Multi-source data fusion hail early warning system

Abstract

The invention discloses a hail early warning system based on multi-source data fusion, and relates to the technical field of meteorological early warning, which comprises the steps of collecting local hail reduction information, cannon station data and radar data, and analyzing hail reduction spatial and temporal distribution, cannon station hail reduction characteristics, hail paths and key areas, radar echo parameters, convective potential physical quantities and cloud system characteristics through a data processing module. The early warning generation module generates early warning information according to the analysis result and a preset rule, and then the early warning information is published by the early warning publishing module through multiple channels. By integrating multi-source data, constructing a correlation model and setting an early warning index threshold, accurate early warning of hail disasters is realized. Compared with a traditional early warning mode, the hail occurrence probability, strength and influence range can be more accurately judged, early warning can be issued in advance, sufficient response time is provided for related departments and the public, and losses caused by hail disasters are effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological warning, and specifically provides a hail warning system for multi-source data fusion. Background Technique

[0002] Traditional hail warning methods have limitations in data acquisition and analysis and processing, and it is often difficult to achieve accurate and timely warnings. Early warning systems with a single data source cannot comprehensively capture the complex meteorological conditions for hail formation, resulting in a low warning accuracy rate and insufficient lead time, making it difficult for relevant departments and the public to take effective preventive measures in the face of hail disasters and causing relatively large losses.

[0003] With the development of meteorological monitoring technology, although some new data acquisition means have emerged, how to effectively fuse multi-source data, explore the laws behind the data, and improve the performance of the warning system remains an urgent problem to be solved. Existing multi-source data fusion technologies are not yet mature in the application of hail warning, and the compatibility and synergy between different data are poor, making it difficult to form an efficient warning system and unable to meet the strict requirements for the accuracy and timeliness of hail warning in practical applications.

[0004] In view of this, this application is specifically proposed. Summary of the Invention

[0005] The purpose of the present invention is to provide a hail warning system for multi-source data fusion to solve the problems mentioned in the above background technique.

[0006] To solve the above technical problems, the hail warning system for multi-source data fusion provided by the present invention includes: a data acquisition module: collecting local hailfall information, local hailfall data from artillery stations, and radar data from local radar stations; a data processing module: classifying and storing the data by time, stations, etc. to build a database, and conducting analysis on the spatio-temporal distribution of hailfall, the characteristics of local hailfall at artillery stations, hail paths and key areas, radar echo parameters, convective potential physical quantities, and cloud system characteristics; a warning generation module: generating hail warning information based on preset rules and indicators in combination with the processing results; a warning release module: releasing warning information to relevant users through multiple channels.

[0007] Further, the spatio-temporal distribution analysis of the data processing module includes: spatial distribution: analyzing the cumulative data of local hailfall station times, and using GIS technology in combination with geographical locations to infer the influence of terrain and climate on hailfall distribution; time distribution: analyzing the annual and monthly changes in hailfall, establishing a linear model of the number of hailfall times and years, calculating the correlation coefficient, and finding that there is more hailfall from March to May and the peak is in April.

[0008] Further, the analysis of the characteristics of hail in township artillery stations by the data processing module includes: Spatial distribution: Count the number of hail occurrences from 2010 to 2023, draw a distribution map, and cluster to find the high-incidence areas; Hourly distribution: Count the number of hail occurrences at different times, draw an hourly graph, and analyze the correlation between meteorological conditions and hail time; Duration and diameter: Count the distribution, draw a chart, and calculate the occurrence probability in different situations.

[0009] Further, the analysis of hail paths and key areas by the data processing module includes: Determining the path: Select hail points, and determine the main path, source area, and direction based on radar echoes; Determining the key area: Combine geographical and hail data to find the key area and explain the reasons.

[0010] Further, the analysis of radar echo parameters by the data processing module includes: Storm type: Count the storm types at hail points, and establish a correlation model with hail intensity and range; 50dBz stretching height: Analyze the data one hour before hail, and establish a change model; Correlation analysis: Analyze the correlation between hail diameter and 50dBz stretching height; Other parameters: Conduct similar analyses on 45dBz stretching height, echo top height, and VIL, and establish a correlation model; Expansion height: Analyze the relationship between expansion height and hail formation; Establishing indicators: Select parameters as early warning reference indicators, and set thresholds to establish a comprehensive model.

[0011] Further, the analysis of convective potential physical quantities by the data processing module includes: Selecting indicators: Select multiple hail day cases and select physical quantity indicators from multiple aspects; Analysis and establishment: Use violin plots to determine indicators, analyze the changes in physical quantities at different times, and establish a correlation model; Monthly analysis: Select the physical quantities of specific months for analysis, and establish a change model to predict the potential.

[0012] Further, the analysis of cloud system characteristics by the data processing module includes: TBB characteristics: Analyze the TBB maps of cases, quantify the characteristics of cloud clusters, and establish a correlation model; TBB gradient: Analyze the TBB gradient map and establish a gradient-based early warning model.

[0013] Further, the early warning generation module generates large hail or hail early warning information according to radar echo parameters, convective potential physical quantity indicators, and cloud system characteristics according to preset rules.

[0014] Further, the early warning release module issues early warnings to local residents, meteorological and agricultural departments through text messages, broadcasts, televisions, and network platforms.

[0015] Compared with the prior art, the beneficial effects of the present invention are:

[0016] 1. Precise analysis of spatio-temporal distribution: By collecting hail data from national stations from 1961 to 2023 and township artillery stations from 2010 to 2023, combined with GIS technology, analyze the spatio-temporal distribution characteristics of hail. It can accurately determine high-hail-occurrence areas such as the urban area of Anshun, and rules such as March to May being the peak period of hail, providing a strong basis for early prevention and helping relevant departments reasonably allocate disaster prevention resources.

[0017] 2. Grasp the characteristics of hail in townships: Conduct a detailed analysis of the hail data from township artillery stations, and statistically analyze their spatial and hourly distribution characteristics, as well as the duration and diameter of hail. Information such as the large number of hail occurrences in Xixiu District and the easy occurrence of hail from afternoon to evening can be understood, providing data support for formulating targeted disaster prevention strategies in township areas and reducing the harm of hail to townships.

[0018] 3. Identify key paths and regions: With the help of radar echo analysis, determine the main hail paths and key hail areas. Warnings can be issued in advance for key regions such as Puding and Zhenning, enabling local residents and relevant departments to make preparations in advance and reducing disaster losses.

[0019] 4. Establish effective warning indicators: Select a variety of radar echo parameters for analysis and establish short-term and imminent warning indicators. Based on the correlation between indicators such as the 50 dBz stretching height and VIL and the hail diameter, set thresholds to generate warning information, improving the scientificity and accuracy of hail warnings.

[0020] 5. Analyze convective potential physical quantities: Select physical quantity indicators from multiple aspects for analysis, and establish correlation models and monthly change models. It can predict the potential of hail occurrence in different months, provide a reference for agricultural production arrangements several months in advance, avoid carrying out key farming activities during high-risk periods, and reduce agricultural losses.

[0021] 6. Research cloud system feature warnings: By analyzing the characteristics of TBB and TBB gradients, establish relevant models. Use cloud system features to assist in warnings, improve the credibility of warnings, make warning information more reliable, and enhance the public's trust in warnings.

[0022] 7. Optimize the application of the warning system: Build and apply a warning system based on actual cases, and continuously optimize parameters and rules. Through continuous improvement, the accuracy and reliability of the warning system are continuously enhanced, better playing the role of warnings and ensuring the safety of people's lives and property.

[0023] 8. Improve the professional capabilities of personnel: During the implementation of the project, the data processing, analysis, and paper writing capabilities of meteorological forecasters were cultivated. The overall technical level of the team was improved, reserve professional talents for the field of meteorological warnings, and promote the sustainable development of this field. Description of the Drawings

[0024] Figure 1In the hail warning system for multi-source data fusion, cumulative distribution of the number of hail occurrences at national weather stations in Anshun City from 1961 to 2023;

[0025] Figure 2 In the hail warning system for multi-source data fusion, annual variation trend of the number of hail occurrences in Anshun City from 1961 to 2023;

[0026] Figure 3 In the hail warning system for multi-source data fusion, correlation coefficient diagram of the number of hail occurrences and years in Anshun City from 1961 to 2023;

[0027] Figure 4 In the hail warning system for multi-source data fusion, monthly variation of the number of hail occurrences in Anshun City from 1961 to 2023;

[0028] Figure 5 In the hail warning system for multi-source data fusion, hail distribution map of counties and districts from 2010 to 2023;

[0029] Figure 6 In the hail warning system for multi-source data fusion, hail distribution map of towns and townships from 2010 to 2023;

[0030] Figure 7 In the hail warning system for multi-source data fusion, hourly distribution map of hail gun stations in towns and townships from 2010 to 2023;

[0031] Figure 8 In the hail warning system for multi-source data fusion, hail duration diagram from 2010 to 2023;

[0032] Figure 9 In the hail warning system for multi-source data fusion, hail duration proportion diagram from 2010 to 2023;

[0033] Figure 10 In the hail warning system for multi-source data fusion, hail diameter diagram from 2010 to 2023;

[0034] Figure 11 In the hail warning system for multi-source data fusion, hail diameter proportion from 2010 to 2023;

[0035] Figure 12 In the hail warning system for multi-source data fusion, northwest hail path map of Anshun City;

[0036] Figure 13 In the hail warning system for multi-source data fusion, key hail area of the northwest path in Anshun City;

[0037] Figure 14 In the hail warning system for multi-source data fusion, west or southwest hail path map of Anshun City;

[0038] Figure 15 In the hail warning system for multi-source data fusion, it is the map of the hail falling area in the western or southwestern path of Anshun City;

[0039] Figure 16 In the hail warning system for multi-source data fusion, it is the map of the proportion of storm types in Anshun City;

[0040] Figure 17 In the hail warning system for multi-source data fusion, it is the violin plot of the 50 dBz stretching height one hour before hailfall;

[0041] Figure 18 In the hail warning system for multi-source data fusion, it is the variation diagram of the 50 dBz stretching height one hour before hailfall;

[0042] Figure 19 In the hail warning system for multi-source data fusion, it is the correlation coefficient diagram between the 50 dBz stretching height and the hail diameter one hour before hailfall;

[0043] Figure 20 In the hail warning system for multi-source data fusion, it is the violin plot of the 45 dBz stretching height one hour before hailfall;

[0044] Figure 21 In the hail warning system for multi-source data fusion, it is the variation diagram of the 45 dBz stretching height one hour before hailfall;

[0045] Figure 22 In the hail warning system for multi-source data fusion, it is the correlation coefficient diagram between the 45 dBz stretching height and the hail diameter one hour before hailfall;

[0046] Figure 23 In the hail warning system for multi-source data fusion, it is the violin plot of the echo top height one hour before hailfall;

[0047] Figure 24 In the hail warning system for multi-source data fusion, it is the variation diagram of the echo top height one hour before hailfall;

[0048] Figure 25 In the hail warning system for multi-source data fusion, it is the correlation coefficient diagram between the echo top height and the hail diameter one hour before hailfall;

[0049] Figure 26 In the hail warning system for multi-source data fusion, it is the violin plot of the VIL one hour before hailfall;

[0050] Figure 27 In the hail warning system for multi-source data fusion, it is the variation diagram of the VIL one hour before hailfall;

[0051] Figure 28In the hail warning system for multi-source data fusion, it is the correlation coefficient diagram between VIL and hail diameter 1 hour before hailfall;

[0052] Figure 29 In the hail warning system for multi-source data fusion, it is the variation diagram of the extended height reaching the 0°C layer height and -20°C layer height;

[0053] Figure 30 In the hail warning system for multi-source data fusion, it is the violin diagram of physical quantities at 08:00;

[0054] Figure 31 In the hail warning system for multi-source data fusion, it is the violin diagram of physical quantities at 20:00;

[0055] Figure 32 In the hail warning system for multi-source data fusion, it is the TBB diagram of typical cases;

[0056] Figure 33 In the hail warning system for multi-source data fusion, it is the TBB gradient diagram of typical cases;

[0057] Figure 34 In the hail warning system for multi-source data fusion, it is the short-term warning index for large hailstones; the table of short-term warning indexes for hailstones mentioned in the text;

[0058] Figure 35 In the hail warning system for multi-source data fusion, it is the case table of physical quantities of hail convection potential at 08:00;

[0059] Figure 36 In the hail warning system for multi-source data fusion, it is the case table of physical quantities of hail convection potential at 20:00;

[0060] Figure 37 In the hail warning system for multi-source data fusion, it is the monthly physical quantity index of spring convection potential at 08:00;

[0061] Figure 38 In the hail warning system for multi-source data fusion, it is the monthly physical quantity index of spring convection potential at 20:00. Specific implementation manner

[0062] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0063] Please refer to Figures 1 - 38 , the present invention provides a technical solution: a hail warning system for multi-source data fusion, including:

[0064] Example of the Short-term Hail Warning System in Anshun City

[0065] Based on the data of national stations in Anshun City from 1961 to 2023, the data of township artillery stations from 2010 to 2023, and the data of Guiyang Radar Station, this example deeply analyzes the hail characteristics and constructs a warning system, aiming to improve the short-term hail forecasting ability and reduce disaster losses.

[0066] I. Data Collection and Sorting

[0067] 1. Multi-source data collection: Comprehensively collect the hail information of national stations in Anshun City from 1961 to 2023, covering the number of hail stations, time, location, etc. At the same time, obtain the hail data of township artillery stations from 2010 to 2023, including elements such as spatial-temporal distribution, duration, diameter, etc. In addition, collect the radar data of Guiyang Radar Station, including parameters such as storm type, 50dBz stretching height, 45dBz stretching height, echo top height, VIL, etc. These extensive data sources provide rich information for subsequent analysis.

[0068] 2. Data sorting and storage: Sort the collected data according to dimensions such as time and stations. The hail data of national stations are sorted by year, month, and station, the data of township artillery stations are sorted by county, township, and time, and the radar data are stored in association with other data according to the hail time and station, constructing an ordered database for convenient subsequent analysis and retrieval.

[0069] II. Analysis of the Spatial-temporal Distribution Characteristics of Hail

[0070] 1. Spatial distribution analysis: Process the cumulative data of the number of hail stations at national stations and draw a spatial distribution map (see Figure 1 ). It can be visually obtained from the map that the most hails occur in the urban area of Anshun (129 times), and the least occur in the urban area of Pingba (73 times). Using Geographic Information System (GIS) technology, combine the hail data with geographical locations to analyze factors such as terrain and climate in different regions. For example, the urban area of Anshun may be more likely to form an environment conducive to hail formation due to its terrain and special climate conditions, resulting in a relatively large number of hail occurrences.

[0071] 2. Temporal distribution analysis: Calculate the annual variation of the number of hail occurrences, and obtain the annual average number of stations as 10.2 times. The most occurred in 1971 (25 times), and the least occurred in 2010 (0 times). Use the linear regression analysis method to establish a linear model of the number of hail occurrences and years y = a + bx (where y is the number of hail occurrences, x is the year, a is the intercept, and b is the slope). After calculation, the correlation coefficient is -0.473, and it passes the significance test of 0.01 (see Figure 2 and Figure 3 ), indicating that the number of hail occurrences is negatively correlated with the year, that is, the annual average number of hails generally shows a downward trend. Statistically calculate the number of hail occurrences per month and draw a monthly variation map (seeFigure 4 ) It was found that the most hailstorms occurred from March to May, with the peak in April, showing a single-peak distribution. This is because the atmospheric instability is the strongest from March to May, providing favorable conditions for the formation of hailstones.

[0072] III. Analysis of Hailstorm Characteristics in Townships (Gun Stations)

[0073] 1. Spatial distribution characteristics of hailstorms: The number of hailstorms at township gun stations from 2010 to 2023 was statistically analyzed, and the hailstorm distribution maps of counties (districts) and townships were drawn (see Figure 5 and Figure 6 ). It can be seen from the figure that the hailstorms show a trend of more in the northwest and less in the southeast in the county (district) distribution. The number of hailstorms in Xixiu District is the largest (121 station-times), and the least in Ziyun (24 station-times). Further analysis found that the number of hailstorms in Jichang, Puding and Fenghuang, Pingba is the largest (13 times). Using the cluster analysis method to cluster the hailstorm sites, the high-incidence areas of hailstorms can be determined, providing a basis for key prevention.

[0074] 2. Hourly distribution characteristics of hailstorms: The occurrence times of hailstorms at different times were statistically analyzed, and the hourly distribution map was drawn (see Figure 7 ). It was found that hailstorms mainly occur from afternoon to evening (15-23 hours), accounting for more than 80% of this period, and the peak is at 17:00. This is because from afternoon to evening, the ground is heated and the temperature rises, and the unstable energy accumulates and is released, which is conducive to the generation of severe convective weather. By analyzing the meteorological conditions such as temperature, humidity, and wind shear at different times, a correlation model between meteorological conditions and hailstorm time is established to further explain the hourly distribution of hailstorms.

[0075] 3. Analysis of hail duration and diameter: The distribution of hail duration and diameter was statistically analyzed, and the corresponding charts were drawn (see Figure 8 , Figure 9 , Figure 10 and Figure 11 ). It was found that most of the hail duration is within 20 minutes (the proportion reaches 96.6%), and the most within 5 minutes (accounting for 62.7%); most of the hail diameters are within 0-10 mm (appearing 279 times, the proportion reaches 75.4%), and 66 times within 10-20 mm (accounting for 17.8%). Using probability statistics methods, the occurrence probabilities of different durations and diameters are calculated to provide a reference for evaluating the hail hazard degree.

[0076] IV. Analysis of Hail Paths and Key Hail Areas

[0077] 1. Determine the main hail paths: Select 96 hail points on 54 hailstorm days, and according to the displacement and development of radar echoes (see Figure 12 and Figure 14) Analysis shows that the main paths are the northwest path (appearing 71 times) and the southwest or west-southwest path (appearing 17 times). Through the trajectory analysis method, the movement trajectory of the radar echo is tracked, and it is determined that the northwest path is generated in the northwestern regions of Bijie and Liupanshui and moves in the southeast direction of Anshun; the southwest or west-southwest path is generated in the marginal regions at the junction of Liupanshui, southwestern Guizhou, or the southern part of Bijie and Anshun and moves in the eastern or northeastern direction of Anshun.

[0078] 2. Determine the key hail areas: For the northwest path, the key hail areas are in Puding and Zhenning (see Figure 13 ); for the southwest or west-southwest path, the key hail area is in Zhenning (see Figure 15 ). Using geographical information and hail data, analyze the terrain and meteorological conditions in the key areas. For example, the terrain in places such as Puding and Zhenning may block and lift the air flow, promoting the aggregation of water vapor and the development of convection, thus easily becoming key hail areas.

[0079] V. Analysis of Short-Term Warning Thresholds for Radar Echo Parameters

[0080] 1. Storm type analysis: Statistically analyze the storm types at 96 hail points (see Figure 16 ). The ordinary single cell appears the most times (52 times, accounting for 54.2%), followed by the multi-cell (37 times, accounting for 38.5%). By comparing the hail conditions under different storm types, establish a correlation model between the storm type and the hail intensity and scope. For example, it is found that the hail intensity and scope under the ordinary single cell storm may have certain characteristic patterns, providing a reference for early warning.

[0081] 2. Analysis of the 50 dBz stretching height: Analyze the 50 dBz stretching height data of 10 volume scans in the 1 hour before hail at 96 hail points (see Figure 17 and Figure 18 ), calculate the median (between 7.0 - 7.7 km) and the upper quartile (between 8.3 - 9.4 km) for each volume scan. It is observed that the upper quartile shows an overall upward trend in the 1 hour before hail and has two sudden increase peaks 42 minutes and 18 minutes before hail. Using the time series analysis method, establish a change model for the 50 dBz stretching height to predict its future change trend, providing a basis for judging the possibility and intensity of hail occurrence.

[0082] 3. Correlation analysis between hail diameter and 50 dBz stretching height: Divide the hail diameter into 5 levels from small to large (D < 5 is level 1, 5 < D < 10 is level 2, 10 < D < 15 is level 3, 15 < D < 20 is level 4, > 20 is level 5), and conduct a correlation analysis with the 50 dBz stretching height in the 1 hour before hail. Use the Pearson correlation coefficient formula (where x i is the 50 dBz stretching height, y iis the grading value of hail diameter, is the corresponding mean value, and n is the number of samples), it is found that the correlation is the best 48 minutes and 18 minutes before hailfall (passing the significance test of 0.01), and 36 minutes, 30 minutes, 24 minutes, and 12 minutes pass the significance test of 0.05 (see Figure 19 ). This indicates that there is an obvious correlation between the 50 dBz stretching height and the hail diameter, which can be used to predict the size of the hail diameter.

[0083] 4.45 dBz stretching height, echo top height, and VIL change analysis: Similarly, analyze the 45 dBz stretching height, echo top height, and VIL (see Figure 20 , Figure 21 , Figure 23 , Figure 24 , Figure 26 and Figure 27 ). The overall trend of the 45 dBz stretching height is upward, the median is between 7.9 - 8.4 km, and the upper quartile is between 9.5 - 10.1 km. The correlation with the hail diameter is not as obvious as that of the 50 dBz stretching height; the upper quartile of the echo top height shows an upward trend, the median is between 10.9 - 11.4 km, and the upper quartile is between 12.9 - 13.6 km. The correlation with the hail diameter passes the 0.05 correlation test 6 minutes and 12 minutes before hailfall; the median of VIL is between 24.6 - 35.1 g.m -3 and the upper quartile is between 42.7 - 51.7 g.m -3 . The correlation with the hail diameter is obvious and passes the significance test at multiple time periods (see Figure 22 , Figure 25 and Figure 28 ). Establish the correlation models between these parameters and the hail diameter respectively, and comprehensively evaluate their roles in hail warning.

[0084] 5. Expansion height analysis: Analyze the change in the number of times the expansion height reaches the 0℃ layer height and the -20℃ layer height 1 hour before hailfall (see Figure 29 ), and it is found that the peak value of the number of times reaching the 0℃ layer height appears 24 minutes before hailfall (90 times), and the first peak value of the number of times reaching the -20℃ layer height appears 30 minutes before hailfall. Use statistical analysis methods to study the relationship between the expansion height and the formation and development of hail, and provide a reference for warning.

[0085] 6. Establish short-term warning indicators for radar echo parameters: Select the upper quartiles of the 50 dBz and 45 dBz stretching heights, echo top height, and Vil 1 hour before hailfall as the reference indicators for large hail warning, and the median as the reference indicator for hail warning (see Figure 34) According to the degree of significant correlation, the 50 dBz stretching height and VIL are defined as the first-level indicators, and the 45 dBz stretching height and echo top height are defined as the second-level indicators. By setting the thresholds of different indicators, a comprehensive early warning model is established. For example, when the 50 dBz stretching height is greater than 8.7 km and VIL is greater than 50.7 g.m -3 a severe hail warning is issued.

[0086] VI. Analysis of Physical Quantities of Hail Convection Potential

[0087] 1. Case Selection and Physical Quantity Index Selection: On the basis of the original 11 cases, 44 new cases are added, and potential analysis is carried out on 15 physical quantities at 08:00 and 20:00 on 54 hail days from 2013 to 2023 (see Figure 35 and Figure 36 ). Physical quantity indicators are selected from aspects such as stratification stability (such as K-index, Si, Li, etc.), dynamic and thermodynamic conditions (such as SWEAT, DCAPE, CAPE, etc.), vertical wind shear (wind vector difference at 3 km and 6 km), and special stratification heights (0℃ layer height, -20℃ layer height, IQ, etc.).

[0088] 2. Analysis and Establishment of Physical Quantity Indicators: By plotting violin plots (see Figure 30 and Figure 31 ), the median of the physical quantities in the figure is selected as the physical quantity indicator of hail convection potential, and the upper quartile is used as the severe hail indicator. Analyze the changes of each physical quantity at different times (08:00 and 20:00). For example, the value of the K-index increases at 20:00 compared with 08:00, indicating that the degree of stratification instability increases at 20:00; CAPE is stronger at 20:00 than at 08:00, indicating that convection develops more strongly at 20:00. Using statistical analysis methods, an association model between physical quantity indicators and the possibility of hail occurrence is established. For example, through logistic regression analysis, the influence weights of each physical quantity on hail occurrence are determined.

[0089] 3. Monthly Analysis of Physical Quantity Indicators of Convection Potential: Select the physical quantities during hailfall from March to May for analysis (see Figure 37 and Figure 38), it was found that in terms of stratification stability, the K index and T85 were the largest in May at 08:00 from March to May, LI, Ls, and SI were gradually decreasing, and the K index also gradually increased at 20:00, LI, Ls, and SI were gradually decreasing, T85 and T75 were the largest in March, and the overall stratification instability gradually increased; in terms of dynamic and thermal conditions, DCAPE, CAPE, etc. were gradually increasing at 08:00 and 20:00 from March to May, and SWEAT was close to the critical value of strong thunderstorms; in terms of vertical wind shear, Vws had an insignificant monthly increase trend at 08:00, and the 3km and 6km wind vector differences decreased monthly at 08:00 and 20:00; in terms of special height layers and the integral of the specific humidity of the whole layer, the 0℃ layer, -20℃ layer, and the integral height of the specific humidity of the whole layer all had an obvious upward trend at 08:00 and 20:00 from March to May. By establishing a monthly physical quantity change model, the potential for hail occurrence in different months is predicted.

[0090] 7. Analysis of hail-causing cloud characteristics

[0091] 1. Analysis of TBB characteristics: Analysis of 9 individual TBB graphs one hour before the occurrence of hail (see Figure 32 ), it was found that convective clouds were mainly small and medium-sized, the shape of the cloud cover was not very regular, mainly blocky or small elliptical, and the central value intensity was between -20 and -60 ° C. Using image processing technology, the shape, scale and intensity of the cloud were quantitatively analyzed, and a correlation model between cloud characteristics and hail occurrence was established.

[0092] 2. TBB gradient analysis: Analyze the TBB gradient diagrams of 9 individual hailstorms one hour before they occurred (see Figure 33 ) and found that before the hail occurred, there was an obvious temperature gradient in Anshun or its marginal areas, and the gradient showed an obvious band distribution with a gradient value between 20-50°C. By analyzing the relationship between temperature gradient and hail occurrence, a hail warning model based on TBB gradient was established. For example, when the temperature gradient is greater than 30°C, the credibility of the hail warning is increased.

[0093] 8. Construction and application of early warning system

[0094] 1. Construction of early warning system: Integrate the above analysis results to build a multi-source data fusion Anshun hail early warning system. The system includes data acquisition module, data processing module, early warning generation module and early warning release module. The data acquisition module is responsible for collecting multi-source data; the data processing module analyzes and processes the data and calculates various indicators and parameters; the early warning generation module generates hail early warning information according to the preset early warning rules and indicators; the early warning release module releases the early warning information to relevant users through various channels.

[0095] 2. Application of the early warning system: Take the hail process in Jichangpo, Puding on May 11, 2015 as an example. The system collects hail information, radar data and physical quantity data of the site and its surroundings. After analysis by the data processing module, it is found that the 50dBz extension height, VIL and other indicators have reached the warning threshold. Combined with the cloud system characteristics and the results of the convective potential physical quantity analysis, the early warning generation module generates hail warning information and releases it to local residents and relevant departments in a timely manner through the early warning release module to remind them to take preventive measures. Through the application of multiple actual cases, the parameters and rules of the early warning system are continuously optimized to improve the accuracy and reliability of the early warning.

[0096] From the above, we can see that:

[0097] (1) According to the national stations in Anshun City from 1961 to 2023, the urban area of ​​Anshun had the most hail, while the urban area of ​​Pingba had the least hail. The annual average number of hail showed an overall downward trend, and the number of hail was negatively correlated with the year.

[0098] (2) From 2010 to 2023, the overall distribution of township artillery stations in Anshun City in counties (districts) showed a trend of more in the northwest and fewer in the southeast, with the largest number of stations (times) in Xixiu District, followed by 84 stations (times) in Zhenning County.

[0099] (3) The monthly variation of hail showed a unimodal distribution, with April being the annual peak. Hail mainly occurred in the afternoon and evening, with the peak period from 15:00 p.m. to 23:00 p.m., and the peak period of hail at 17:00. The duration of hail was mostly within 20 minutes, accounting for 96.6%. The density of hail was mostly within 0-10 mm, which occurred 279 times, accounting for 75.4%; 10-20 mm occurred 66 times, accounting for 17.8%.

[0100] (4) The hail path that affects Anshun is mainly the northwest path, and the key hail areas of this path are mainly concentrated in Puding, Zhenning, etc. The second is the southwest or west path, and the key hail area of ​​this path is in Zhenning.

[0101] (5) The monomers that affect the hail in Anshun are mainly ordinary monomers, followed by multiple monomers. The hail diameter has a significant correlation with the 50dBz extension height and VIL, but the correlation with the 45dBz extension height and echo top height is not as obvious as the 50dBz extension height.

[0102] (6) Convective clouds that cause hail are mainly small and medium-sized. The shape of the cloud cover is not very regular, and is mainly block-shaped or small elliptical. There is an obvious temperature gradient in Anshun or its marginal areas.

[0103] In summary: By collecting the hail information from national stations and township artillery stations in Anshun City and the data from Guiyang Radar Station, and using means such as GIS technology, statistical analysis methods, and model establishment, the present invention deeply analyzes the spatio-temporal distribution of hail, the characteristics of hail in township artillery stations, the hail path and key areas, radar echo parameters, convective potential physical quantities, and cloud system characteristics, and constructs a multi-source data fusion hail warning system. This system realizes the accurate warning of hail disasters, can timely issue warning information, helps relevant departments and the public to take preventive measures in advance, and effectively reduces the losses caused by hail disasters.

Claims

1. A hail warning system for multi-source data fusion, characterized in that: It includes: Data acquisition module: Collect local hail information, local hail data of artillery stations, and radar data of local radar stations; Data processing module: Classify and store the data by time and station to build a database, and conduct analyses on the spatio-temporal distribution of hail, the characteristics of local hail at artillery stations, hail paths and key areas, radar echo parameters, convective potential physical quantities, and cloud system characteristics; Early warning generation module: Generate hail early warning information based on preset rules and indicators, combined with the processing results; Early warning release module: Release early warning information to relevant users through multiple channels.

2. The hail warning system for multi-source data fusion according to claim 1, characterized in that: The spatio-temporal distribution analysis of hail in the data processing module includes: Spatial distribution: Analyze the cumulative data of local hail station times, and use GIS technology combined with geographical locations to infer the influence of terrain and climate on hail distribution; Time distribution: Analyze the annual and monthly changes of hail, and establish a linear model between the number of hail occurrences and years.

3. The hail warning system for multi-source data fusion according to claim 1, characterized in that: The analysis of the characteristics of hail at township artillery stations in the data processing module includes: Spatial distribution: Statistically analyze the number of hail occurrences from 2010 to 2023, draw a distribution map, and cluster to find the high-incidence areas; Hourly distribution: Statistically analyze the number of hail occurrences at different times, draw an hourly map, and analyze the correlation between meteorological conditions and hail time; Duration and diameter: Statistically analyze the distribution, draw charts, and calculate the occurrence probabilities of different situations.

4. The hail warning system for multi-source data fusion according to claim 1, wherein: The analysis of hail paths and key areas in the data processing module includes: Determine the path: Select hail points, and determine the main path, source area, and direction based on radar echoes; Determine the key area: Combine geographical and hail data to find the key area and explain the reasons.

5. The hail warning system for multi-source data fusion according to claim 1, wherein: The analysis of radar echo parameters in the data processing module includes: Storm type: Statistically analyze the storm types at hail points, and establish a correlation model with hail intensity and range; 50dBz stretching height: Analyze the data one hour before hail, and establish a change model; Correlation analysis: Analyze the correlation between hail diameter and 50dBz stretching height; Other parameters: Conduct similar analyses on 45dBz stretching height, echo top height, and VIL, and establish a correlation model; Expansion height: Analyze the relationship between expansion height and hail formation; Establish indicators: Select parameters as early warning reference indicators, and set thresholds to build a comprehensive model.

6. The hail warning system for multi-source data fusion according to claim 1, characterized in that: The analysis of convective potential physical quantities in the data processing module includes: Select indicators: Select multiple hail day cases and select physical quantity indicators from multiple aspects; Analysis and establishment: Use violin plots to determine indicators, analyze the changes of physical quantities at different times, and establish a correlation model; Monthly analysis: Select the physical quantities of specific months for analysis, and establish a change model to predict the potential.

7. The hail warning system for multi-source data fusion according to claim 1, characterized in that: The analysis of cloud system characteristics in the data processing module includes: TBB characteristics: Analyze the TBB maps of cases, quantify the characteristics of cloud clusters, and establish a correlation model; TBB gradient: Analyze the TBB gradient map and establish a gradient-based early warning model.

8. The hail warning system for multi-source data fusion according to claim 1, characterized in that: The early warning generation module generates large hail or hail early warning information according to radar echo parameters, convective potential physical quantity indicators, and cloud system characteristics, in accordance with preset rules.

9. The hail warning system for multi-source data fusion according to claim 1, characterized in that: The early warning release module releases early warnings to local residents, meteorological and agricultural departments through text messages, radio, television, and online platforms.

Images