A unified weather object construction method and system based on multi-source data

Abstract

The application relates to the technical field of weather observation, in particular to a unified weather object construction method and system based on multi-source data, which comprises the following steps: acquiring multi-source meteorological data in a target area, performing unified space-time organization processing, generating a plurality of meteorological element feature sequences, and mapping the sequences to corresponding unified grids; constructing a weather object generation index corresponding to each unified grid to determine a plurality of candidate weather object regions; analyzing object space aggregation indexes corresponding to any two candidate weather object regions, judging whether each candidate weather object region meets an aggregation condition, if the aggregation condition is met, constructing a unified weather object model corresponding to a to-be-constructed weather object; calculating an object matching index corresponding to the to-be-constructed weather object, and judging whether an object matching condition is met to determine an object matching type of the to-be-constructed weather object; and establishing a weather object life cycle chain. The application can realize automatic construction of weather objects.

Description

Technical Field

[0001] This invention relates to the field of weather observation technology, and in particular to a method and system for constructing a unified weather object based on multi-source data. Background Technology

[0002] Current meteorological operational systems are primarily built around observational and product data. Weather radar systems provide data in the form of combined reflectivity, CAPPI, and echo top height; satellite systems provide data in the form of cloud images, cloud cover, and cloud top height; automatic weather station systems provide observational data on temperature, humidity, precipitation, wind speed, and visibility; and numerical models provide future weather forecasts in the form of grid forecast fields.

[0003] While existing systems can access, display, and analyze meteorological data, their organization primarily remains at the grid, cell, or layer level. For operational personnel, the real focus is often not on individual grid values, but on weather entities with complete spatial extent and evolutionary processes. For example, a rainstorm corresponds to a precipitation area, a severe convective event corresponds to a cluster of thunderstorm cells, a strong wind event corresponds to a wind-affected area, and low visibility weather corresponds to a fog area.

[0004] Currently, meteorological operational systems lack a unified system for representing weather objects. Different systems often develop independent algorithms for single weather phenomena, resulting in a lack of unified management of weather objects and hindering advanced applications such as weather event construction, weather process cognitive analysis, and automatic generation of digital twin weather scenarios. With the development of the low-altitude economy, digital twin cities, and artificial intelligence technologies, there is a need for a method that can automatically convert multi-source meteorological data into unified weather objects, enabling the object-oriented representation and management of weather processes. Therefore, proposing a method for automatically constructing weather objects based on multi-source meteorological data is of great significance. Summary of the Invention

[0005] To address this, the present invention provides a unified weather object construction method and system based on multi-source data, in order to overcome the problem of the lack of object-oriented representation of weather processes in the prior art.

[0006] To achieve the above objectives, on the one hand, the present invention provides a method for constructing a unified weather object based on multi-source data, comprising: Multi-source meteorological data within the target area is acquired, and unified spatiotemporal organization processing is performed based on the data type and the object type of the weather object to be constructed to generate several meteorological element feature sequences corresponding to the weather object to be constructed, and mapped to the corresponding unified grid. The multi-source meteorological data includes at least one of weather radar data, meteorological satellite data, ground observation data, lightning location data, wind profile data, radiosonde data, and numerical forecast data. The target area includes several unified grids. Based on the meteorological element feature sequences and data quality coefficients, a weather object generation index is constructed for each unified grid to determine several candidate weather object regions, wherein any candidate weather object region includes several candidate grids. Analyze the spatial proximity, attribute similarity, intensity continuity, and temporal continuity among the candidate weather object regions to obtain the object spatial aggregation index corresponding to any two candidate weather object regions; Based on the spatial aggregation index of each object, it is determined whether each candidate weather object region meets the aggregation condition. If the aggregation condition is met, the candidate weather object regions are aggregated to construct a unified weather object model corresponding to the weather object to be constructed. The search range is established based on the unified weather object model to calculate the object matching index corresponding to the weather object to be constructed, and to determine whether the object matching conditions are met, so as to determine the object matching type of the weather object to be constructed. The lifecycle state of the weather object to be constructed is determined based on the object matching type and the weather object generation index, so as to establish a weather object lifecycle chain.

[0007] Furthermore, constructing the weather object generation index corresponding to any of the aforementioned unified grids includes: The meteorological element feature sequence corresponding to the unified grid is normalized to obtain the corresponding standard meteorological element feature sequence. The data quality coefficient sequence corresponding to the standard meteorological element feature sequence is determined based on the data quality control results in the unified spatiotemporal organization and processing process. The basic weight sequence corresponding to the standard meteorological element feature sequence is determined based on the preset weight calibration rule; The weather object generation index corresponding to the unified grid is constructed based on the standard meteorological element feature sequence, the data quality coefficient sequence, and the basic weight sequence.

[0008] Further, several candidate weather target areas are identified, including: Several core candidate grids are determined based on the comparison results between the weather object index corresponding to each unified grid and the preset core threshold. Based on the comparison results between the weather object index corresponding to each unified grid and the preset core threshold and preset edge threshold, several edge undetermined grids are determined. Based on the distribution relationship of each core candidate grid and each edge undetermined grid, several candidate weather object regions are determined.

[0009] Furthermore, the object spatial aggregation index corresponding to any two candidate weather object regions is calculated, including: The spatial proximity between the two candidate weather object regions is determined based on the shortest distance between their boundaries and the distance scale parameter. Based on the object type, construct the regional attribute vectors corresponding to the two candidate weather object regions to determine the attribute similarity between the two candidate weather object regions. The shortest connection zone is established between the adjacent boundaries of the two candidate weather object regions, and the intensity continuity between the two candidate weather object regions is determined based on the weather object generation index and the average intensity difference within the connection zone. The temporal continuity between the two candidate weather object areas is determined based on the degree of overlap between the two candidate weather object areas and the previous time weather object area. The corresponding object spatial aggregation index is determined based on the spatial proximity, attribute similarity, intensity continuity, and temporal continuity between the two candidate weather object regions.

[0010] Furthermore, based on the relative condition that the object space aggregation index between any two candidate weather object areas is greater than a preset aggregation threshold, it is determined that the two candidate weather object areas meet the aggregation condition.

[0011] Further, the object matching index corresponding to the weather object to be constructed is calculated, including: Based on the unified weather object model and the historical movement speed of the previous weather object, the current centroid position of the weather object to be constructed is predicted to establish the search range; The historical weather objects within the search range and the weather objects to be constructed are matched to obtain the object matching index corresponding to the weather objects to be constructed.

[0012] Further, determining the object matching type of the weather object to be constructed includes: Whether the object matching condition is met is determined based on the comparison result between the object matching index corresponding to the weather object to be constructed at any time and the preset matching index threshold. Specifically, if a historical weather object and a weather object to be constructed in consecutive time periods meet the object matching conditions, the object matching type is determined to be a split object type. If several historical weather objects and any weather object to be constructed in any time period meet the object matching conditions, the object matching type is determined to be a merged object type. If the object matching conditions are not met, the object matching type is determined to be a newly generated object type.

[0013] Furthermore, establish a lifecycle chain for weather objects, including: If the weather object to be constructed does not meet the object matching condition with the previous time object and the weather object generation index corresponding to the unified weather object model meets the generation threshold condition, then the life cycle state is determined to be the generation state. If the lifecycle state of the weather object to be constructed is the generation state, and the weather object to be constructed continues to exist in the next time step, or the weather object generation index corresponding to the unified weather object model meets the high intensity threshold condition, then the weather object lifecycle chain is confirmed to be established.

[0014] Furthermore, it also includes: The object correlation index is determined based on the spatial correlation, temporal correlation, physical attribute correlation, and evolution synchronization among several different types of weather objects within the target area. A weather object relationship network is constructed based on the object association index between different types of weather objects. The weather object relationship network has weather objects as nodes and object relationships as edges. The object relationships are determined based on the object association index.

[0015] On the other hand, the present invention also provides a unified weather object construction system, comprising: The feature mapping module is used to acquire multi-source meteorological data within the target area, perform unified spatiotemporal organization processing based on data type and object type of the weather object to be constructed, generate several meteorological element feature sequences corresponding to the weather object to be constructed, and map them to the corresponding unified grid. The multi-source meteorological data includes at least one of weather radar data, meteorological satellite data, ground observation data, lightning location data, wind profile data, radiosonde data, and numerical weather prediction data. The target area includes several unified grids. The region division module is used to construct the weather object generation index corresponding to each unified grid based on the feature sequence of each meteorological element and the data quality coefficient, so as to determine several candidate weather object regions, wherein any candidate weather object region includes several candidate grids. The aggregation analysis module is used to analyze the spatial proximity, attribute similarity, intensity continuity and temporal continuity among the candidate weather object areas to obtain the object spatial aggregation index corresponding to any two candidate weather object areas. The model building module is used to determine whether each candidate weather object region meets the aggregation conditions based on the spatial aggregation index of each object. If the aggregation conditions are met, the candidate weather object regions are aggregated to build a unified weather object model corresponding to the weather object to be built.

[0016] The object matching module is used to establish a search range based on the unified weather object model, calculate the object matching index corresponding to the weather object to be constructed, and determine whether the object matching conditions are met, so as to determine the object matching type of the weather object to be constructed, and determine the life cycle state of the weather object to be constructed based on the object matching type and the weather object generation index, so as to establish a weather object life cycle chain.

[0017] Compared with existing technologies, the advantages of this invention lie in its ability to analyze the data types of multi-source meteorological data within the target area and the object types of the weather objects to be constructed, and to perform targeted unified spatiotemporal organization processing to generate corresponding weather element feature sequences for each unified grid, thereby achieving refined management of weather processes. By comprehensively considering the meteorological element feature sequences and data quality coefficients, the weather object generation index corresponding to each unified grid is determined. A data quality adaptive mechanism is introduced to quantify the weather characteristics of each unified grid, accurately identifying candidate weather object regions. Furthermore, the corresponding object spatial index is calculated based on the multi-dimensional correlation relationships of each candidate weather object region, analyzing whether there is a possibility of aggregation among the candidate weather objects. This allows for the construction of a complete unified weather object model for the weather objects to be constructed, realizing the object-oriented expression and management of weather processes. Finally, a weather object lifecycle chain is established through continuous time-series object matching, achieving reliable matching of weather objects across consecutive time periods and accurate identification of complex evolutionary relationships.

[0018] Furthermore, by introducing a data quality coefficient, adaptive correction of observation reliability is achieved. For the normalized standard meteorological element feature sequence, the corresponding basic weights are calibrated. The weather object generation index constructed in this way realizes the effective fusion of multi-source features, which can improve the accuracy of subsequent weather object analysis and identification.

[0019] Furthermore, high-confidence identification is ensured through core candidate grids, while object boundary regions are preserved through edge undetermined grids, overcoming the defects of boundary shrinkage or noise introduction in single-threshold segmentation. Connectivity is determined based on the spatial distribution relationship between core candidate grids and edge undetermined grids, ensuring the physical continuity of candidate regions and realizing rational identification from grid-level response to object-level candidate regions, effectively adapting to complex weather environments.

[0020] Furthermore, the aggregation conditions among candidate regions are comprehensively evaluated from four independent dimensions: spatial, attribute, intensity, and temporal. Spatial proximity identifies regions that are discontinuous but belong to the same system; attribute similarity ensures physical consistency; intensity continuity characterizes the transitional state of the connecting zone; and temporal continuity utilizes historical information to enhance decision-making stability. The fusion of the four dimensions forms a unified aggregation criterion, enabling accurate merging of the same weather system and reliable differentiation of regions with different properties. Attached Figure Description

[0021] Figure 1 This is a structural block diagram of the unified weather object construction system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the unified weather object construction method based on multi-source data according to an embodiment of the present invention. Figure 3 A schematic diagram illustrating the process of generating an index for a weather object corresponding to any unified grid in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the process of determining several candidate weather object areas in an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0023] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0024] Please see Figure 1 The diagram shown is a structural block diagram of the unified weather object construction system according to an embodiment of the present invention; the unified weather object construction system provided by the embodiment of the present invention includes: The feature mapping module is used to acquire multi-source meteorological data within the target area, perform unified spatiotemporal organization processing based on data type and object type of the weather object to be constructed, generate several meteorological element feature sequences corresponding to the weather object to be constructed, and map them to the corresponding unified grid. The multi-source meteorological data includes at least one of weather radar data, meteorological satellite data, ground observation data, lightning location data, wind profile data, radiosonde data, and numerical weather prediction data. The target area includes several unified grids. The region division module, which is connected to the feature mapping module, is used to construct the weather object generation index corresponding to each unified grid based on the feature sequence of each meteorological element and the data quality coefficient, so as to determine several candidate weather object regions, wherein any candidate weather object region includes several candidate grids. The aggregation analysis module, which is connected to the region division module, is used to analyze the spatial proximity, attribute similarity, intensity continuity and temporal continuity among the candidate weather object regions to obtain the object spatial aggregation index corresponding to any two candidate weather object regions. The model building module is connected to the aggregation analysis module and the region division module respectively. It is used to determine whether each candidate weather object region meets the aggregation conditions based on the spatial aggregation index of each object. If the aggregation conditions are met, the candidate weather object regions are aggregated to build a unified weather object model corresponding to the weather object to be built. The object matching module, which is connected to the module construction module and the region division module respectively, is used to establish a search range based on the unified weather object model, calculate the object matching index corresponding to the weather object to be constructed, and determine whether the object matching conditions are met, so as to determine the object matching type of the weather object to be constructed, and determine the life cycle state of the weather object to be constructed based on the object matching type and the weather object generation index, so as to establish a weather object life cycle chain.

[0025] Please see Figure 2 The diagram shown is a flowchart illustrating the unified weather object construction method based on multi-source data according to an embodiment of the present invention. The method provided by this embodiment includes: Step S1: Obtain multi-source meteorological data within the target area, perform unified spatiotemporal organization processing based on data type and object type of the weather object to be constructed, generate several meteorological element feature sequences corresponding to the weather object to be constructed, and map them to the corresponding unified grid. The multi-source meteorological data includes at least one of weather radar data, meteorological satellite data, ground observation data, lightning location data, wind profile data, radiosonde data, and numerical forecast data. The target area includes several unified grids. In this embodiment, the latitude and longitude boundaries of the target area are determined according to the scope of business applications. The scope of business applications can be a province, a river basin, or a specific urban cluster, etc., and a uniform spatial resolution and a uniform time interval are set. For example, for severe convective weather monitoring, the uniform spatial resolution can be set to 1 km × 1 km, and the uniform time interval to 6 minutes; for regional precipitation monitoring, it can be set to 3 km × 3 km with a 10-minute interval; for large-scale cloud system analysis, it can be set to 5 km × 5 km with a 30-minute or 1-hour interval. In actual implementation, the resolution and time interval can be flexibly adjusted according to computing resources, data density, and business timeliness requirements. The target area is uniformly divided into several uniform grids, each grid having a unique row and column index and center coordinates. A data time window is established for each uniform time period. The width of the time window can be set according to the data type. For example, the radar data time window can be set to ±3 minutes, satellite data to ±10 minutes, and ground station data to ±5 minutes. Various meteorological data falling within this time window are grouped into the same processing batch. If a certain type of data exists in multiple observation times within the time window, the record closest to the uniform time period is selected for calculation.

[0026] Understandably, for regular grid data output by weather radar, meteorological satellites, and numerical weather prediction models, spatial resampling is performed based on the original grid center coordinates and the unified grid center coordinates. For continuous meteorological elements such as combined reflectivity, vertical integral liquid water content, echo top height, precipitation, temperature, humidity, wind speed, cloud cover, and cloud top height, bilinear interpolation or area-weighted interpolation methods are used to map the original grid values ​​onto the unified grid. For discrete elements such as cloud type, weather phenomenon codes, and quality labels, nearest neighbor interpolation is used to avoid interpolation producing intermediate category values ​​without physical meaning. For radar three-dimensional volume scan data, a CAPPI (contour plane position display) grid field or a three-dimensional grid field is first formed according to preset height layers (such as 0.5km, 1km, 2km, 3km, 5km, 7km, 10km), and then mapped to the unified grid. For discrete point data such as those from automatic weather stations and lightning location systems, a unified grid is used as the statistical unit for gridding. For cumulative elements such as precipitation and lightning strikes, the cumulative amount or density per unit area of ​​all observations falling within the grid is calculated. For elements characterized by extreme values ​​such as maximum wind speed, the maximum value of all observations within the grid is taken. For continuous elements such as temperature, humidity, average wind speed, and visibility, the average or median of all valid observations within the grid is taken. If the observation points within the grid are sparse, the nearest station value or inverse distance weighted interpolation can also be used to obtain the grid value. The specific statistical method used needs to be determined based on the physical characteristics of the weather object to be constructed: for example, when constructing a thunderstorm object, the lightning density is the cumulative number of strikes divided by the grid area; when constructing a strong wind object, the maximum wind speed is taken as the maximum value within the grid; when constructing a low visibility object, the visibility is taken as the average value within the grid. When multiple observation stations of the same type exist within a unified grid, appropriate statistics are selected based on the physical mechanism triggered by the object. For weather objects triggered by extreme events, such as strong winds and heavy precipitation, the maximum value or high percentile value of the observations within the grid is used to reflect the extreme intensity. For objects triggered by regional background conditions, such as temperature, humidity, and cloud cover, the average value or median of the observations within the grid is used to reflect the regional average conditions.

[0027] Understandably, when a unified grid lacks valid observation data, the first step is to determine if the number of its neighboring valid grids reaches a preset limit. If so, interpolation is performed using inverse distance weighting or neighborhood averaging methods to supplement the data. If valid data is insufficient, the grid is marked as missing, and the data quality coefficient of the corresponding feature of that grid is set to zero in subsequent calculations. The missing data marker is transmitted along with the meteorological element feature sequence to ensure that subsequent steps can detect the data loss. For each data source, if the deviation between its original observation time and the unified time does not exceed the allowable time difference (e.g., 3 minutes for radar and 5 minutes for ground stations), the data from that time is directly used. If two adjacent time periods are valid and the target time is located between them, linear time interpolation is used to obtain the value of the target time. If the deviation exceeds the allowable time difference, the data is not directly used in the calculation, or its data quality coefficient is reduced, for example, from 1.0 to 0.6. The time deviation information is recorded as part of the data quality coefficient.

[0028] After the above processing, each unified grid cell generates a meteorological element feature sequence, which includes: grid identifier (row and column number), center coordinates (longitude and latitude), unified time, values ​​of each meteorological feature factor (such as reflectivity, precipitation, wind speed, lightning density, etc.), data source identifier (such as radar, satellite, ground station, etc.), data quality coefficients corresponding to each feature (reflecting data validity, timeliness, interpolation status and quality control results), and validity identifier (valid / missing data). Among them, the feature factors can be dynamically selected according to the type of object to be constructed. For example, for thunderstorm objects, combined reflectivity, VIL, echo top height, and lightning density are selected, and for precipitation objects, radar reflectivity, surface precipitation, grid precipitation products, etc. are selected. Each meteorological element feature sequence includes several meteorological element features.

[0029] Step S2: Construct a weather object generation index corresponding to each unified grid based on the feature sequence of each meteorological element and the data quality coefficient, so as to determine several candidate weather object regions, wherein any candidate weather object region includes several candidate grids. Specifically, in step S2, constructing the weather object generation index corresponding to any of the unified grids includes: Step S21: Normalize the meteorological element feature sequence corresponding to the unified grid to obtain the corresponding standard meteorological element feature sequence. In this embodiment, since the characteristic dimensions and numerical ranges of each meteorological element are different, the system first converts the characteristic factors of each meteorological element to the range of 0 to 1 before calculating the Weather Object Formation Index (WOI). For positive factors where a larger characteristic value is more conducive to the formation of weather objects, the normalized value corresponding to the i-th meteorological element characteristic is: ,in, For the first The actual values ​​of each meteorological element characteristic For the first The characteristics of meteorological elements begin to support the lower limit of weather object formation. For the first When the characteristics of a meteorological element reach the upper limit of sufficient support, hour, ;when hour, `min()` is a preset function to determine the minimum value, and `max()` is a preset function to determine the maximum value. For inverse factors such as visibility, where smaller values ​​are more favorable for weather object formation, the normalized value corresponding to the i-th meteorological feature is: ,in, This indicates the low threshold value corresponding to severe weather conditions. The high threshold value at which the object begins to form.

[0030] Step S22: Determine the data quality coefficient sequence corresponding to the standard meteorological element feature sequence based on the data quality control results in the unified spatiotemporal organization and processing process; In this embodiment, each meteorological element feature has a corresponding data quality coefficient. The data quality coefficient can quantify the reliability of each meteorological element feature value in subsequent fusion calculations. The value ranges from 0 to 1. For example, for data with normal original data, data time deviation within the allowable range and passing data quality control, the data quality coefficient takes the maximum value of 1. For data generated by spatial interpolation or temporal interpolation, or data with slight time deviation, the data quality coefficient takes a medium value, such as 0.5 to 0.9. The greater the deviation, the lower the data quality coefficient. For data with missing original data, unqualified quality control, or time deviation exceeding the allowable range, the data quality coefficient takes 0. Meteorological element feature factors with a quality coefficient of 0 do not participate in subsequent weighted calculations. This data quality coefficient sequence corresponds one-to-one with the standard meteorological element feature sequence, ensuring that each meteorological element feature value carries reliability information to participate in fusion in subsequent steps.

[0031] Step S23: Determine the basic weight sequence corresponding to the standard meteorological element feature sequence based on the preset weight calibration rules; In this embodiment, the basic weights reflect the relative contribution of each meteorological element characteristic factor to the formation of the weather object to be constructed. The preset weight calibration rule adopts a combination of expert knowledge and historical sample correction. First, initial expert values ​​are set based on weather mechanisms, operational experience, data stability, and spatial representativeness. Then, the weights are corrected using confirmed historical weather object samples. The expert weights and sample correction weights are combined according to the following formula: ,in, As the initial weights for experts, Adjusting weights for historical samples, The combination coefficient is set between 0.5 and 0.8. When historical samples are scarce, the expert weight ratio is increased; when historical samples are abundant, the sample correction ratio is increased. After combination, all weights are normalized so that their sum equals 1, ensuring numerical comparability of weather object generation indices corresponding to different object types. Historical sample correction employs a grid search or stepwise trial method, adjusting each weight in steps of 0.05 or 0.1. Hit rate, false alarm rate, and critical success index are calculated on confirmed object samples and non-object samples. The comprehensive evaluation function is expressed as: ,in, For hit rate, This is the critical success index. For false alarm rate, , and As for the weighting of the evaluation indicators, the weights of each indicator can be adjusted according to the business focus. Larger weightings are used as historical sample correction weights.

[0032] Step S24: Construct the weather object generation index corresponding to the unified grid based on the standard meteorological element feature sequence, the data quality coefficient sequence, and the basic weight sequence.

[0033] In this embodiment, the Weather Object Generation Index (WOI) is calculated according to the following formula: ,in, For the first The basic weights of meteorological element characteristics For the first Data quality coefficients corresponding to each meteorological element characteristic For the first The standard meteorological element characteristic value. The weather object generation index is a dimensionless standardized comprehensive index with a value range between 0 and 1. It represents the comprehensive confidence that the unified grid supports the existence of the weather object to be constructed: the closer the value is to 1, the higher the probability that the weather object to be constructed exists in the grid; the closer the value is to 0, the lower the probability that the weather object to be constructed exists in the grid.

[0034] In a specific embodiment, if the weather object to be constructed is a thunderstorm object, then the combined reflectivity CREF, vertical integral liquid water content VIL, echo top height ET, and lightning density LD are selected as feature factors: The combined reflectivity directly reflects the intensity of convective echoes, and therefore has the highest weight; VIL reflects the liquid water content and potential precipitation intensity inside the convective cloud; echo top height reflects the degree of vertical development; and lightning density reflects the electrical activity of the cloud. For example, during a thunderstorm, the normalized lower and upper limits of the combined reflectivity are 35 dBZ and 55 dBZ, respectively; the lower and upper limits of VIL are 10 kg / m² and 40 kg / m², respectively; the lower and upper limits of echo top height are 6 km and 12 km, respectively; and the lightning density threshold is determined based on historical samples from the region.

[0035] Specifically, by introducing a data quality coefficient, adaptive correction of observation reliability is achieved. For the normalized standard meteorological element feature sequence, the corresponding basic weights are calibrated. The weather object generation index constructed in this way realizes the effective fusion of multi-source features, which can improve the accuracy of subsequent weather object analysis and identification.

[0036] Specifically, in step S2, several candidate weather object areas are determined, including: Step S25: Determine several core candidate grids based on the comparison results between the weather object index corresponding to each unified grid and the preset core threshold. Step S26: Based on the comparison results between the weather object index corresponding to each unified grid and the preset core threshold and preset edge threshold, determine a number of edge undetermined grids; Step S27: Determine several candidate weather object regions based on the distribution relationship of each core candidate grid and each edge undetermined grid.

[0037] In this embodiment, WOI is greater than or equal to a preset core threshold. The unified grid is determined as the core candidate grid. The preset core threshold is usually set at a relatively high level to ensure that the core candidate grid has high confidence and can reliably represent the core area of ​​the weather object to be constructed. The preset core threshold can be set according to the object type and business needs. For example, in thunderstorm monitoring scenarios, a higher preset core threshold can be set to ensure detection accuracy, while in post-disaster assessment scenarios, a relatively lower preset core threshold can be set to improve comprehensiveness. The WOI is then placed at the preset edge threshold. With preset core threshold The unified grid between them is determined as the edge undetermined grid, where, The preset edge threshold is usually set between the preset core threshold and the background noise level, so that the grid between the two thresholds represents a position with a certain support strength but whose confidence has not yet reached the core level. These positions are usually the boundary transition areas of weather objects.

[0038] Understandably, based on the core candidate grid, several initial connected regions are formed through neighborhood connection analysis. It is then determined whether the edge candidate grid has an 8-grid neighborhood connection with the core candidate grid, and whether the physical characteristics of the grid meet the preset continuity condition. If both conditions are met, the corresponding edge candidate grid is included in the candidate weather object region; otherwise, the edge candidate grid is not included in the candidate weather object region and is only regarded as a background region in subsequent processing.

[0039] In this embodiment, after the candidate weather object region is initially formed, isolated point filtering, small region removal, and hole filling can be performed. Isolated regions with an area smaller than the minimum area threshold and a maximum WOI that does not reach the strong object retention threshold are removed. Non-candidate weather object region grids that are surrounded by the same candidate weather object region and have an area smaller than the hole threshold are filled. In practice, the implementer can set the corresponding minimum area threshold, strong object retention threshold, and hole threshold based on the actual situation, without making specific limitations here.

[0040] Specifically, the core candidate grid ensures high-confidence identification, while the edge undetermined grid preserves the object boundary region, overcoming the defects of boundary shrinkage or noise introduction in single-threshold segmentation. Based on the spatial distribution relationship between the core candidate grid and the edge undetermined grid, connectivity is determined to ensure the physical continuity of the candidate region, realizing the rational identification from grid-level response to object-level candidate region, and effectively adapting to complex weather environments.

[0041] Step S3: Analyze the spatial proximity, attribute similarity, intensity continuity, and temporal continuity among the candidate weather object regions to obtain the object spatial aggregation index corresponding to any two candidate weather object regions. Specifically, in step S3, the object spatial aggregation index corresponding to any two candidate weather object areas is calculated, including: The spatial proximity between the two candidate weather object regions is determined based on the shortest distance between their boundaries and the distance scale parameter. In this embodiment, the boundary point sets of two candidate weather object regions P and Q are extracted, and the shortest distance between the boundaries of the two regions is calculated. Then the spatial proximity is: ,in, This is a distance scale parameter used to control the rate at which spatial proximity decays with increasing distance. It can be set to 1 to 3 times the edge length of a uniform grid. When two regions are directly adjacent, , This indicates that two areas are directly connected or adjacent in space, exhibiting the highest tendency for spatial aggregation. As distance increases, The spatial proximity gradually decreases, with closer regions having a higher spatial proximity. When the shortest distance exceeds the preset maximum aggregation distance, the spatial proximity is set to 0, indicating that the two regions are too far apart and should not be considered for aggregation. For example, in raster data implementation, the calculation can be simplified based on grid adjacency relationships: when two candidate weather object regions have an 8-neighbor connection, the spatial proximity is directly set to 1; when two candidate weather object regions are separated by one grid, the spatial proximity is a value between 0.5 and 0.8; when the number of grids between them exceeds the preset maximum aggregation distance, the spatial proximity is set to 0. The introduction of spatial proximity ensures that spatial aggregation only occurs between geographically adjacent candidate regions, avoiding the inclusion of spatially distant irrelevant regions in the merging consideration, effectively limiting the search range and reducing computational complexity.

[0042] Based on the object type, construct the regional attribute vectors corresponding to the two candidate weather object regions to determine the attribute similarity between the two candidate weather object regions. In this embodiment, different object types use different methods to represent their physical attributes. For example, a thunderstorm object should construct a regional attribute vector containing average reflectivity, maximum reflectivity, average vertical integral liquid water content, and average echo top height; a precipitation object should construct a regional attribute vector containing average precipitation intensity, maximum precipitation intensity, and average reflectivity; and a strong wind object should construct a regional attribute vector containing average wind speed, maximum wind speed, and wind direction consistency. In actual implementation, the dimensions and specific physical attribute items included in the regional attribute vector are dynamically determined based on the object type of the weather object to be constructed. For the first... The attributes, with normalized differences, are: ,in, and For the two candidate weather object regions One attribute, This represents the valid range of variation for this attribute. To prevent tiny positive numbers with a denominator of 0, the attribute similarity is: ,in, For attribute internal weights, all The sum of the physical attributes of two regions is 1. The closer the physical attributes of two regions are, the closer the attribute similarity is to 1, indicating that they are more likely to belong to the same weather system. Conversely, the greater the difference in attributes, the lower the attribute similarity, indicating that even if they are spatially adjacent, they may be different weather phenomena. Attribute similarity ensures that the merged weather object has consistent physical attributes, effectively avoiding the erroneous merging of weather phenomena of different natures into the same object.

[0043] The shortest connection zone is established between the adjacent boundaries of the two candidate weather object regions, and the intensity continuity between the two candidate weather object regions is determined based on the weather object generation index and the average intensity difference within the connection zone. In this embodiment, the closest pair of boundary points are determined between the boundaries of candidate weather object region P and candidate weather object region Q. Based on these points, a shortest connecting line is established between the two regions, and this connecting line is extended into a connecting band of a certain width. The width of the connecting band can be set according to the grid resolution, for example, set to 1 to 3 grid widths to ensure coverage of the main grid in the transition area between the two regions. After the connecting band is established, the intensity continuity is evaluated from two dimensions. The first dimension is the boundary intensity difference. The average intensity values ​​of candidate weather object region P and candidate weather object region Q within the adjacent boundary band are calculated. The average intensity value can be a weather object generation index, or it can be a major physical quantity characterizing the intensity level of the weather object to be constructed, such as the reflectivity of a thunderstorm object or the wind speed of a strong wind object. The continuity corresponding to the boundary intensity difference is: ,in, and The average intensity of the adjacent boundary zone between the two regions. Within the permissible range of strength difference, the smaller the strength difference between the two sides of the boundary, the closer this continuity component is to 1, indicating a smooth transition in strength between the two regions. The second dimension is the effective proportion of the connecting band. , This represents the number of grid cells within the connection band that reach the edge candidate threshold. The total number of grid cells in the connecting zone indicates a higher effective proportion of connecting zones, signifying a continuous high-exponential connection between the two regions rather than an abrupt break. The closer this continuity component is to 1, the stronger the continuity between the two candidate weather object regions is represented as: Among them, the smaller the intensity difference between two candidate weather object regions and the more connected grids that reach the edge threshold, the better. The larger the intensity continuity, the better. Intensity continuity reflects the local intensity distribution of the connecting bands between candidate regions, providing crucial local evidence for merging decisions.

[0044] The temporal continuity between the two candidate weather object areas is determined based on the degree of overlap between the two candidate weather object areas and the previous time weather object area. In this embodiment, temporal continuity is used to determine whether two current candidate weather object regions originated from the same historical object in the previous time period. It obtains the spatial range of each confirmed weather object region from the previous time period. For two candidate weather object regions P and Q in the current time period, they are merged into a temporary region (P∪Q). The spatial overlap between this temporary region and each historical object region from the previous time period is calculated, and the maximum value is taken as the temporal continuity. The spatial overlap can be calculated as the ratio of the intersection area to the union area of ​​the two regions. Let the weather object in the previous time period be... Then the time continuity When two candidate regions originated from the same historical weather object in a previous time period, the temporal continuity is high, indicating that merging these two candidate regions into one object in the current time period conforms to historical evolution patterns and helps maintain the consistency of the object's identity. When two candidate regions belonged to different historical objects in a previous time period, the temporal continuity is low, indicating that merging should be done with caution. For initial time periods that are formed for the first time and do not have any historical objects, Take a neutral value of 0.5 and redistribute its corresponding weights. , and This ensures that the aggregation decision can still be completed normally within this time frame.

[0045] The corresponding object spatial aggregation index is determined based on the spatial proximity, attribute similarity, intensity continuity, and temporal continuity between the two candidate weather object regions.

[0046] In this embodiment, after preliminary connectivity analysis, candidate regions may exhibit both the same weather system separated by weak discontinuities and spatially adjacent regions with different physical properties. The system uses an object spatial aggregation index to determine whether two candidate regions meet the aggregation conditions. and Its corresponding object space aggregation index ,in, These are the corresponding weights. , For spatial proximity, For attribute similarity, For strength continuity, Temporal continuity is considered. Spatial proximity is a fundamental condition for aggregation and carries a high weight; attribute similarity ensures consistency in physical properties; intensity continuity identifies weak discontinuities; and temporal continuity maintains the stability of the object structure across different time periods. The object spatial aggregation index is a dimensionless comprehensive score ranging from 0 to 1. It comprehensively reflects the overall aggregation tendency of two candidate regions across four dimensions: spatial location, physical attributes, intensity distribution, and historical evolution. The closer the object spatial aggregation index is to 1, the more likely the two candidate weather object regions should be aggregated into the same weather object; the closer the object spatial aggregation index is to 0, the more likely the two candidate weather object regions should remain independent.

[0047] Specifically, the aggregation conditions among candidate regions are comprehensively evaluated from four independent dimensions: spatial, attribute, intensity, and temporal. Spatial proximity identifies regions that are discontinuous but belong to the same system; similar attributes ensure physical consistency; continuous intensity characterizes the transitional state of the connecting zone; and continuous time utilizes historical information to enhance decision-making stability. The fusion of the four dimensions forms a unified aggregation criterion, enabling accurate merging of the same weather system and reliable differentiation of regions with different properties.

[0048] Step S4: Based on the spatial aggregation index of each object, determine whether each candidate weather object region meets the aggregation condition. If the aggregation condition is met, aggregate each candidate weather object region to construct a unified weather object model corresponding to the weather object to be constructed. Specifically, in step S4, based on the relative condition that the object space aggregation index between any two candidate weather object areas is greater than a preset aggregation threshold, it is determined that the two candidate weather object areas meet the aggregation condition, and then the two candidate weather object areas are aggregated.

[0049] In this embodiment, each generated candidate weather object region is treated as an initial object, and its spatial boundary, internal grid set, and basic attributes are recorded. All initial objects are paired, and the object spatial aggregation index of each pair is calculated. Region pairs with object spatial aggregation indices greater than a preset aggregation threshold are added to the aggregation candidate queue and sorted from high to low according to their object spatial aggregation indices. The preset aggregation threshold can be set according to the object type and business requirements, or based on the average object spatial aggregation index of the candidate weather object regions generated during the historical weather object formation process. Region pairs are sequentially taken from the aggregation candidate queue, and for each pair of regions, it is determined whether the following condition is met: the object spatial aggregation index is greater than the preset aggregation threshold. If the condition is met, the two candidate weather object regions are merged into a new region. In a specific embodiment, aggregation conditions can be added: spatial proximity greater than a preset first threshold and attribute similarity greater than a preset second threshold. The spatial proximity condition and attribute similarity condition are necessary conditions to prevent a high-scoring indicator from masking other obviously unreasonable indicators. For example, when two regions have extremely low spatial proximity or extremely low attribute similarity, they should not be merged even if the object spatial aggregation index reaches the aggregation threshold. When all conditions are met, the two candidate weather object regions are merged into a new region.

[0050] Understandably, after each merge, the spatial aggregation index of the merged new region and the surrounding regions that have not yet participated in the merge is recalculated. If the aggregation conditions are met, the merge continues, and the above process is repeated until there are no regions that meet the aggregation conditions. This iterative mechanism ensures that multiple candidate regions are merged step by step to form a complete weather object. For example, three spatially adjacent candidate regions may first be merged with region B, then with region C, and finally form a single object.

[0051] Understandably, for each final weather object formed after aggregation, a unique object identifier is generated. This identifier consists of the object type code, the time of first generation, and a sequence number. For example, "TH-202605070200-001" represents the first thunderstorm object generated at 2:00 AM on May 7, 2026. Then, various attributes of the weather object are calculated to establish a unified weather object model: spatial attributes include the boundary polygon, area, perimeter, centroid, circumscribed rectangle, major axis, minor axis, azimuth angle, and the top, bottom, and volume of the three-dimensional object; physical attributes are determined according to the object type and at least include the maximum, minimum, and average values ​​of the main meteorological characteristics. The model uses high percentile values ​​and standard deviations, and applies quality weights to each attribute. Statistical attributes include the number of grids, the proportion of valid data, the number of data sources, duration, historical maximum area, historical maximum intensity, and cumulative movement distance. Evolutionary attributes include area change rate, intensity change rate, movement speed, movement direction, lifecycle status, and splitting / merging information. Relationship attributes include associated object identifiers, relationship types, association indices, start times, and end times. The unified weather object model provides a data foundation for subsequent continuous tracking and lifecycle management. Impact attributes include the impact range, impact level, and spatial relationships with targets such as airports, low-altitude air routes, schools, hospitals, and administrative regions.

[0052] Step S5: Establish a search range based on the unified weather object model to calculate the object matching index corresponding to the weather object to be constructed, and determine whether the object matching conditions are met, so as to determine the object matching type of the weather object to be constructed. In this embodiment, weather objects undergo dynamic changes such as location shifts, morphological changes, and intensity evolution over consecutive time periods. By establishing predicted locations and search ranges, the weather object of the current time period is matched with historical weather objects of the previous time period to determine the identity correspondence of the same weather object across consecutive time periods.

[0053] Specifically, in step S5, the object matching index corresponding to the weather object to be constructed is calculated, including: Based on the unified weather object model and the historical movement speed of the previous weather object, the current centroid position of the weather object to be constructed is predicted to establish the search range; In this embodiment, for the previous time object The system predicts the current centroid position based on historical movement speeds: ,in, and This represents the movement speed of the object in the previous stage in both directions. When there is no historical movement speed, the original centroid is used as the predicted position for the time interval, and the search range is established according to the maximum reasonable movement distance of the object type.

[0054] The historical weather objects within the search range and the weather objects to be constructed are matched to obtain the object matching index corresponding to the weather objects to be constructed.

[0055] In this embodiment, the object matching index is calculated for historical weather objects and the current weather object to be constructed within the search range: ,in, For the boundary intersection and union ratio, For centroid distance similarity, For area similarity, For intensity similarity, For height or vertical structural similarity, when a two-dimensional object does not have a height attribute, the weight of the height item is reassigned to other metrics. The boundary intersection-union ratio is the ratio of the area of ​​the intersection of the boundaries of two objects to the area of ​​their union. Centroid distance similarity is calculated based on the centroid distance and a preset distance scale parameter; the closer the distance, the higher the similarity. Area similarity is calculated based on the relative difference in the areas of the two objects; the smaller the difference, the higher the similarity. Intensity similarity is calculated based on the relative difference in the overall intensity of the two objects. Vertical structure similarity is calculated based on the relative difference in vertical features such as echo top height or cloud top height, and is applicable to objects with three-dimensional structures. The object matching index is a dimensionless comprehensive score, ranging from 0 to 1, comprehensively reflecting the confidence that historical weather objects and the current weather object to be constructed belong to the same lifecycle chain. For two-dimensional objects without height attributes, the weight of vertical structure similarity can be redistributed to the other four indicators.

[0056] Specifically, in step S5, determining the object matching type of the weather object to be constructed includes: The object matching condition is determined based on the comparison result between the object matching index corresponding to the weather object to be constructed at any time and the preset matching index threshold. If the object matching index is greater than or equal to the preset matching index threshold, it is determined that the current weather object to be constructed and the corresponding historical weather object meet the object matching condition, and the two belong to the same life cycle chain of continuous objects. The current object inherits the object identifier of the historical object. If the object matching index is less than the preset matching index threshold, it is determined that the object matching condition is not met.

[0057] Understandably, if a historical weather object meets the object matching condition with multiple weather objects to be constructed in consecutive time periods, it is determined that the historical weather object has split in the corresponding time period, and the object matching type is split object type. If multiple historical weather objects meet the object matching condition with the same weather object to be constructed in the current time period, it is determined that the multiple historical weather objects are merged into the same weather object in the current time period, and the object matching type is merged object type. If the weather object to be constructed in the current time period does not meet the object matching condition with any historical weather object, it is determined that the weather object to be constructed is a newly generated object, and the object matching type is newly generated object type, and a new weather object identifier is assigned to it. If a historical weather object does not match any current weather object within a consecutive preset number of time periods (such as 2 to 3 time periods), it is determined that the historical weather object has been destroyed.

[0058] Step S6: Determine the lifecycle state of the weather object to be constructed based on the object matching type and the weather object generation index, so as to establish a weather object lifecycle chain.

[0059] Specifically, in step S6, the lifecycle chain of the weather object is established, including: If the weather object to be constructed does not meet the object matching condition with the previous time object and the weather object generation index corresponding to the unified weather object model meets the generation threshold condition, then the life cycle state is determined to be the generation state. If the lifecycle state of the weather object to be constructed is in the generating state, and the weather object continues to exist in the next time period, or the weather object generation index corresponding to the unified weather object model meets the high-intensity threshold condition, then the weather object lifecycle chain is confirmed to be established. For objects that only appear once and have low intensity, they can be marked as short-term pending confirmation objects and not included in the formal lifecycle chain. Implementers can set the generation threshold condition and the high-intensity threshold condition based on the actual situation.

[0060] Understandably, for weather objects whose lifecycle chains have been confirmed, when a weather object exists continuously for at least two time periods and its evolution index is continuously not less than a preset development threshold, its lifecycle state is determined to be in a development state. This state indicates that the object is strengthening or expanding, and its intensity is trending upward. When a weather object persists for a preset duration, such as existing continuously for more than a certain number of time periods, and its evolution index is within a preset stable range, indicating that changes in attributes such as area and intensity are slowing down, and its current comprehensive intensity reaches or exceeds a preset proportion of its historical maximum intensity, its lifecycle state is determined to be in a mature state. This state indicates that the object is in the peak stage of its lifecycle, and its intensity is stable at a high level. When the object's evolution index is continuously not greater than a preset decay threshold, its lifecycle state is determined to be in a decay state. If at least two of the following occur simultaneously: area shrinkage, intensity reduction, and height decrease, the confidence level of the decay judgment can be improved. This state indicates that the object is weakening or shrinking. In a specific embodiment, the object evolution index is: ,in, The rate of change of area, For the rate of change of intensity, For a high rate of change, The rate of change of movement is the rate of change of area. The rate of change in intensity is: The height change rate is: The rate of change of movement is used to describe the acceleration or deceleration of an object and is corrected for by directional continuity: ,in, and Used to prevent abnormal amplification of the rate of change of small or weak objects. For the reference range of height variation, As a reference range for velocity changes, when a purely two-dimensional object does not use a height parameter, it will... Set to zero and renormalize the other weights.

[0061] Understandably, to reduce frequent lifecycle state switching caused by observational fluctuations, state transition constraints are set: the generated state can be converted to the developing state, mature state, or directly disappear; the developing state can be converted to the mature state, decaying state, or continue to maintain the developing state; the mature state can be converted to the developing state (object secondary enhancement), decaying state, or continue to maintain the mature state; the decaying state can return to the developing state when its intensity significantly increases, otherwise it will continue to decay until it disappears. State determination requires at least 1 to 2 consecutive time intervals to meet the corresponding conditions before confirmation.

[0062] Specifically, after step S6, the following steps are also included: The object correlation index is determined based on the spatial correlation, temporal correlation, physical attribute correlation, and evolution synchronization among several different types of weather objects within the target area. In this embodiment, the specific method for determining the object association index includes: for objects and : ,in, For spatial correlation, For time correlation, For physical attribute correlation, For evolutionary synchronization, , , and These are the corresponding weights. Spatial correlation considers both intersection-union ratio and inclusion ratio: , , When there is no overlap but the distance is relatively short, the nearest neighbor score is supplemented based on the shortest distance at the boundary. The temporal correlation is: ,in, For the time scale corresponding to the object type, the weather objects at the same time... The physical attribute correlation is set to 1. Relationship rules are established based on object type. For example, the correlation between thunderstorms and precipitation is calculated based on the correspondence between reflectivity and precipitation intensity, consistency of intensity levels, and distance between the centers of intensity values. Evolutionary synchronization is based on the correlation coefficient of the intensity change sequences of two objects over a number of recent time periods. calculate: When the number of historical data points is insufficient, A neutral value of 0.5 is used. An object relationship is established when the Object Association Index (OAI) is not less than the association threshold. Implementers can set the association threshold based on design requirements. The Object Association Index is a dimensionless comprehensive score, ranging from 0 to 1, which comprehensively reflects the confidence level that there is an inherent relationship between two different objects.

[0063] A weather object relationship network is constructed based on the object association index between different types of weather objects. The weather object relationship network has weather objects as nodes and object relationships as edges. The object relationships are determined based on the object association index.

[0064] In this embodiment, a symbiotic relationship is established when two weather objects continuously overlap or are adjacent to each other during the same time period and there is no clear sequential relationship; a triggering relationship is established when one object is generated or enhanced first, and then another object appears in the neighboring area with a sequential evolution; an inclusion relationship is established when one object is mainly located inside another object; an influence relationship is established when an object overlaps with a business objective or administrative region and reaches an influence threshold; and an evolutionary relationship is established when objects of the same type match in consecutive time periods, or when they split, merge, or transform. The object relationship network uses weather objects as nodes and object relationships as edges. Each edge records at least the source object identifier, target object identifier, relationship type, OAI value, start time, end time, and main determination criteria.

[0065] Understandably, to determine whether an object is suitable for inclusion in the formal object repository, the system further calculates an object completeness index: ,in, For spatial integrity, For attribute completeness, For time continuity, For data source integrity. The corresponding weights are as follows: Spatial integrity is calculated based on the degree of boundary closure, the proportion of internal voids, and the proportion of isolated grids; Attribute integrity is calculated based on the proportion of valid attributes among the required attributes of the target object; Temporal continuity is calculated based on whether there has been a successful match in the most recent time intervals; Data source integrity is calculated based on the ratio of the planned data source to the actual valid data source. Objects with an OCI of not less than 0.8 are marked as high-trust objects, those between 0.6 and 0.8 are generally trustworthy objects, and those below 0.6 are low-trustworthy objects. Low-trustworthy objects are reserved as objects to be verified and are not directly used for relational network or digital twin entity generation.

[0066] Understandably, when weather objects are mapped to a digital twin scene, the system first selects a weather entity template based on the object type, and then converts the object's boundaries, centroid, top height, bottom height, intensity level, direction of movement, and lifecycle status into the scene entity's position, horizontal range, vertical range, density, transparency, color, particle count, movement speed, and display status. Cloud objects use two-dimensional boundaries to determine the horizontal range of the cloud entity, cloud base and top height to determine the vertical range, and cloud cover and type to determine coverage, density, and morphology templates. Precipitation objects use boundaries to determine the precipitation particle range, and precipitation intensity to determine particle count, particle size, and falling speed. Thunderstorm objects establish a three-dimensional spatial range based on object boundaries and echo top height, and determine entity density, color, and lightning effects based on maximum reflectivity, VIL, and lightning density. Strong wind objects generate wind-affected areas or risk bodies based on boundaries and direction of movement, and determine streamline velocity, particle velocity, or risk level based on wind speed level. Low visibility objects use boundaries to determine the fog area range, and use visibility values ​​to inversely map fog density. In consecutive time intervals, the digital twin system maintains the continuous existence of entities based on object identifiers, and interpolates and updates position, range, density, transparency, and motion state between adjacent data intervals. When an object is in the generation state, it gradually appears; in the development state, it expands or strengthens; in the mature state, it remains relatively stable; in the decay state, its density or range decreases; and in the dying state, it fades away and releases resources.

[0067] This invention analyzes the data types of multi-source meteorological data within the target area and the object types of the weather objects to be constructed, and performs targeted unified spatiotemporal organization processing to generate corresponding weather element feature sequences for each unified grid, achieving refined management of weather processes. By comprehensively considering the meteorological element feature sequences and data quality coefficients, the weather object generation index corresponding to each unified grid is determined. A data quality adaptive mechanism is introduced to quantify the weather characteristics of each unified grid, accurately identifying candidate weather object regions. Furthermore, the corresponding object spatial index is calculated based on the multi-dimensional correlations of each candidate weather object region, analyzing whether there is a possibility of aggregation among the candidate weather objects. This constructs a complete unified weather object model for the weather objects to be constructed, realizing the object-oriented expression and management of weather processes. Finally, a weather object lifecycle chain is established through continuous time-series object matching, achieving reliable matching of weather objects across consecutive time periods and accurate identification of complex evolutionary relationships.

[0068] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for constructing a unified weather object based on multi-source data, characterized in that, include: Multi-source meteorological data within the target area is acquired, and unified spatiotemporal organization processing is performed based on the data type and the object type of the weather object to be constructed to generate several meteorological element feature sequences corresponding to the weather object to be constructed, and mapped to the corresponding unified grid. The target area includes several unified grids. Based on the meteorological element feature sequences and data quality coefficients, a weather object generation index is constructed for each unified grid to determine several candidate weather object regions, wherein any candidate weather object region includes several candidate grids. Analyze the spatial proximity, attribute similarity, intensity continuity, and temporal continuity among the candidate weather object regions to obtain the object spatial aggregation index corresponding to any two candidate weather object regions; Based on the spatial aggregation index of each object, it is determined whether each candidate weather object region meets the aggregation condition. If the aggregation condition is met, the candidate weather object regions are aggregated to construct a unified weather object model corresponding to the weather object to be constructed. The search range is established based on the unified weather object model to calculate the object matching index corresponding to the weather object to be constructed, and to determine whether the object matching conditions are met, so as to determine the object matching type of the weather object to be constructed. The lifecycle state of the weather object to be constructed is determined based on the object matching type and the weather object generation index, so as to establish a weather object lifecycle chain.

2. The method for constructing a unified weather object based on multi-source data according to claim 1, characterized in that, Constructing the weather object generation index corresponding to any of the aforementioned unified grids includes: The meteorological element feature sequence corresponding to the unified grid is normalized to obtain the corresponding standard meteorological element feature sequence. The data quality coefficient sequence corresponding to the standard meteorological element feature sequence is determined based on the data quality control results in the unified spatiotemporal organization and processing process. The basic weight sequence corresponding to the standard meteorological element feature sequence is determined based on a preset weight calibration rule. The weather object generation index corresponding to the unified grid is constructed based on the standard meteorological element feature sequence, the data quality coefficient sequence, and the basic weight sequence.

3. The method for constructing a unified weather object based on multi-source data according to claim 2, characterized in that, Determine several candidate weather object areas, including: Several core candidate grids are determined based on the comparison results between the weather object index corresponding to each unified grid and the preset core threshold. Based on the comparison results between the weather object index corresponding to each unified grid and the preset core threshold and preset edge threshold, several edge undetermined grids are determined. Based on the distribution relationship of each core candidate grid and each edge undetermined grid, several candidate weather object regions are determined.

4. The method for constructing a unified weather object based on multi-source data according to claim 3, characterized in that, Calculate the object spatial aggregation index corresponding to any two candidate weather object regions, including: The spatial proximity between the two candidate weather object regions is determined based on the shortest distance between their boundaries and the distance scale parameter. Based on the object type, construct the regional attribute vectors corresponding to the two candidate weather object regions to determine the attribute similarity between the two candidate weather object regions. The shortest connection zone is established between the adjacent boundaries of the two candidate weather object regions, and the intensity continuity between the two candidate weather object regions is determined based on the weather object generation index and the average intensity difference within the connection zone. The temporal continuity between the two candidate weather object areas is determined based on the degree of overlap between the two candidate weather object areas and the previous time weather object area. The corresponding object spatial aggregation index is determined based on the spatial proximity, attribute similarity, intensity continuity, and temporal continuity between the two candidate weather object regions.

5. The method for constructing a unified weather object based on multi-source data according to claim 4, characterized in that, Based on the relative condition that the object space aggregation index between any two candidate weather object areas is greater than a preset aggregation threshold, it is determined that the two candidate weather object areas meet the aggregation condition.

6. The method for constructing a unified weather object based on multi-source data according to claim 5, characterized in that, Calculating the object matching index corresponding to the weather object to be constructed includes: Based on the unified weather object model and the historical movement speed of the previous weather object, the current centroid position of the weather object to be constructed is predicted to establish the search range; The historical weather objects within the search range and the weather objects to be constructed are matched to obtain the object matching index corresponding to the weather objects to be constructed.

7. The method for constructing a unified weather object based on multi-source data according to claim 6, characterized in that, Determining the object matching type of the weather object to be constructed includes: Whether the object matching condition is met is determined based on the comparison result between the object matching index corresponding to the weather object to be constructed at any time and the preset matching index threshold. Specifically, if a historical weather object and a weather object to be constructed in consecutive time periods meet the object matching conditions, the object matching type is determined to be a split object type. If several historical weather objects and any weather object to be constructed in any time period meet the object matching conditions, the object matching type is determined to be a merged object type. If the object matching conditions are not met, the object matching type is determined to be a newly generated object type.

8. The method for constructing a unified weather object based on multi-source data according to claim 7, characterized in that, Establish a lifecycle chain for weather objects, including: If the weather object to be constructed does not meet the object matching condition with the previous time object and the weather object generation index corresponding to the unified weather object model meets the generation threshold condition, then the life cycle state is determined to be the generation state. If the lifecycle state of the weather object to be constructed is the generation state, and the weather object to be constructed continues to exist in the next time step, or the weather object generation index corresponding to the unified weather object model meets the high intensity threshold condition, then the weather object lifecycle chain is confirmed to be established.

9. The method for constructing a unified weather object based on multi-source data according to claim 8, characterized in that, Also includes: The object correlation index is determined based on the spatial correlation, temporal correlation, physical attribute correlation, and evolution synchronization among several different types of weather objects within the target area. A weather object relationship network is constructed based on the object association index between different types of weather objects. The weather object relationship network has weather objects as nodes and object relationships as edges. The object relationships are determined based on the object association index.

10. A unified weather object construction system, which employs the unified weather object construction method based on multi-source data as described in any one of claims 1-9, characterized in that, include: The feature mapping module is used to acquire multi-source meteorological data within the target area, perform unified spatiotemporal organization processing based on data type and object type of the weather object to be constructed, generate several meteorological element feature sequences corresponding to the weather object to be constructed, and map them to the corresponding unified grid. The multi-source meteorological data includes at least one of weather radar data, meteorological satellite data, ground observation data, lightning location data, wind profile data, radiosonde data, and numerical weather prediction data. The target area includes several unified grids. The region division module is used to construct the weather object generation index corresponding to each unified grid based on the feature sequence of each meteorological element and the data quality coefficient, so as to determine several candidate weather object regions, wherein any candidate weather object region includes several candidate grids. The aggregation analysis module is used to analyze the spatial proximity, attribute similarity, intensity continuity and temporal continuity among the candidate weather object areas to obtain the object spatial aggregation index corresponding to any two candidate weather object areas. The model building module is used to determine whether each candidate weather object region meets the aggregation conditions based on the spatial aggregation index of each object. If the aggregation conditions are met, the candidate weather object regions are aggregated to build a unified weather object model corresponding to the weather object to be built. The object matching module is used to establish a search range based on the unified weather object model, calculate the object matching index corresponding to the weather object to be constructed, and determine whether the object matching conditions are met, so as to determine the object matching type of the weather object to be constructed, and determine the life cycle state of the weather object to be constructed based on the object matching type and the weather object generation index, so as to establish a weather object life cycle chain.

Images