the dispersion benchmark, performs robust scaling on the multidimensional spatiotemporal feature matrix to generate a robust scaling vector;

[0034] Y140, determine whether the element values ​​in the robust scaling vector exceed a preset threshold;

[0035] Y150, if the element value exceeds a preset threshold, then the abnormal events that meet this condition will be removed, and the scaled feature value will be extracted as the scaled feature value of the lightning event.

[0036] Specifically, this invention lays the foundation for subsequent data processing by acquiring raw lightning event data and constructing a multidimensional spatiotemporal feature matrix. Given the potential for outliers in the raw data, to ensure robustness of feature scaling, this scheme further calculates the median and dispersion benchmark of the statistical distribution of the multidimensional spatiotemporal feature matrix. These statistics are insensitive to outliers and can more accurately reflect the central tendency and dispersion of the data. Based on this, the multidimensional spatiotemporal feature matrix is ​​robustly scaled using the median and dispersion benchmark to generate a robust scaling vector. This scaling method effectively reduces the impact of extreme values ​​on feature representation, allowing data from different feature dimensions to be compared on a uniform scale. Subsequently, by determining whether the element values ​​in the robust scaling vector exceed a preset threshold, lightning events that still significantly deviate from the normal pattern after robust processing can be accurately identified. Finally, outliers are removed, and the scaled feature values ​​are extracted as the scaled feature values ​​for the lightning events. This series of steps ensures that a more accurate and robust feature representation of lightning events can be obtained in the initial stage of lightning signal reconstruction. This provides high-quality input data for subsequent lightning signal reconstruction based on satellite observation geometry, thereby avoiding the negative impact of anomalous lightning in the original data on the overall reconstruction process. This step performs robust scaling on all lightning events in the original data, and the generated feature values ​​are used for all subsequent calculations. The threshold judgment is only used for anomalous event detection, but does not affect the integrity of the scaled feature values.

[0037] As a specific implementation method, the present invention is implemented as follows: First, raw lightning event data is acquired from a satellite lightning imager (such as an LMI sensor). This data may include information such as the time of occurrence (UTC), longitude, latitude, and radiation energy of each lightning event. This information is organized into a multi-dimensional spatiotemporal feature matrix. For example, each row represents a lightning event, and the columns are time, longitude, latitude, and radiation energy, respectively. Next, for each feature dimension of the matrix, the median and interquartile range (IQR) are calculated as a benchmark for the degree of dispersion. For example, for the time dimension, the median and IQR of the occurrence times of all lightning events are calculated. Then, robust scaling is performed on the data of each feature dimension. Specifically, the scaled feature value = (original feature value - median) / IQR. After scaling, a robust scaling vector is obtained. A preset threshold is set. For example, if the absolute value of the scaled feature value of a lightning event in any dimension exceeds 3, the event is considered an anomalous event. All anomalous events that meet this condition are removed. Finally, the scaled feature values ​​are extracted and used as scaled feature values ​​of the lightning event for subsequent steps such as satellite nadir angle calculation.

[0038] Through the above technical solution, this invention effectively addresses the noise and outlier issues present in the original lightning event data during the initial stage of lightning signal reconstruction. By introducing median and dispersion benchmarks for robust scaling, the interference of extreme data points on feature representation can be significantly reduced, ensuring the accuracy and stability of the scaling results. This robust processing makes the scaled feature values ​​of the lightning event more realistically reflect the essential characteristics of lightning activity, avoiding reconstruction bias caused by data quality issues. Therefore, this scheme provides a more reliable and accurate input for subsequent lightning signal reconstruction based on satellite observation geometry, thereby improving the overall accuracy and robustness of lightning signal reconstruction.

[0039] In one specific embodiment, the present invention first acquires raw lightning event data, constructs a multi-dimensional spatiotemporal feature matrix, calculates the median and quantile range for each feature, performs robust scaling, and extracts the scaled feature values ​​of the lightning events. In practical applications, taking a severe convective event in the Guangdong and Guangxi regions on June 1, 2018, as an example, the lightning event observed using the FY-4A LMI geostationary satellite lightning imager is used. The event feature values ​​processed in this step are used for subsequent satellite nadir angle calculation and cluster analysis. The feature dimensions include longitude, latitude, and time. The median and quantile ranges are calculated based on the actual data, with a preset threshold of 3. Events with an absolute value exceeding 3 after scaling are considered anomalous events. After removing anomalous events, the scaled feature values ​​of the lightning events are extracted for subsequent steps.

[0040] Y200, based on the scaled eigenvalues ​​of lightning events, uses spherical geometry formulas to calculate the geocentric angle and combines the Earth's radius and satellite altitude to determine the satellite nadir angle for each lightning event.

[0041] In one implementation, step Y200 may further include:

[0042] Y210, obtain the latitude and longitude coordinates of the location of the lightning event and the latitude and longitude coordinates of the satellite nadir point contained in the scaled feature value of the lightning event;

[0043] Y220, calculate the geocentric angle value based on the latitude and longitude coordinates of the location where the lightning event occurred and the latitude and longitude coordinates of the satellite's nadir point;

[0044] Y230, a spatial triangle is constructed by combining the aforementioned geocentric angle value, the Earth's average radius value, and the orbital altitude value of the satellite sensor at the observation time;

[0045] Y240 calculates the angle between the satellite observation line-of-sight vector and the geocentric perpendicular vector based on the geometric relationship of the spatial triangle, and determines the angle as the satellite nadir angle of the lightning event.

[0046] Specifically, this invention obtains the precise location of a lightning event and the coordinates of the satellite's nadir point, calculates the geocentric angle using spherical geometry, and then constructs a spatial geometric model by combining the Earth's radius and the satellite's orbital altitude, ultimately calculating the nadir angle when the satellite observes the lightning event. This step provides crucial geometric parameters for subsequent dynamic adjustment of the spatial clustering radius.

[0047] The geocentric angle refers to the angle between the point of a lightning event and the nadir point of a satellite, as observed from the Earth's center. It reflects the central angle corresponding to the great circle distance between the two points on the Earth's surface. The nadir angle is the angle between the satellite's line of sight and the perpendicular to the Earth's center. This angle directly determines the pixel size effect and atmospheric path length of satellite observations.

[0048] Specifically, firstly, the latitude and longitude coordinates of the lightning event are extracted from the scaled feature values ​​of the lightning event, and the coordinates of the satellite's nadir point at the corresponding observation time are obtained (Y210). Then, the geocentric angle between the two points is calculated based on their latitude and longitude using the great circle distance formula in spherical geometry (Y220). Next, a spatial triangle is constructed using the Earth's center, the lightning event point, and the satellite sensor as three vertices, utilizing the known geocentric angle, the Earth's average radius, and the satellite's orbital altitude (Y230). Finally, within this triangle, the angle between the satellite's line of sight and the perpendicular to the Earth's center is calculated using the sine or cosine theorem, thus obtaining the satellite's nadir angle (Y240).

[0049] As a specific implementation method, the geocentric angle is calculated. The inverse cosine formula can be used: ,in The longitude of the satellite's nadir point. and These are the latitude and longitude of the lightning event, respectively. Furthermore, the satellite nadir angle... It can be calculated using the arctangent formula: ,in The average radius of the Earth This represents the satellite's orbital altitude. Using this formula, the satellite's nadir angle corresponding to each lightning event can be accurately calculated.

[0050] The above technical solution enables precise calculation of the geometric angles of each lightning event under satellite observation, providing accurate input for subsequent calculations of the pixel size effect factor and atmospheric path length factor. This precise calculation based on spherical geometry avoids errors caused by planar approximation, ensures the accuracy of spatial scale scaling factor calculation, and thus improves the accuracy of lightning signal reconstruction.

[0051] Y300 uses the satellite nadir angle to calculate the pixel size effect factor and atmospheric path length factor, and obtains the spatial scale scaling factor through nonlinear transformation.

[0052] In one implementation, step Y300 may further include:

[0053] Y310, obtain the satellite nadir angle, and combine the sensor parameters to calculate the geometric distortion ratio and atmospheric transmission distance ratio, which are respectively determined as the pixel size effect factor and atmospheric path length factor;

[0054] Y320 constructs a power-law response function for the pixel size effect factor and an exponential decay function for the atmospheric path length factor;

[0055] Y330, the output value of the power-law response function and the output value of the exponential decay function are coupled and calculated to generate a composite correction coefficient;

[0056] Y340 performs an inverse transformation on the composite correction coefficients to obtain the spatial scale scaling factor.

[0057] Specifically, this invention calculates the geometric distortion ratio by combining satellite nadir angle with sensor optical parameters to reflect changes in pixel projected area, thus determining it as the pixel size effect factor; simultaneously, it calculates the atmospheric transmission distance ratio to reflect the degree of signal attenuation, thus determining it as the atmospheric path length factor. A power-law response function is constructed for the pixel size effect factor to simulate its nonlinear impact on spatial resolution; an exponential decay function is constructed for the atmospheric path length factor to simulate the absorption and scattering effects of the atmosphere on the signal. The output values ​​of the two functions are coupled to generate a composite correction coefficient, which is then processed through an inverse transform to obtain the spatial scale scaling factor. This factor dynamically compensates for changes in spatial resolution and signal strength caused by observation geometry.

[0058] As a specific implementation method, the solution of the present invention is implemented as follows: obtaining the satellite nadir angle of a lightning event. By combining sensor parameters such as the field of view of a lightning imager (e.g., LMI) and the detector pixel size, the ratio of the ground pixel area to the nadir pixel area is calculated using a geometric projection model. This ratio is used as the geometric distortion ratio, i.e., the pixel size effect factor. Using an atmospheric radiative transfer model or a simplified geometric model, the ratio of the signal transmission path length to the vertical path length is calculated and used as the atmospheric transmission distance ratio, i.e., the atmospheric path length factor. .against Construct a power-law response function, for example ,in These are constants determined based on the sensor characteristics; for Construct an exponential decay function, for example ,in This is the atmospheric attenuation coefficient. and Perform multiplication coupling operations to generate composite correction coefficients. Subsequently, regarding Perform inverse transformation processing, such as taking the reciprocal or applying a predefined inverse function, to obtain the final spatial scale scaling factor. This process can be summarized by the following formula: ,in and As a sensitivity factor, and for Obtain by performing a nonlinear transformation This is to eliminate the impact of outliers.

[0059] The above technical solution enables the dynamic calculation of the spatial scale scaling factor based on satellite observation geometry, accurately quantifies the combined effects of pixel size effect and atmospheric attenuation effect, provides a reliable basis for subsequent scale correction of spatial distance, and thus improves the adaptability of lightning signal reconstruction to the differences in observation conditions in different regions.

[0060] In one specific embodiment, the present invention is based on satellite nadir angle. The spatial scale scaling factor is calculated using the method described above. For example, in satellite observation geometric calculations, formulas are used. ,in and Determined by sensor parameters and atmospheric models, respectively. and Based on experience, and for Perform nonlinear transformations (such as logarithmic transformations or piecewise function processing) to obtain This is used for subsequent spatiotemporal distance calculations.

[0061] Y400, based on the aforementioned spatial scale scaling factor, uses spherical distance to measure the spatial distance between lightning events, and combines this with the temporal distance obtained by processing using the Huber loss function to generate a comprehensive spatiotemporal distance.

[0062] Specifically, this invention applies a spatial scale scaling factor to spatial distance calculation and combines it with temporal distance processing that considers the intermittent and continuous characteristics of lightning discharges to generate a comprehensive distance that accurately reflects the spatiotemporal correlation of lightning events. The scheme first calculates the spherical arc length based on the latitude and longitude coordinates of the lightning event and corrects it using a spatial scale scaling factor to eliminate the influence of geometric distortions in satellite observations. Simultaneously, based on the intermittent and continuous characteristics of lightning discharges, the time difference is normalized using the Huber loss function. Finally, the two are fused using a preset spatiotemporal scale factor to form a comprehensive spatiotemporal distance, providing an accurate metric for subsequent clustering.

[0063] In one implementation, step Y400 may further include:

[0064] Y410: Obtain the latitude and longitude coordinates of the first lightning event and the second lightning event, as well as the corresponding spatial scale scaling factor. Calculate the spherical arc length based on the latitude and longitude coordinates, and multiply the spherical arc length by the spatial scale scaling factor to obtain the scale-corrected spatial distance.

[0065] Among them, the spherical arc length refers to the shortest distance between two points on the Earth's surface along a great circle path. Its calculation needs to take into account the Earth's curvature and is usually solved using spherical trigonometric formulas. The spatial scale scaling factor is a correction coefficient calculated based on the satellite nadir angle and is used to compensate for spatial resolution differences and signal attenuation effects caused by changes in observation geometry.

[0066] Specifically, for any two lightning events, obtain their latitude and longitude coordinates ( , )and( , The arc length of the sphere is calculated using the spherical cosine theorem or the Haversine formula.

[0067] As a specific implementation method, the following formula can be used for calculation:

[0068] ,in, , This is the average radius of the Earth.

[0069] Next, the calculated spherical arc length Multiplying the two events by the corresponding spatial scale scaling factors (e.g., averaging or choosing based on event location) yields the scale-corrected spatial distance: ,in Spatial scale factor, This is the spatial scale scaling factor.

[0070] Y420: Obtain the timestamps of the first lightning event and the second lightning event, calculate the absolute value of their time difference, and use a preset threshold parameter to perform a robust transformation on the absolute value of the time difference through the Huber loss function to obtain a normalized time distance.

[0071] The timestamp refers to the precise time value recording the moment a lightning event occurs, usually expressed in UTC time. The Huber loss function is a robust loss function that uses squared loss when the error is less than a threshold and linear loss when the error is greater than the threshold, thereby reducing the impact of outliers.

[0072] Specifically, obtain the timestamps of the two lightning events. and Calculate the absolute value of the time difference Then, using a preset threshold parameter τ, the Huber loss function is applied to... The transformation is performed to obtain the normalized time distance. As one specific implementation method, The calculation formula is as follows:

[0073] ;

[0074] in, This is a time scale factor used to adjust the dimensions and importance of time distances. This process smooths out distance variations within a normal time difference range, while imposing a linear penalty on abnormally large time differences to prevent them from having an excessive impact on clustering.

[0075] Y430, based on a preset spatiotemporal scale factor, performs scale fusion on the scale-corrected spatial distance and the normalized temporal distance to generate the comprehensive spatiotemporal distance.

[0076] The preset spatiotemporal scale factor refers to the relative importance of spatial and temporal dimensions in the overall distance, set according to actual application needs. It can be adjusted based on lightning type, geographical region, season, and weather system to ensure that time and spatial distance are not completely coupled, while maintaining comparable development rates and reflecting the actual proximity of lightning events. Scale fusion refers to combining two distances with different dimensions into a single comprehensive distance metric through weighting.

[0077] Specifically, based on the preset spatial scale factor and time scale factor (satisfy Spatial distance after scale correction and normalized time distance Perform scale fusion to generate comprehensive spatiotemporal distance .

[0078] As a specific implementation method, the combined distance can also be directly generated using the following formula:

[0079] ;

[0080] in, Spatial scale factor is already included. and scaling factor The correction, and Includes time scale factor The adjustment is equivalent to embedding both time and spatial scales within the distance, ensuring the comparability of spatiotemporal distances.

[0081] Based on the above analysis, this invention achieves accurate calculation of the spatiotemporal distance between lightning events. By introducing a spatial scale scaling factor, it effectively compensates for the spatial resolution variations caused by the Earth's curvature in high-orbit observations, ensuring consistent comparability of the spatial distances of lightning events in different regions. Simultaneously, considering the intermittent and continuous characteristics of lightning discharges, the Huber loss function is used to handle temporal distances, enhancing the algorithm's robustness to abnormal time intervals and avoiding interference from random noise in the clustering results. The resulting comprehensive spatiotemporal distance more accurately reflects the actual correlation between lightning events, providing a reliable foundation for subsequent clustering analysis and thus improving the accuracy of lightning signal reconstruction.

[0082] Y500, based on the comprehensive spatiotemporal distance, introduces a balanced weight and performs normalization processing to calculate the balanced spatiotemporal distance for clustering, and constructs an objective function based on the balanced spatiotemporal distance.

[0083] In this context, balancing weights refer to the weights assigned to different clusters during the clustering process. The aim is to adjust the influence of each cluster on the overall objective function, preventing the clustering results from being biased towards larger clusters due to differences in cluster size, thus achieving a more balanced clustering effect. Normalization involves adjusting the weights to a specific range (e.g., between 0 and 1) to eliminate the influence of dimensions and ensure the comparability of weights in subsequent calculations. Balanced spatiotemporal distance is the distance metric after weight adjustment; it reflects the true similarity between lightning events after considering differences in cluster size. The objective function is the function that needs to be minimized during the optimization of the clustering algorithm; its value reflects the quality of the current clustering result.

[0084] In one implementation, step Y500 may further include:

[0085] Y510, based on the comprehensive spatiotemporal distance, calculates the balance weight according to the number of lightning events contained in each cluster.

[0086] The balancing weights are calculated based on the cluster size (i.e., the number of lightning events contained in a cluster), and are usually inversely proportional to the cluster size to prevent large clusters from dominating the objective function.

[0087] Specifically, the balance weights can be calculated using the inverse of the cluster size, the inverse of the square root, or an exponential function, for example, ,in It is a balancing factor. It is the number of samples in cluster j. It is the maximum number of clustered samples.

[0088] Y520, the balance weights are normalized, and the normalized balance weights are used to weight the integrated spatiotemporal distance to obtain the balanced spatiotemporal distance.

[0089] Normalization refers to adjusting the calculated balance weights to a specific range (e.g., between 0 and 1) to eliminate the influence of dimensions and ensure that the weights are comparable in subsequent calculations.

[0090] Specifically, min-max normalization can be used to linearly scale the weights to the [0,1] interval, or Z-score standardization can be used to convert the weights into a distribution with a mean of 0 and a standard deviation of 1. Weighting refers to applying the normalized balanced weights to the overall spatiotemporal distance to adjust their relative importance in the clustering process, thereby obtaining a balanced spatiotemporal distance. Weighting can be achieved by multiplying the overall spatiotemporal distance by the corresponding normalized balanced weights; for example, in lightning events... To the cluster center Equilibrium spacetime distance ,in It is the normalized balance weight.

[0091] Y530 uses the sum of squares of the equilibrium spatiotemporal distances of all lightning events to their respective cluster centers as the objective function.

[0092] The objective function is the function that needs to be minimized during the optimization process of the clustering algorithm, and its value reflects the quality of the current clustering result.

[0093] Specifically, for N lightning events and K cluster centers, the objective function can be expressed as: ,in It is an indicator matrix representing lightning events. Does it belong to clustering? , It was a lightning event. To the cluster center The goal is to find a balance between spatial and temporal distances. By minimizing this objective function, the clustering algorithm can find a cluster partition that minimizes the intra-cluster distances and maximizes the inter-cluster distances.

[0094] Specifically, the present invention optimizes the distance metric and objective function in the clustering process by introducing a balancing weight mechanism. In the initial stage of clustering iteration or in each iteration, the system calculates the balancing weight based on the number of lightning events contained in each cluster. For example, the balancing weight can be calculated using an exponential function: ,in It is a balancing factor. It is clustering The number of samples, This represents the maximum number of clustered samples. Subsequently, these balancing weights are normalized, for example, by dividing each weight by the sum of all weights, so that the sum of the normalized weights is 1. The balanced spatiotemporal distance is obtained by weighting the comprehensive spatiotemporal distance from each lightning event to its respective cluster center using normalized balanced weights: Finally, the objective function is the sum of the squares of the equilibrium spatiotemporal distances of all lightning events to their respective cluster centers: By minimizing this objective function, the clustering algorithm is guided to form cluster structures that are not only internally compact but also more balanced in scale.

[0095] Through the above technical solution, this invention effectively solves the problem of clustering result bias caused by the uneven distribution of lightning events in each cluster during the clustering process of lightning signal reconstruction. By introducing balanced weights and weighting the comprehensive spatiotemporal distance, the clustering algorithm can treat clusters of different sizes fairly when optimizing the objective function. This avoids the excessive influence of large clusters on the cluster center location, ensuring that even clusters containing fewer lightning events can be accurately identified and characterized. Therefore, this scheme can significantly improve the accuracy and stability of lightning signal reconstruction, especially in complex observation scenarios where lightning events are unevenly distributed, and can obtain lightning signal reconstruction results that are more consistent with the actual situation.

[0096] Y600 uses principal component analysis to split clusters along the principal direction, merge neighboring clusters, and update cluster centers until the objective function converges, thus obtaining the reconstructed lightning signal.

[0097] In one implementation, step Y600 may further include:

[0098] Y610 constructs a covariance matrix for the initial cluster set to extract the principal direction, and cuts the clusters in the initial cluster set into sub-clusters based on the projection variance of the sample points on the principal direction;

[0099] Y620, calculate the spatiotemporal distance between the sub-cluster centers, merge cluster pairs whose spatiotemporal distance is less than a preset proximity threshold, and update the cluster centers until the objective function converges, and lock the final cluster structure;

[0100] Y630, based on the final clustering structure, identifies lightning events belonging to the same cluster as a reconstructed lightning event.

[0101] Specifically, this invention extracts the main clustering direction through principal component analysis and cuts the clusters based on projection variance, achieving refined splitting of the initial clusters; it calculates the spatiotemporal distance between sub-cluster centers and compares it with a preset threshold to achieve corrective fusion of over-splitting; and iteratively updates the cluster centers and monitors the convergence of the objective function to ensure the stability of the cluster structure. This iterative optimization strategy of splitting first and then fusing enables the clustering process to adaptively adjust the cluster boundaries, thereby accurately capturing the true spatiotemporal distribution patterns of lightning events.

[0102] As a specific implementation method, the present invention is implemented as follows: For each initial cluster, the covariance matrix of all lightning events within it in spatiotemporal coordinates (such as longitude, latitude, and time) is calculated, and the principal direction with the largest variance is obtained through eigenvalue decomposition. All events within the cluster are projected onto this principal direction to form a one-dimensional distribution; if the distribution exhibits multiple peaks or the projection variance exceeds a preset threshold (e.g., a multiple of the average cluster distance), the cluster is cut into two or more sub-clusters along the valley. Subsequently, the equilibrium spatiotemporal distance between any two sub-cluster centers is calculated (this distance is calculated in step Y500). If this distance is less than a preset proximity threshold (e.g., a composite threshold considering a spatial distance of 5 kilometers and a time difference of 100 milliseconds), the two sub-clusters are merged into a new cluster. After each split or fusion, the cluster centers are recalculated, and the objective function value (i.e., the sum of squares of the equilibrium spatiotemporal distances from all events to their respective cluster centers) is calculated. The above splitting and fusion process is repeated until the rate of change of the objective function is less than the convergence threshold, at which point the cluster structure is stable, and the iteration stops.

[0103] Through the above technical solution, this invention can effectively handle complex situations such as uneven lightning distribution and irregular spatiotemporal extension. Initial clustering, through splitting along the main direction, avoids mistakenly merging multiple independent lightning bolts into one due to excessively large clusters; the fusion of neighboring sub-clusters prevents individual lightning bolts from being over-segmented due to noise or observational geometric effects. This dynamic adjustment mechanism significantly improves the accuracy and robustness of lightning signal reconstruction, making the reconstruction results more closely reflect the spatiotemporal continuity of actual lightning activity.

[0104] In one specific embodiment, the present invention uses the following mathematical formula for iterative optimization: For N lightning events and K cluster centers, the objective function (inertia) is defined as: ;in For the indicator matrix, To balance the spatiotemporal distance, the balancing weights are calculated as follows: ,in As a balance factor, For clustering The number of samples, The maximum number of clustered samples. After normalization, the balanced spatiotemporal distance is obtained: Through iterative optimization, principal component analysis is used to split clusters along the principal direction or merge neighboring clusters, updating cluster centers until convergence. The convergence condition is: ,in This is the convergence threshold.

[0105] In practical applications, such as Figure 4 As shown in a specific embodiment of the present invention, the main processing steps of a lightning signal reconstruction method considering satellite observation geometry are as follows:

[0106] 1. Handling Abnormal Lightning Events:

[0107] For the original lightning event, while preserving the original feature discrimination, the impact of outliers and extreme values ​​on clustering is reduced, and its spatiotemporal features are scaled to a specific range.

[0108] Features Let the original feature value be x. (j) =[x1 (j) x2 (j) ,...,x n (j) ] T Then the scaled feature value is: ;

[0109] in, The median, For the quantile range, To standardize the eigenvalues, For the first The lightning event was in the 1st The original observations in each feature dimension.

[0110] 2. Satellite observation geometric calculations:

[0111] Calculate the geocentric angle using spherical geometry formulas:

[0112] ;

[0113] in, Let λ be the geocentric angle, λ0 be the longitude of the nadir point, and φ and λ be the latitude and longitude of the observation point, respectively.

[0114] Furthermore, the satellite nadir angle for each lightning event is obtained:

[0115] ;

[0116] in, For the satellite nadir angle, For the Earth's radius, This is the satellite's altitude.

[0117] The pixel size effect is calculated based on the satellite nadir angle, while also considering the atmospheric path length effect, to obtain the spatial scale scaling factor:

[0118] ;

[0119] in, The initial scaling factor. and These are the pixel size effect factor and the atmospheric path length factor, respectively. and To correspond to the sensitivity factors of the two; to eliminate the influence of outliers, for Perform a nonlinear transformation to obtain , This is the spatial scale scaling factor that is ultimately applied to spatiotemporal distance calculations.

[0120] 3. Calculation of the spatiotemporal distance of a lightning event:

[0121] For two points on the sphere and The spherical distance is:

[0122] ;

[0123] in, The distance is spherical. For the Earth's radius, For intermediate calculation variables, ;

[0124] in, The latitude value of the first lightning event. This is the dimension value of the second lightning event (or cluster center). The longitude value of the first lightning event. This is the longitude value of the second lightning event (or cluster center).

[0125] time and The distance is Based on the intermittent and continuous characteristics of lightning discharge, the Huber loss function is used to handle time distance:

[0126] ;

[0127] in, The time distance after robust processing. As a time scale factor, This is the threshold parameter.

[0128] Regarding lightning events Considering spatial scaling, calculate To the cluster center Spatial-temporal distance:

[0129] ;

[0130] in, For the first A lightning event To the Cluster centers The comprehensive spatiotemporal distance The time distance after robust processing. The spherical distance Spatial scale factor, In order to target the A lightning event The final spatial scale scaling factor calculated.

[0131] The temporal and spatial scale factors can be adjusted according to lightning type, geographical region, season and weather system, so that time and spatial distance are not completely coupled, while having comparable development rates, and can reflect the actual proximity of lightning events.

[0132] 4. Iterative optimization:

[0133] for A lightning event, Cluster centers, allocation matrix The objective function (inertia) is defined as follows:

[0134] ;

[0135] in, Let be the objective function. The total number of all lightning events. This represents the total number of clusters to be identified or divided using a clustering algorithm. It is a binary indicator variable. For the first A lightning event To the Cluster centers The comprehensive spatiotemporal distance.

[0136] Introducing balancing weights:

[0137] ;

[0138] in, For the first The balanced weights of each cluster, It is a balancing factor. It is clustering The number of samples, It is the maximum number of clustered samples.

[0139] The spacetime distance after equilibrium is:

[0140] ;

[0141] in, This is the spatiotemporal distance after adjusting for balancing weights. For the first A lightning event To the Cluster centers The comprehensive spatiotemporal distance For normalized weights, then:

[0142] ;

[0143] in, The objective function that needs to be minimized is... The total number of lightning events participating in the clustering. The total number of clusters the algorithm aims to create. It is a binary indicator variable. This is the spatiotemporal distance after adjusting for balancing weights.

[0144] Finding the allocation matrix and cluster center To minimize the objective function:

[0145] ;

[0146] in, As cluster center, For the indicator matrix, The objective function that needs to be minimized.

[0147] Through iterative optimization, principal component analysis is used to split clusters along the principal direction or merge neighboring clusters, updating cluster centers until convergence. The convergence condition is:

[0148] ;

[0149] in, For the first The objective function value calculated after the next iteration. For the first The objective function value calculated after the current iteration. This is the convergence threshold.

[0150] In practical applications, a severe convective event in the Guangdong and Guangxi regions on June 1, 2018, is used as an example. Lightning events observed using the FY-4A LMI geostationary satellite lightning imager are reconstructed using the method of this invention, and compared with synchronous observations from the ISS LIS low-orbit satellite lightning imager. Figure 5 and Figure 6 As shown. Figure 5 The raw LMI lightning was compared with that of LIS. During this process, 270 lightning bolts were observed in LIS and 100 lightning bolts were observed in LMI. Figure 6 The lightning reconstructed using this method was compared with LIS observations; the number of lightning bolts reconstructed using this method was 249. The lightning observed using this method shows higher consistency with LIS observations.

[0151] Figures 7 to 8 The data presented is from a severe convective event that occurred in northern my country on June 11, 2018. During this event, LIS observed 205 lightning strikes, LMI recorded 22, and the reconstructed lightning strike count using this method was 25. Ignoring the impact of LMI detection limitations (which resulted in a lower number of lightning events), this method shows higher agreement with LIS observations (as indicated by the orange arrow).

[0152] Example 2, Figure 2 This is a schematic diagram of a lightning signal reconstruction system considering satellite observation geometry, as shown in Embodiment 2 of the present invention. Figure 2As shown in Embodiment 2, a lightning signal reconstruction system considering satellite observation geometry is provided, including: a feature value acquisition module, a satellite nadir angle determination module, a scaling factor acquisition module, a spatiotemporal distance generation module, an objective function construction module, and a lightning signal acquisition module. The feature value acquisition module calculates the median and quantile range from the original lightning event data, scales the spatiotemporal features, and obtains scaled feature values ​​for the lightning events. The satellite nadir angle determination module calculates the geocentric angle using spherical geometry formulas based on the scaled feature values ​​of the lightning events, and determines the satellite nadir angle for each lightning event by combining the Earth's radius and satellite altitude. The scaling factor acquisition module calculates the pixel size effect factor and atmospheric path length factor using the satellite nadir angle, and obtains the spatial scale scaling factor through nonlinear transformation. The spatiotemporal distance generation module uses the spatial scale scaling factor, measures the spatial distance between lightning events using spherical distance, and combines it with the temporal distance obtained using the Huber loss function to generate a comprehensive spatiotemporal distance. The objective function construction module is used to calculate the balanced spatiotemporal distance for clustering by introducing balanced weights and normalizing the calculated spatiotemporal distance, and then constructing the objective function based on the balanced spatiotemporal distance. The lightning signal acquisition module is used to split the clusters along the principal direction using principal component analysis, merge neighboring clusters, update the cluster centers, and continue until the objective function converges to obtain the reconstructed lightning signal.

[0153] In this embodiment, the feature value acquisition module includes: a first construction unit, a calculation and statistics unit, a first generation unit, a judgment unit, and an extraction unit. The first construction unit acquires raw lightning event data and constructs a multidimensional spatiotemporal feature matrix based on the raw lightning event data. The calculation and statistics unit calculates the median and dispersion benchmark of the statistical distribution for the multidimensional spatiotemporal feature matrix. The first generation unit performs robust scaling on the multidimensional spatiotemporal feature matrix using the median and the dispersion benchmark to generate a robust scaling vector. The judgment unit determines whether the element values ​​in the robust scaling vector exceed a preset threshold. The extraction unit removes abnormal events that meet this condition if the element values ​​exceed the preset threshold, and extracts the scaled feature values ​​as the scaled feature values ​​of the lightning events.

[0154] In this embodiment, the satellite nadir angle determination module includes: an acquisition unit, a first calculation unit, a second construction unit, and a first determination unit. The acquisition unit acquires the latitude and longitude coordinates of the lightning event's location and the satellite's nadir point, which are included in the scaled feature values ​​of the lightning event. The first calculation unit calculates the geocentric angle value based on the latitude and longitude coordinates of the lightning event's location and the satellite's nadir point. The second construction unit constructs a spatial triangle by combining the geocentric angle value, the Earth's average radius value, and the satellite sensor's orbital altitude value at the observation time. The first determination unit calculates the angle between the satellite's line-of-sight vector and the geocentric perpendicular vector based on the geometric relationship of the spatial triangle, and determines the angle as the satellite nadir angle of the lightning event.

[0155] In this embodiment, the scaling factor obtaining module includes a second determining unit, a third constructing unit, a second generating unit, and a first obtaining unit. The second determining unit is used to acquire the satellite nadir angle, and calculate the geometric distortion ratio and atmospheric transmission distance ratio based on sensor parameters, respectively determining them as a pixel size effect factor and an atmospheric path length factor. The third constructing unit is used to construct a power-law response function for the pixel size effect factor and an exponential decay function for the atmospheric path length factor. The second generating unit is used to couple the output value of the power-law response function with the output value of the exponential decay function to generate a composite correction coefficient. The first obtaining unit is used to perform an inverse transform on the composite correction coefficient to obtain the spatial scale scaling factor.

[0156] In this embodiment, the spatiotemporal distance generation module includes a second obtaining unit, a third obtaining unit, and a third generating unit. The second obtaining unit is used to acquire the latitude and longitude coordinates of the first lightning event and the second lightning event, and the corresponding spatial scale scaling factor. Based on the latitude and longitude coordinates, it calculates the spherical arc length and multiplies the spherical arc length by the spatial scale scaling factor to obtain the scale-corrected spatial distance. The third obtaining unit is used to acquire the timestamps of the first lightning event and the second lightning event, calculate the absolute value of their time difference, and use a preset threshold parameter to perform a robust transformation on the absolute value of the time difference through the Huber loss function to obtain a normalized spatiotemporal distance. The third generating unit is used to perform scale fusion of the scale-corrected spatial distance and the normalized spatiotemporal distance according to a preset spatiotemporal scale factor to generate the comprehensive spatiotemporal distance.

[0157] In this embodiment, the objective function construction module includes a second calculation unit, a fourth obtaining unit, and a balancing unit. The second calculation unit calculates a balancing weight based on the comprehensive spatiotemporal distance and the number of lightning events in each cluster. The fourth obtaining unit normalizes the balancing weights and uses the normalized balancing weights to weight the comprehensive spatiotemporal distance to obtain the balanced spatiotemporal distance. The balancing unit uses the sum of the squares of the balanced spatiotemporal distances from all lightning events to their respective cluster centers as the objective function.

[0158] In this embodiment, the lightning signal acquisition module includes a clustering segmentation unit, a locking unit, and an identification unit. The clustering segmentation unit constructs a covariance matrix for an initial cluster set to extract the principal direction, and segments the clusters in the initial cluster set into sub-clusters based on the projection variance of sample points along the principal direction. The locking unit calculates the spatiotemporal distance between the centers of the sub-clusters, merges clusters with a spatiotemporal distance less than a preset proximity threshold, and updates the cluster centers until the objective function converges, locking the final cluster structure. The identification unit identifies lightning events belonging to the same cluster as a reconstructed lightning event based on the final cluster structure.

[0159] The various variations and specific examples of the lightning signal reconstruction method considering satellite observation geometry provided in Embodiment 1 are also applicable to the lightning signal reconstruction system considering satellite observation geometry provided in this embodiment. Through the foregoing detailed description of the lightning signal reconstruction method considering satellite observation geometry, those skilled in the art can clearly understand the implementation method of the lightning signal reconstruction system considering satellite observation geometry in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0160] Example 3, Figure 3 This is a schematic diagram of the structure of an electronic device according to Embodiment 3 of the present invention, as shown below. Figure 3 As shown, Embodiment 3 also provides an electronic device 300, which may include a processor 301 and a memory 302.

[0161] Memory 302 is used to store programs. Memory 302 may include volatile memory, such as random-access memory (RAM), such as static random-access memory (SRAM), double data rate synchronous dynamic random-access memory (DDR SDRAM), etc.; memory may also include non-volatile memory, such as flash memory. Memory 302 is used to store computer programs (such as application programs, functional modules, etc. that implement the above methods), computer instructions, etc. The computer programs, computer instructions, etc., can be partitioned and stored in one or more memories 302. Furthermore, the computer programs, computer instructions, data, etc., can be accessed by processor 301.

[0162] The aforementioned computer programs and instructions can be stored in one or more partitions of memory 302. Furthermore, the aforementioned computer programs and instructions can be invoked by processor 301.

[0163] The processor 301 is configured to execute the computer program stored in the memory 302 to implement the various steps of the methods described in the above embodiments. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0164] The processor 301 and the memory 302 can be independent structures or integrated structures. When the processor 301 and the memory 302 are independent structures, the memory 302 and the processor 301 can be coupled together via bus 303.

[0165] The electronic device in this embodiment can execute the technical solution in the above method. Its specific implementation process and technical principle are the same, and will not be repeated here.

[0166] Example 4, also provides a computer-readable storage medium including a computer program and instructions, which, when run on a computer, cause the computer to perform a lightning signal reconstruction method considering satellite observation geometry according to any embodiment of the present invention.

[0167] Computer-readable storage media include various media that can store program code, such as USB flash drives, external hard drives, ROM, RAM, magnetic disks, or optical disks.

[0168] This embodiment also provides a computer program product, which includes: a computer program stored in a readable storage medium, at least one processor of an electronic device can read the computer program from the readable storage medium, and the at least one processor executes the computer program to cause the electronic device to perform the solution provided in any of the above embodiments.

[0169] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this invention disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0170] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. A method for reconstructing lightning signals considering satellite observation geometry, characterized in that, include: By calculating the median and quantile ranges from the raw lightning event data, the spatiotemporal features are scaled to obtain scaled feature values ​​of the lightning events. Based on the scaled characteristic values ​​of lightning events, the geocentric angle is calculated using spherical geometry formulas, and combined with the Earth's radius and satellite altitude, the satellite nadir angle for each lightning event is determined. Using the satellite nadir angle, the pixel size effect factor and atmospheric path length factor are calculated, and the spatial scale scaling factor is obtained through nonlinear transformation. Based on the aforementioned spatial scale scaling factor, a spherical distance is used to measure the spatial distance between lightning events, and combined with the temporal distance obtained by processing using the Huber loss function, a comprehensive spatiotemporal distance is generated. Based on the comprehensive spatiotemporal distance, a balanced weight is introduced and normalized to calculate the balanced spatiotemporal distance for clustering, and an objective function is constructed based on the balanced spatiotemporal distance. Principal component analysis is used to split the clusters along the principal direction, merge neighboring clusters, and update the cluster centers until the objective function converges, thus obtaining the reconstructed lightning signal.

2. The lightning signal reconstruction method considering satellite observation geometry as described in claim 1, characterized in that, The process of calculating the median and quantile ranges from the original lightning event data and scaling the spatiotemporal features to obtain scaled feature values ​​of the lightning events includes: Obtain raw lightning event data, and construct a multidimensional spatiotemporal feature matrix based on the raw lightning event data; Calculate the median and dispersion benchmark of the statistical distribution for the multidimensional spatiotemporal feature matrix; The multidimensional spatiotemporal feature matrix is ​​robustly scaled using the median and the dispersion benchmark to generate a robust scaling vector. Determine whether the element values ​​in the robust scaling vector exceed a preset threshold; If the value of the element exceeds a preset threshold, the abnormal events that meet this condition will be removed, and the scaled feature value will be extracted as the scaled feature value of the lightning event.

3. The lightning signal reconstruction method considering satellite observation geometry as described in claim 1, characterized in that, The geocentric angle is calculated using spherical geometry formulas based on the scaled feature values ​​of the lightning event. Combined with the Earth's radius and satellite altitude, the satellite nadir angle for each lightning event is determined as follows: Obtain the latitude and longitude coordinates of the location of the lightning event and the latitude and longitude coordinates of the satellite nadir point, which are included in the scaled feature values ​​of the lightning event; The geocentric angle value is calculated based on the latitude and longitude coordinates of the location where the lightning event occurred and the latitude and longitude coordinates of the satellite's nadir point; A spatial triangle is constructed by combining the aforementioned geocentric angle value, the Earth's average radius value, and the orbital altitude value of the satellite sensor at the observation time; The angle between the satellite observation line-of-sight vector and the geocentric perpendicular vector is calculated based on the geometric relationship of the spatial triangle, and the angle is determined as the satellite nadir angle of the lightning event.

4. The lightning signal reconstruction method considering satellite observation geometry as described in claim 1, characterized in that, The process involves using the satellite nadir angle to calculate the pixel size effect factor and atmospheric path length factor, and then obtaining the spatial scale scaling factor through nonlinear transformation. The satellite nadir angle is obtained, and the geometric distortion ratio and atmospheric transmission distance ratio are calculated by combining sensor parameters. These are then determined as the pixel size effect factor and atmospheric path length factor, respectively. A power-law response function is constructed for the pixel size effect factor, and an exponential decay function is constructed for the atmospheric path length factor; The output value of the power-law response function is coupled with the output value of the exponential decay function to generate a composite correction coefficient. Perform an inverse transformation on the composite correction coefficients to obtain the spatial scale scaling factor.

5. The lightning signal reconstruction method considering satellite observation geometry as described in claim 1, characterized in that, The method, based on a spatial scale scaling factor, uses spherical distance to measure the spatial distance between lightning events and combines this with the temporal distance obtained using the Huber loss function to generate a comprehensive spatiotemporal distance, including: Obtain the latitude and longitude coordinates of the first lightning event and the second lightning event and the corresponding spatial scale scaling factor, calculate the spherical arc length based on the latitude and longitude coordinates, and multiply the spherical arc length by the spatial scale scaling factor to obtain the scale-corrected spatial distance; The timestamps of the first lightning event and the second lightning event are obtained, the absolute value of their time difference is calculated, and the absolute value of the time difference is robustly transformed by the Huber loss function using a preset threshold parameter to obtain the normalized time distance. Based on a preset spatiotemporal scale factor, the scale-corrected spatial distance and the normalized temporal distance are scale-fused to generate the comprehensive spatiotemporal distance.

6. The lightning signal reconstruction method considering satellite observation geometry as described in claim 1, characterized in that, The method involves introducing balanced weights and normalizing the comprehensive spatiotemporal distance to calculate a balanced spatiotemporal distance for clustering. The objective function constructed based on this balanced spatiotemporal distance includes: Based on the comprehensive spatiotemporal distance, the balance weight is calculated according to the number of lightning events contained in each cluster; The balance weights are normalized, and the integrated spatiotemporal distance is weighted using the normalized balance weights to obtain the balanced spatiotemporal distance. The objective function is the sum of the squares of the equilibrium spatiotemporal distances of all lightning events to their respective cluster centers.

7. The lightning signal reconstruction method considering satellite observation geometry as described in claim 1, characterized in that, The process involves using principal component analysis to split and cluster along the principal direction, merging neighboring clusters, updating cluster centers, and continuing until the objective function converges. The resulting reconstructed lightning signal includes: A covariance matrix is ​​constructed for the initial cluster set to extract the principal direction, and the clusters in the initial cluster set are cut into sub-clusters based on the projection variance of the sample points on the principal direction; Calculate the spatiotemporal distance between the sub-cluster centers, merge cluster pairs whose spatiotemporal distance is less than a preset proximity threshold, and update the cluster centers until the objective function converges, thus locking the final cluster structure. Based on the final clustering structure, lightning events belonging to the same cluster are identified as a reconstructed lightning event.

8. A lightning signal reconstruction system considering satellite observation geometry, characterized in that, include: The feature value acquisition module is used to calculate the median and quantile range from the raw lightning event data, scale the spatiotemporal features, and obtain the scaled feature values ​​of the lightning event. The satellite nadir angle determination module is used to calculate the geocentric angle based on the scaled feature value of the lightning event using spherical geometry formulas, and combine the Earth's radius and satellite altitude to determine the satellite nadir angle for each lightning event; The scaling factor acquisition module is used to calculate the pixel size effect factor and atmospheric path length factor using the satellite nadir angle, and obtain the spatial scale scaling factor through nonlinear transformation; The spatiotemporal distance generation module is used to generate a comprehensive spatiotemporal distance by using a spherical distance metric based on the spatial scale scaling factor and combining it with the temporal distance obtained by processing using the Huber loss function. The objective function construction module is used to introduce balanced weights and perform normalization processing based on the comprehensive spatiotemporal distance, calculate the balanced spatiotemporal distance for clustering, and construct the objective function based on the balanced spatiotemporal distance. as well as The lightning signal acquisition module is used to split the clusters along the principal direction using principal component analysis, merge neighboring clusters, update the cluster centers, and so on until the objective function converges to obtain the reconstructed lightning signal.

9. An electronic device, characterized in that, include: At least one processor; as well as A memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform a lightning signal reconstruction method considering satellite observation geometry according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It includes computer programs and instructions that, when run on a computer, cause the computer to perform a lightning signal reconstruction method that takes into account satellite observation geometry as described in any one of claims 1-7.

Images