Radar observation scheduling method based on real-time lightning activity sensing
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING METEOROLOGICAL OBSERVATION CENT
- Filing Date
- 2026-02-28
- Publication Date
- 2026-05-26
Smart Images
Figure CN122085281A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of meteorological technology, and in particular to a radar observation and scheduling method based on real-time sensing of lightning activity. Background Technology
[0002] Thunderstorms and severe convective weather are among the major hazardous weather events affecting social production, military activities, and public safety. They are often accompanied by strong winds, hail, and short-duration heavy rainfall, causing enormous economic losses and casualties. Therefore, achieving real-time and accurate monitoring and early warning of thunderstorms and severe convective weather is a core requirement for meteorological disaster prevention and mitigation efforts.
[0003] Currently, the monitoring and early warning of thunderstorms and severe convective weather mainly rely on two technical approaches: lightning location technology based on lightning electromagnetic pulse radiation and weather radar technology based on active remote sensing. However, there is no precedent for applying lightning observation data to radar scheduling and feeding it back into lightning observation. Studies have shown that lightning activity is significantly correlated with specific radar echo characteristics. Lightning is mainly concentrated in strong echo areas with radar reflectivity factors greater than 30 dBZ, and most lightning radiation sources are distributed in the altitude range of 6 to 11 kilometers. High-frequency lightning activity often corresponds to strong updrafts and a large number of collisions and growths of ice phase particles (ice crystals, graupel), which usually manifests as strong high-altitude echoes on radar. This physical correlation is the cornerstone of lightning activity pattern analysis and radar collaborative observation and scheduling strategies.
[0004] To enhance monitoring and early warning capabilities, it is necessary to develop "target observation and collaborative observation technologies," establish a collaborative observation system for severe convective weather, and conduct demonstrations of collaborative observation using weather radar as the primary instrument, combined with various other observation equipment. However, existing collaborative observation technologies are mostly based on numerical weather prediction potential data or networked radar observation data for target guidance. Potential forecasting itself has unavoidable forecast biases; and collaborative observation based on networked radar requires waiting for the radar to complete a full volumetric scan cycle (usually 3-6 minutes) and finish data fusion processing before the center of the severe weather target can be identified. By this time, the rapidly developing convective cell center may have moved more than several kilometers, resulting in delayed observation and dispatch instructions and an inability to capture the most intense and critical evolution period of the storm.
[0005] Although existing technologies have made significant progress in data fusion, forecasting and early warning, and assimilation applications, they all share a fundamental bottleneck: they all belong to the "post-observation analysis" or "trend extrapolation" mode, failing to achieve forward, proactive, and precise scheduling of radar observation resources using lightning, the fastest and most direct strong convective electrical activity signal, as a real-time trigger source; at the same time, they also cannot achieve refined observation of the impact of lightning activity on radar echoes. Summary of the Invention
[0006] The purpose of this invention is to design a radar observation and scheduling method based on real-time lightning activity perception in order to solve the above problems.
[0007] The present invention achieves the above objectives through the following technical solutions:
[0008] Radar observation scheduling methods based on real-time lightning activity sensing include:
[0009] S1. Real-time acquisition of lightning data and radar volume scan base data;
[0010] S2. Perform quality control and spatiotemporal grid alignment on lightning data and radar volume scan base data to obtain aligned lightning data and radar mosaic data;
[0011] S3. Use the aligned lightning data to identify lightning clusters and extract key features of lightning clusters;
[0012] S4. Generate the effective radar echo area of the lightning cluster based on the key features of the lightning cluster, and obtain the radar echo characteristics of the lightning cluster.
[0013] S5. Construct a target correlation matrix and jointly match the key features of each lightning cluster Ci with the radar echo features;
[0014] S6. Identify potential targets based on the target association matrix to obtain high-potential targets;
[0015] S7. Extract the enhanced feature vector of high-potential targets;
[0016] S8. Select observation targets based on the enhanced feature vectors of high-potential targets;
[0017] S9. Based on the observation target, schedule the corresponding X-band radar and generate observation scheduling instructions;
[0018] S10. The observation and dispatch instruction is sent to the radar control terminal corresponding to the dispatched X-band radar site.
[0019] S11. The radar performs RHI scanning according to the observation and dispatch instructions.
[0020] The beneficial effects of this invention are as follows: This method transforms the lightning locator from a traditional "early warning information source" and "post-event correlation data" role into a "control signal source" that drives the radar observation system to make real-time, proactive, and intelligent behavior adjustments. It opens up the shortest technical path from "lightning signal perception" to "radar action execution", and constructs a new collaborative observation paradigm with faster response, more accurate guidance, and more intelligent scheduling, thereby fundamentally improving the short-term near-term monitoring and early warning capabilities for lightning and severe convective weather. Attached Figure Description
[0021] Figure 1 This is a schematic diagram of the radar observation and scheduling method based on real-time lightning activity sensing of the present invention. Detailed Implementation
[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0023] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0024] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0025] In the description of this invention, it should be understood that the terms "upper," "lower," "inner," "outer," "left," "right," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used to facilitate the description of this invention and to simplify the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.
[0026] Furthermore, the terms "first," "second," etc., are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.
[0027] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, terms such as "set" and "connection" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.
[0028] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
[0029] like Figure 1 As shown, the radar observation scheduling method based on real-time lightning activity sensing includes:
[0030] S1. Real-time acquisition of lightning data and radar volume scan base data; specifically: continuously receiving and parsing lightning data (lightning time, location, intensity, type) and radar volume scan base data conforming to the "Weather Radar Base Data Format Standard" from the lightning locator network. For each received data point, preliminary format checks and error corrections are performed, including verifying the integrity of the data frame, the validity of the timestamp, and whether the geographic coordinates are within the valid range. A data reception status log is recorded to ensure the integrity and availability of subsequent processing. Simultaneously, the data decryption module is activated to parse the acquired categorized data and convert it into internal data objects according to the standard.
[0031] S2. Perform quality control and spatiotemporal grid alignment on lightning data and radar volume scan base data to obtain aligned lightning data and radar mosaic data;
[0032] The multi-level progressive quality control of lightning data specifically involves: first, removing data points with unreasonable metadata such as signal strength and positioning error; then, removing noise data caused by equipment interference or ground object reflection through outlier detection (e.g., IQR method) and spatial consistency checks (e.g., local density filtering) to obtain processed lightning data; and the quality control of radar volume scan base data specifically involves: applying ground object clutter suppression, velocity ambiguity removal, and calibration checks to obtain quality-controlled radar data.
[0033] Spatiotemporal grid alignment specifically involves defining a unified 3km x 3km horizontal grid and a spatiotemporal cube with a 1-minute time window. The processed lightning data and quality-controlled radar data are mapped onto a standardized Cartesian grid and a standard height layer using inverse distance weighting and nearest neighbor interpolation. This ensures that the multi-source data are strictly aligned in the three dimensions of time, space, and quantity, and records the maximum combined reflectivity of the radar mosaic data in the grid and its corresponding height.
[0034] S3. Utilize the aligned lightning data to identify lightning clusters and extract key features of the lightning clusters; specifically including:
[0035] S301. Using a 1-minute sliding window, maintain a time series with a sliding length of 15 minutes for each 3km×3km grid cell. Define a detection window and a background window. Perform initial screening based on the jump condition. Grids that meet the jump condition are designated as jump grids, while those that do not meet the jump condition are filtered out. The jump condition is the jump rate. And the total frequency of lightning within the detection window ; ,in, An exponentially weighted moving average of the frequency of lightning strikes within the background window;
[0036] S302. Extract all three-dimensional lightning points within a preset range centered on the leap grid, and cluster the three-dimensional lightning points using the density-based DBSCAN spatial clustering algorithm to obtain at least one lightning cluster Ci; specifically:
[0037] The distance function of the DBSCAN spatial clustering algorithm is defined as follows: the horizontal spatial distance uses the Haversine semi-positive distance formula, the vertical spatial distance uses the height difference, and a time constraint T = 60 seconds is set. Clustering is performed only on pairs of points whose time difference is within the time constraint T. Within the spatiotemporal constraints, the normalized spatial distance calculation formula is: ,in, , For horizontal and vertical distances, , Empirical scale parameters of 3 km and 2 km were selected respectively. (Setting...) (Right now (adjacent) Clustering is performed. Clusters with a spatial overlap greater than 50% generated by adjacent time windows (such as a 30-second sliding window) are automatically merged and assigned a unique ID to form a lightning cluster Ci. The update log and status of the lightning cluster Ci are continuously tracked and updated in the log in a timely manner.
[0038] S303. Extract the key features of each lightning cluster. The key features include the three-dimensional center (Center_i), three-dimensional range, dynamic features, and vertical structure features. The three-dimensional center (Center_i) is the weighted average of the longitude, latitude, and altitude of the lightning point (based on signal strength). The three-dimensional range includes the equivalent ellipse (major axis, minor axis, and direction) of the horizontal projection and the vertical thickness (Vertical_Span_i). The dynamic features include the current 1-minute lightning frequency (Freq_i) and the rate of change (ΔFreq_i) relative to the previous window. The vertical structure features include the lightning's highest height (Height_Top_i), the lightning's average height (Height_Center_i), and the lightning density vertical gradient (Vertical_Gradient_i) (calculated through linear fitting).
[0039] S4. Generate the effective radar echo area of the lightning cluster based on its key characteristics, and obtain the radar echo characteristics of the lightning cluster; specifically including:
[0040] S401. Obtain the reflectivity CR of the three-dimensional center of the lightning cluster Ci at the current moment or the most recent radar mosaic combination.
[0041] S402, using radar combined reflectivity Based on the contour area, a 10km distance buffer is drawn outward to form the effective radar echo zone;
[0042] S403. Determine whether the three-dimensional center Center_i of lightning cluster Ci is located within the effective radar echo area through spatial topology. If so, proceed to S404; otherwise, eliminate the lightning cluster and do not perform subsequent calculations or radar scheduling.
[0043] S404. Query the maximum reflectivity of each height layer within the horizontal range of the lightning cluster Ci, and use it as the radar echo characteristic of the lightning cluster.
[0044] S5. Construct a target correlation matrix and jointly match the key features of each lightning cluster Ci with the radar echo features.
[0045] S6. Identify potential targets based on the target correlation matrix to obtain high-potential targets. Specifically, if a lightning cluster Ci simultaneously satisfies the first and second conditions, but not the third condition, then the lightning cluster Ci is considered a high-potential target. The first condition is: the three-dimensional center Center_i of the lightning cluster Ci is located at the radar combined reflectivity... The first condition is that the strong echo core is located in the vicinity of the lightning cluster Ci; the second condition is that the lightning cluster Ci is in a rapid development stage; the third condition is that the maximum height of the lightning Height_Top_i is less than 5km for the lightning cluster Ci.
[0046] S7. Extracting enhanced feature vectors of high-potential targets This includes the logarithmic frequency increment log(ΔFreq_i+1), the maximum height of lightning Height_Top_i, the vertical thickness Vertical_Span_i, and the maximum radar combined reflectivity CR_max of the grid.
[0047] S8. Select observation targets based on the enhanced feature vectors of high-potential targets; specifically including:
[0048] S801. Analyze the priority score of each high-potential target. , is represented as: ,in, , , and The weights for frequency variation, maximum lightning height, vertical thickness, and reflectivity are respectively. , , These correspond to the enhanced feature vector values log(ΔFreq_i+1), Height_Top_i, Vertical_Span_i, and CR_max of the high-potential target, respectively. The weight configuration file is dynamically loaded according to the weather type: for example, under the "severe thunderstorm" weather system, the weights w1 and w2 are increased to 0.4 and 0.3, respectively; under the "short-term heavy precipitation" weather system, the weights w3 and w4 are even higher.
[0049] S802, Scoring based on priority Sort the targets in descending order, select the top 3 high-potential targets as observation targets, and record the sorting results and weight usage in the decision log.
[0050] S9. Based on the observation target, schedule the corresponding X-band radar and generate observation scheduling instructions; specifically including:
[0051] S901: Acquire X-band radars within a 75-kilometer radius that are currently in "scheduling" mode;
[0052] S902. Select all X-band radar stations whose relative azimuth to the observed target is within a preset azimuth as pre-selected radar stations.
[0053] S903. Based on the highest height of the observed target (Height_bottom_i) and the location of the pre-selected radar station, calculate the highest detection elevation angle of the pre-selected radar station, and use the pre-selected radar station whose highest detection elevation angle is within the preset detection elevation angle as the scheduled X-band radar station.
[0054] S904. Generate observation scheduling instructions for the X-band radar sites to be scheduled. The observation scheduling instructions include the radar preset azimuth, preset scanning angle and radar control terminal IP. When the azimuth deviation is within ±5°, the observation scheduling instruction is a continuous scheduling command; when the azimuth deviation exceeds ±5°, the observation scheduling instruction is a single scheduling command.
[0055] S10. The observation and scheduling instructions are sent to the radar control terminal corresponding to the X-band radar site being scheduled via a low-latency communication link.
[0056] S11. The radar executes an RHI scan according to the observation and dispatch instructions; specifically, after receiving the instructions, the radar control software immediately cuts off the existing scan and controls the antenna to turn to the designated azimuth, starting the RHI scan according to the calculated elevation angle value. At the same time, the radar returns a radar status report to the control center, preparing for the next round of radar dynamic scheduling and adjustment.
[0057] This method dynamically and adaptively concentrates limited rapid and customized scanning resources on the most likely to cause disasters and the most important convective targets, thereby effectively improving the resource utilization efficiency and overall monitoring effectiveness of the entire observation network during severe convective weather events.
[0058] This method constructs a closed-loop control paradigm of "perception-decision-execution" based on real-time event triggering. It uses lightning location data streams as real-time scheduling commands to drive the dynamic allocation of radar observation resources, thus solving the core operational pain point in severe convective weather monitoring where "radar scanning speed cannot keep up with the speed of weather evolution." Specifically, it manifests as follows:
[0059] Event triggering: The system continuously monitors a high-frequency lightning event stream that updates every minute.
[0060] Intelligent decision-making: Through the scheduling algorithm described below, "high-priority three-dimensional lightning clusters" with high development potential are identified in real time.
[0061] Command generation and execution: The three-dimensional dynamic characteristics of the lightning cluster (such as spatial location, vertical structure, and evolution trend) are parameterized in real time into a set of non-periodic control commands that can be directly parsed by the radar's underlying control system. This command instructs the radar (such as an X-band fast-scan radar) to immediately interrupt the ongoing conventional volumetric scan (VCP) and switch to performing a customized RHI (range-height indicator) fine scan of the target cluster's core region.
[0062] Feedback optimization: The acquired high-resolution observation data is fed back to the system to optimize subsequent decision-making models.
[0063] This method utilizes real-time knowledge of lightning to determine and "command" the radar "when, where, and how" to focus its observations, thus achieving a leap from static analysis to dynamic control.
[0064] This method employs an adaptive decision-making process comprised of "coarse screening + fine agglomeration + intelligent sorting," specifically:
[0065] Lightning frequency jump detection based on a fixed spatiotemporal grid (e.g., 3km × 3km, 1min) is introduced as a "smart gate" for computational resources. By comparing the frequency of the detection window with that of the background window (calculating the jump rate R), detection is only performed when conditions are met (e.g.,...). Fine-grained analysis is initiated on the "jump grid" and its surrounding area. This method enables adaptive allocation of computing resources, filters out most inactive areas, and is a key design feature to ensure the real-time performance of the system.
[0066] Accurate perception is achieved through three-dimensional spatiotemporal density clustering: For the lightning points in the coarse screening area, the three-dimensional DBSCAN algorithm, which integrates horizontal distance, vertical height difference and strict time constraints, is used for clustering.
[0067] The custom distance function is: normalized spatial distance within a time window T. ,when The nodes are determined to be adjacent. This is fundamentally different from traditional two-dimensional planar clustering. The generated "lightning clusters" are three-dimensional physical entities with a clear vertical structure and spatiotemporal continuity.
[0068] Contextualized dynamic prioritization enables intelligent decision-making: Multi-dimensional dynamic features such as vertical thickness, lightning peak height, and frequency variation rate are extracted from 3D lightning clusters. (Ranking function) The weights can be dynamically loaded with preset configurations based on the diagnosed weather type (such as severe thunderstorms or short-duration heavy rainfall). This allows the ranking criteria to align with the most pressing disaster early warning needs, achieving a shift from "finding the strongest target" to "finding the target that needs the most attention".
[0069] This method's adaptive decision-making evolves the isolated, static judgments of traditional schemes into a continuous, dynamic, and closely related intelligent perception and decision-making process, which can more keenly capture the critical period of "initial surge" of convection, and has significant early warning significance.
[0070] This method implements a closed-loop feedback control system for a "dynamic resource scheduler".
[0071] Core roles are defined as follows: the lightning positioning network is the trigger, the data processing center is the intelligent controller, and the radar is the controlled actuator.
[0072] Closed-loop resource scheduling: It not only completes a single closed loop of "event triggering → command execution", but its core controller also has the ability to prioritize and dynamically schedule multiple targets and resources. When multiple high-priority targets are generated at the same time, or when multiple radars are available, the scheduler will comprehensively consider factors such as target priority, real-time working status of radars, and geographical geometric relationships to dynamically allocate tasks and ensure optimal global observation benefits.
[0073] Task-oriented interface: The interface between the system and the radar is not simply a data interface, but a task command interface. The commands encapsulate scanning parameters, execution priorities, and possible task queue management information, enabling the radar to integrate into a higher-level collaborative observation task flow.
[0074] This architecture draws on the resource optimization principles of radar collaborative detection, transforming radar from a fixed-period data acquisition device into an intelligent sensing node driven by real-time physical events and capable of dynamically reconfiguring its sensing behavior. This forms the engineering foundation for achieving "intelligent scheduling" and "precise observation."
[0075] In summary, this method provides a complete and implementable approach and system for improving the timeliness and relevance of refined observations of severe convective weather, elevating it to a new level of closed-loop control and resource optimization.
[0076] By using lightning, a highly predictive and frequently updated detection signal, as the core scheduling basis, this method shortens the response time for refined observation of strong convection cores from the traditional passive waiting that relies on the radar's own scanning cycle (about 3 to 6 minutes) to within 1 minute of a sudden surge in lightning activity, thus gaining crucial decision-making lead time for short-term nowcasting and early warning.
[0077] The technical solutions of the present invention are not limited to the specific embodiments described above. Any technical modifications made in accordance with the technical solutions of the present invention fall within the protection scope of the present invention.
Claims
1. A radar observation and scheduling method based on real-time lightning activity sensing, characterized in that, include: S1. Real-time acquisition of lightning data and radar volume scan base data; S2. Perform quality control and spatiotemporal grid alignment on lightning data and radar volume scan base data to obtain aligned lightning data and radar mosaic data; S3. Use the aligned lightning data to identify lightning clusters and extract key features of lightning clusters; S4. Generate the effective radar echo area of the lightning cluster based on the key features of the lightning cluster, and obtain the radar echo characteristics of the lightning cluster. S5. Construct a target correlation matrix and jointly match the key features of each lightning cluster Ci with the radar echo features; S6. Identify potential targets based on the target association matrix to obtain high-potential targets; S7. Extract the enhanced feature vector of high-potential targets; S8. Select observation targets based on the enhanced feature vectors of high-potential targets; S9. Based on the observation target, schedule the corresponding X-band radar and generate observation scheduling instructions; S10. The observation and dispatch instruction is sent to the radar control terminal corresponding to the dispatched X-band radar site. S11. The radar performs RHI scanning according to the observation and dispatch instructions.
2. The radar observation and scheduling method based on real-time lightning activity perception according to claim 1, characterized in that, In S2, the quality control of lightning data is specifically carried out as follows: first, points with unreasonable metadata such as signal strength and positioning error are removed; then, noise data caused by equipment interference or ground object reflection is removed through outlier detection and spatial consistency check to obtain processed lightning data; the quality control of radar volume scan base data is specifically carried out as follows: ground object clutter suppression, velocity ambiguity removal and calibration check are applied to obtain quality-controlled radar data. The spatiotemporal grid alignment specifically involves defining a unified 3km x 3km horizontal grid and a spatiotemporal cube with a 1-minute time window. The processed lightning data and quality-controlled radar data are uniformly mapped onto a standardized Cartesian grid and a standard height layer using inverse distance weighting and nearest neighbor interpolation. The maximum combined reflectivity of the radar mosaic data in the grid and its corresponding height are recorded.
3. The radar observation and scheduling method based on real-time lightning activity sensing according to claim 1, characterized in that, S3 includes: S301. Using a 1-minute sliding window, maintain a time series with a sliding length of 15 minutes for each 3km×3km grid cell. Define the detection window and background window. Perform initial screening based on the jump condition. Grids that meet the jump condition are designated as jump grids, while grids that meet the jump condition are filtered out. S302. Extract all three-dimensional lightning points within a preset range centered on the leap grid, and cluster the three-dimensional lightning points to obtain at least one lightning cluster Ci; S303. Extract the key features of each lightning cluster, including the three-dimensional center (Center_i), three-dimensional range, dynamic features, vertical structure features, and spatial compactness.
4. The radar observation and scheduling method based on real-time lightning activity perception according to claim 3, characterized in that, S4 includes: S401. Obtain the reflectivity CR of the three-dimensional center of the lightning cluster Ci at the current moment or the most recent radar mosaic combination. S402, using radar combined reflectivity Based on the contour area, a 10km distance buffer is drawn outward to form the effective radar echo zone; S403. Determine whether the three-dimensional center Center_i of lightning cluster Ci is located within the effective radar echo area by using spatial topology. If so, proceed to S404; otherwise, remove lightning cluster Ci. S404. Query the maximum reflectivity of each height layer within the horizontal range of the lightning cluster Ci, and use it as the radar echo characteristic of the lightning cluster.
5. The radar observation and scheduling method based on real-time lightning activity perception according to claim 1, characterized in that, In S6, potential target identification is specifically performed as follows: if a lightning cluster Ci simultaneously satisfies the first and second conditions, but does not satisfy the third condition, then the lightning cluster Ci is considered a high-potential target; the first condition is: the three-dimensional center Center_i of the lightning cluster Ci is located at the radar combined reflectivity. The strong echo core or its upwind side; The second condition is: the lightning cluster Ci is in a rapid development stage; the third condition is: the lightning cluster Ci has a height_Top_i of less than 5km.
6. The radar observation and scheduling method based on real-time lightning activity sensing according to claim 1, characterized in that, S8 includes: S801. Analyze the priority score of each high-potential target. , represented as: ,in, , , and The weights for frequency variation, maximum lightning height, vertical thickness, and reflectivity are respectively. , , These correspond to the enhanced feature vector values log(ΔFreq_i+1), Height_Top_i, Vertical_Span_i, and CR_max of the high-potential target, respectively. S802, Scoring based on priority Sort the targets in descending order and select the top 3 high-potential targets as observation targets.
7. The radar observation and scheduling method based on real-time lightning activity sensing according to claim 1, characterized in that, In S9, specifically including: S901: Acquire X-band radars within a 75-kilometer radius that are currently in "scheduling" mode; S902. Select all X-band radar stations whose relative azimuth to the observed target is within a preset azimuth as pre-selected radar stations. S903. Based on the highest height of the observed target (Height_bottom_i) and the location of the pre-selected radar station, calculate the highest detection elevation angle of the pre-selected radar station, and use the pre-selected radar station whose highest detection elevation angle is within the preset detection elevation angle as the scheduled X-band radar station. S904. Generate observation scheduling instructions for the X-band radar sites to be scheduled. The observation scheduling instructions include the radar preset azimuth, preset scanning angle and radar control terminal IP. When the azimuth deviation is within ±5°, the observation scheduling instruction is a continuous scheduling command; when the azimuth deviation exceeds ±5°, the observation scheduling instruction is a single scheduling command.