Hail landing time prediction and evaluation method based on multivariate data

Abstract

The invention provides a hail landing time prediction and evaluation method based on multivariate data, and the method comprises the steps: extracting the ascending airflow driving data of a height corresponding to the distribution range of a key region, analyzing the fluctuation amplitude and frequency change of the ascending airflow speed of different height layers along with the time, and obtaining the spatial-temporal characteristics of the convection intensity change, determining the relevance between the convection intensity change and the ice crystal nucleus generation rate, and outputting the corrected ice crystal nucleus generation rate according to the relevance; and dynamically adjusting simulation parameters of the ice crystal nucleus generation rate according to the correlation coefficient of the convection intensity and the corrected ice crystal nucleus generation rate, obtaining the distribution of the adjusted ice crystal nucleus generation rate, and simulating the evolution path of the hail inoculation process by combining the spatial-temporal characteristics of the change of the convection intensity to obtain a dynamic simulation result of the hail inoculation process.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for predicting and assessing hailfall time based on multivariate data. Background Technology

[0002] The formation process of hail in clouds is a key area of ​​meteorological research, directly related to extreme weather forecasting and disaster prevention. With climate change, the increasing frequency and intensity of hail events make accurate prediction of their formation time and fall window crucial. Currently, hail prediction methods based on cloud temperature gradients and convection intensity have a certain theoretical foundation, but significant shortcomings remain in practical applications. A core challenge in this research is how the dynamic changes in internal cloud temperature distribution affect the ice crystal nucleation rate. Since temperature differences at different altitudes of clouds directly determine the rate of water vapor condensation and ice crystal formation, the principle is that supercooled water droplets in the low-temperature region of the upper atmosphere remain liquid when they reach below the freezing point. When the temperature drops further to the heterogeneous nucleation threshold (i.e., the minimum temperature required for supercooled water droplets to form ice crystals on the surface of condensation nuclei, usually between -5℃ and -40℃, depending on the physicochemical properties of the condensation nuclei), water vapor molecules rapidly aggregate around the condensation nuclei to form ice crystals. The greater the temperature gradient and the higher the degree of supercooling, the faster the probability and rate of ice crystal nucleation. The non-uniformity of temperature distribution may lead to significant fluctuations in the rate of ice crystal nucleation, which in turn affects the initial formation conditions of hail. This fluctuation further triggers unstable changes in the intensity of convection within the cloud layer. Specifically, when the vertical temperature gradient of the cloud increases, the density difference between the warm air at the bottom and the cold air at the top increases, creating stronger buoyancy and accelerating the updraft. Fluctuations in the ice crystal nucleus formation rate alter the distribution of latent heat release within the cloud. Increased latent heat release in localized areas further intensifies the upward movement of warm, moist air in those areas, leading to pulsating changes in convection intensity. When convection intensity strengthens or weakens, the hail formation process may be accelerated or delayed, making the prediction of the landing time window more complex. Existing technologies, when fusing multi-level temperature data, lack dynamic adjustment strategies and cannot effectively capture these interrelated changes. Therefore, how can we dynamically adjust the simulation strategy for ice crystal nucleus formation rate and convection intensity by fusing multi-level cloud temperature data to adapt to the rapidly changing cloud environment? Summary of the Invention

[0003] This invention provides a method for predicting and evaluating hailfall time based on multivariate data, mainly including: Temperature distribution data and vertical wind speed data at different cloud altitudes are acquired. Temperature values ​​and updraft speeds at each altitude are recorded and preliminarily processed in real time to obtain a complete dataset of cloud temperature distribution data and updraft-driven data. Temperature gradient changes between altitudes are analyzed, and a stratified calculation method is used to determine the water vapor condensation rate under the influence of the temperature gradient, obtaining the distribution characteristics of the water vapor condensation rate. Based on the distribution characteristics of the water vapor condensation rate and combined with the temperature gradient data, the dynamic changes in the ice crystal nucleus formation rate are simulated. If the water vapor condensation rate at a certain altitude exceeds a preset threshold, that altitude is marked as a key region for ice crystal nucleus formation, and the distribution range of the key region is obtained. If the water vapor condensation rate at a certain altitude does not exceed the preset threshold, that altitude is marked as an inactive region, and its water vapor condensation rate change trend continues to be monitored. Updraft-driven data corresponding to the distribution range of the key regions are extracted. The fluctuation amplitude and frequency changes of updraft speed at different altitudes over time are analyzed to obtain the spatiotemporal characteristics of convection intensity changes, determining the correlation between convection intensity changes and ice crystal nucleus formation rate. Based on the correlation, a corrected ice crystal nucleus formation rate is output. Based on the convection intensity and... The correlation coefficient of the corrected ice crystal nucleus formation rate is dynamically adjusted to modify the simulation parameters of the ice crystal nucleus formation rate, resulting in the adjusted distribution of the ice crystal nucleus formation rate. Combined with the spatiotemporal characteristics of convection intensity changes, the evolution path of the hail formation process is simulated, yielding dynamic simulation results. These results are used to analyze the temporal characteristics of hail from formation to descent. The moment when the hail particle size growth rate reaches its maximum value is recorded as the key turning point. The maturity moment is defined as when the hail particle size exceeds the critical size and the updraft can no longer support its weight. This is compared with the peak value of the convection intensity. The time difference between the current time and the hail maturity time is used to determine the potential hail landing time window. If the change in convection intensity during a certain period reaches a preset critical condition, then that period is marked as the potential landing time window, and the preliminary hail landing time range is determined. Based on the preliminary hail landing time range, real-time data on cloud environment changes are integrated, and the temperature gradient values ​​at various cloud heights at the current time are extracted and compared with the average temperature gradient values ​​at the same heights in historical hail cases of the same period. When the difference exceeds a set threshold, the boundary of the potential landing time window is dynamically corrected by combining time series characteristics and key turning points to obtain the corrected hail landing time prediction result.

[0004] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a method for predicting and assessing hail landing time based on multivariate data. By acquiring temperature distribution and vertical wind speed data at different cloud altitudes, analyzing temperature gradient changes, and calculating water vapor condensation rates, it combines updraft-driven data to simulate the ice crystal nucleus formation process, constructing a multi-level data fusion model to output a corrected ice crystal nucleus formation rate. This invention further analyzes the correlation between convection intensity changes and ice crystal nucleus formation, simulates the evolution path of hail formation, records key turning points, and determines the hail maturity time. It then dynamically corrects the potential landing time window using real-time cloud environment data, ultimately obtaining an accurate hail landing time prediction. This method, through multi-dimensional data analysis and dynamic simulation, achieves in-depth research on the hail formation mechanism and accurate prediction of landing time, providing important technical support for disaster prevention and mitigation. Attached Figure Description

[0005] Figure 1 This is a flowchart of a hailfall time prediction and evaluation method based on multivariate data according to the present invention. Detailed Implementation

[0006] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0007] like Figure 1 This embodiment of a hailfall time prediction and assessment method based on multivariate data may specifically include: S101. Obtain temperature distribution data and vertical wind speed data at different cloud altitudes. Record and preliminarily organize the temperature value and updraft speed at each altitude in real time to obtain a complete dataset of cloud temperature distribution data and updraft driving data. Analyze the temperature gradient changes between different altitudes and use a hierarchical calculation method to determine the water vapor condensation rate under the influence of the temperature gradient, and obtain the distribution characteristics of the water vapor condensation rate.

[0008] Temperature distribution data and vertical wind speed data at different cloud altitudes are acquired from multi-source meteorological observation equipment. Temperature values ​​and updraft speeds at each altitude are recorded in real time. Temperature data are stratified according to preset altitude intervals, and the ratio of the temperature difference between adjacent altitude layers to the altitude difference is calculated to obtain the temperature gradient values ​​between each layer. Based on the temperature gradient values ​​and the corresponding vertical wind speed data, actual water vapor pressure data for each altitude layer is obtained from meteorological observation equipment. The Clausius-Clapeyron equation is used to calculate the saturated water vapor pressure at that temperature, where the saturated water vapor pressure Es = 6.11 × 10^(7.5T / (237.3+T)), where T is the temperature in Celsius. The difference between the actual water vapor pressure and the saturated water vapor pressure is calculated. Combined with the influence coefficient of the temperature gradient on the saturated water vapor pressure, the water vapor supersaturation distribution data for each altitude layer is obtained. Based on the water vapor supersaturation distribution data, combined with the vertical wind speed distribution and the air density at each altitude layer, the product of vertical wind speed, air density, and water vapor mixing ratio per unit time is calculated to determine the water vapor flux through each altitude layer. The difference in water vapor flux between adjacent altitude layers is divided by the altitude difference and then multiplied by the supersaturation of the corresponding layer to determine the water vapor condensation rate value at each altitude layer, thus obtaining the distribution characteristics of water vapor condensation rate within the cloud layer.

[0009] Specifically, data acquisition by multi-source meteorological observation equipment involves the coordinated work of various observation methods, such as ground meteorological stations, weather balloons, and weather radars.

[0010] In one possible implementation, a ground-based weather station records near-surface temperature and wind speed data every 10 minutes, while a weather balloon continuously measures temperature and wind speed at different altitudes during its vertical ascent. Weather radar detects the vertical wind speed distribution within the cloud layer using the Doppler effect. This multi-source data fusion method can obtain more accurate and continuous information on the vertical structure of clouds. The stratified calculation of temperature gradients is crucial for understanding the thermal structure within clouds.

[0011] Specifically, after obtaining temperature data from the ground to an altitude of 8000 meters, the altitude is divided into 500-meter increments, and the temperature difference between adjacent layers is calculated. For example, if the temperature is 5°C at 2000 meters and 2°C at 2500 meters, the temperature gradient between these layers is -6°C / km. This negative temperature gradient indicates that the temperature decreases with increasing altitude, which is conducive to the cooling and condensation of water vapor. The Clausius-Clapeyron equation plays a crucial role in calculating saturated vapor pressure. This equation describes the exponential relationship between saturated vapor pressure and temperature; for every 10°C increase in temperature, the saturated vapor pressure approximately doubles.

[0012] In one embodiment, when the temperature at a certain altitude is 10°C, the saturated vapor pressure calculated by the equation is 12.27 hPa, while the actual measured vapor pressure is 13.5 hPa, resulting in a supersaturation of 1.23 hPa. This supersaturated state is a necessary condition for cloud droplet formation and growth. The calculation of vapor flux requires comprehensive consideration of three factors: vertical wind speed, air density, and vapor-water mixing ratio.

[0013] It should be noted that vertical wind speed determines the rate of water vapor transport, air density reflects the mass of air per unit volume, and the water vapor mixing ratio indicates the proportion of water vapor in the air. When the updraft speed is 2 m / s, the air density is 0.8 kg / m³, and the water vapor mixing ratio is 8 g / kg, the water vapor flux is the product of these three factors. By calculating the difference in water vapor flux between adjacent altitude layers, it is possible to determine how much water vapor undergoes a phase change at that layer. The vertical distribution characteristics of water vapor condensation rates reflect the key processes in precipitation formation within clouds.

[0014] Preferably, in the lower and middle parts of the cloud layer, due to higher temperatures and stronger updrafts, the water vapor condensation rate is usually higher; while in the upper part of the cloud layer, although the supersaturation may be higher, the condensation rate decreases due to the reduced water vapor content. This vertical distribution characteristic directly affects the evolution of cloud droplet spectra and the formation efficiency of precipitation particles, providing an important basis for accurately predicting precipitation intensity and duration.

[0015] S102. Based on the distribution characteristics of water vapor condensation rate and combined with temperature gradient data, simulate the dynamic changes in ice crystal nucleus formation rate. If the water vapor condensation rate at a certain altitude exceeds a preset threshold, mark that altitude as a critical region for ice crystal nucleus formation and obtain the distribution range of the critical region. If the water vapor condensation rate at a certain altitude does not exceed the preset threshold, mark that altitude as an inactive region and continue to monitor the changing trend of its water vapor condensation rate.

[0016] Based on the distribution characteristics of water vapor condensation rate and combined with temperature gradient data, the ice crystal nucleus formation rate at each altitude level is calculated according to the Bergieron process principle. The Bergieron process describes the physical process of ice crystal growth through water vapor transfer in a supercooled water droplet environment. The ice crystal nucleus formation rate is calculated by multiplying the water vapor condensation rate by the absolute value of the temperature gradient and then by the supercooling coefficient. Continuous calculations are performed for each altitude level to obtain dynamic data on the ice crystal nucleus formation rate over time and altitude. Based on this dynamic data, the ice crystal nucleus formation rate at each altitude level is compared with a preset threshold. If the ice crystal nucleus formation rate at a certain altitude exceeds the preset threshold, that altitude is marked as a critical region for ice crystal nucleus formation, and the location information of that altitude is recorded. All altitudes marked as critical regions are summarized to determine the continuous distribution range of the critical regions. For height layers outside the continuous distribution range of the key region, if the ice crystal nucleus formation rate at that height does not exceed a preset threshold, it is marked as an inactive region. The water vapor condensation rate in the inactive region is continuously monitored, the change in water vapor condensation rate at adjacent times is calculated, the trend of change is determined, and combined with the already determined distribution range of the key region, the complete regional distribution range is obtained.

[0017] Specifically, the Bergeron process is the core theory in cloud physics that explains ice crystal growth. This process is based on the difference in saturated water vapor pressure between ice and water surfaces.

[0018] In one possible implementation, when ice crystals and supercooled water droplets coexist in the cloud, water vapor evaporates from the surface of the supercooled water droplets and sublimates on the surface of the ice crystals because the saturated vapor pressure on the ice surface is lower than that on the water surface. The rate of this water vapor transfer process depends on the temperature conditions and the water vapor concentration gradient.

[0019] Specifically, calculating the ice crystal nucleation rate requires comprehensive consideration of multiple physical parameters. When the water vapor condensation rate at a certain altitude is 0.5 g / m³, the absolute temperature gradient is 8°C / km, and the supercooling coefficient is 0.2, the ice crystal nucleation rate is the product of these three factors. The supercooling coefficient reflects the degree to which the temperature at that altitude is below the freezing point; the lower the temperature, the larger the supercooling coefficient, and the more active the ice crystal nucleation. This calculation method can quantitatively describe the activity level of ice crystal formation at different altitudes. Acquiring dynamic data involves a continuous spatiotemporal monitoring process.

[0020] It should be noted that the ice crystal nucleus formation rate is not a constant value, but fluctuates with changes in atmospheric conditions. By collecting data at various altitudes every 5 minutes, a time series of ice crystal nucleus formation rates can be constructed. Vertically, a monitoring layer is set every 200 meters from the cloud base to the cloud top, forming a complete three-dimensional dynamic dataset. Threshold determination plays a crucial role in region identification.

[0021] For example, when the preset threshold is 1.0 g / m³, altitudes exceeding this value are considered to have significant ice crystal formation activity. Within the 4000-5500 m altitude range, if the ice crystal formation rate at multiple consecutive altitudes exceeds the threshold, this continuous interval is marked as a critical region. This method can identify the altitude range within clouds most conducive to precipitation formation. Monitoring inactive regions is equally important; these areas, while currently having low ice crystal formation rates, have the potential to transform into active regions.

[0022] In one embodiment, the development trend of a region can be determined by calculating the difference in water vapor condensation rates between adjacent time points. If the water vapor condensation rate of an inactive region shows an increasing trend for three consecutive time periods, and the growth rate exceeds 0.1 g / m³ per minute, then the region may transform into a critical region in the future. The complete regional distribution range is obtained by integrating information on critical and inactive regions.

[0023] Preferably, key areas are marked in red, inactive but growing areas are marked in yellow, and stable inactive areas are marked in blue. This classification method not only reflects the current ice crystal formation status but also predicts future trends, providing precise target area location for weather modification operations.

[0024] S103. Extract the updraft driving data corresponding to the distribution range of key areas at different altitudes, analyze the fluctuation amplitude and frequency of updraft velocity at different altitudes over time to obtain the spatiotemporal characteristics of convection intensity changes, determine the correlation between convection intensity changes and ice crystal nucleus formation rate, and output the corrected ice crystal nucleus formation rate based on the correlation.

[0025] By analyzing the distribution range of key regions, updraft-driven data at corresponding altitudes are extracted. The standard deviation of the updraft velocity at each altitude level over consecutive time points is calculated to obtain the fluctuation amplitude. The number of velocity direction changes per unit time is counted to obtain the frequency change value. The convection intensity value at each altitude level is determined by multiplying the fluctuation amplitude and the frequency change value, thus obtaining the distribution data of convection intensity values ​​changing with time and altitude. Based on this distribution data of convection intensity values ​​changing with time and altitude, the Pearson correlation coefficient method is used to calculate the point-by-point correlation between the updraft velocity time series and the corresponding ice crystal nucleus formation rate time series at each altitude level, obtaining the correlation coefficient value for each altitude level. Altitude levels where the absolute value of the correlation coefficient exceeds a preset threshold are identified, determining the correlation distribution between convection intensity changes and ice crystal nucleus formation rate. For this correlation distribution, a data fusion calculation process based on decision tree ensemble is constructed. Temperature gradient values, water vapor condensation rate values, and updraft velocity values ​​at each altitude level are used as input parameters. A weighted combination calculation is performed to obtain fusion feature values. These fusion feature values ​​are then corrected based on the correlation distribution, outputting the corrected ice crystal nucleus formation rate.

[0026] Specifically, the extraction of updraft-driven data involves the precise measurement of vertical wind speeds within key areas.

[0027] In one possible implementation, radial velocity data acquired by Doppler weather radar, combined with vertical wind speed measurements from wind profiler radar, can construct a complete three-dimensional wind field structure within a key area. Once a certain altitude layer is marked as a key area for ice crystal nucleation, the system automatically retrieves updraft data within a 500-meter range before and after that altitude layer, forming a continuous vertical wind speed profile. The calculation of convection intensity values ​​needs to comprehensively consider the temporal variation characteristics of wind speed.

[0028] Specifically, the fluctuation amplitude is quantified by calculating the standard deviation to assess the instability of the updraft. When the updraft velocity at a certain altitude is 2, 5, 3, 6, and 4 m / s over 10 minutes, the standard deviation is approximately 1.58 m / s, reflecting the intensity of the airflow fluctuation. The frequency variation is determined by counting the number of changes in velocity direction. If the updraft repeatedly transforms into a downdraft within a unit of time, it indicates intense convective activity in the region. The Pearson correlation coefficient plays a crucial role in analyzing the correlation between meteorological elements.

[0029] It should be noted that this method calculates the correlation coefficient, ranging from -1 to 1, by dividing the product of the covariance and standard deviation of the two variable sequences. Positive values ​​indicate a positive correlation, negative values ​​indicate a negative correlation, and the closer the absolute value is to 1, the stronger the correlation. During ice crystal formation, the updraft velocity and the ice crystal nucleus formation rate are usually positively correlated because strong updrafts can continuously transport water vapor and maintain supersaturation conditions. Determining the correlation distribution provides a weighting basis for subsequent data fusion.

[0030] For example, when the correlation coefficient at 4000 meters is 0.85, at 5000 meters it is 0.92, and at 6000 meters it is 0.78, it indicates that the updraft at 5000 meters has the most significant impact on ice crystal nucleation. This vertical distribution of correlations reflects the varying degrees of influence of atmospheric dynamic processes on cloud microphysical processes at different altitudes. The data fusion calculation process based on decision tree ensemble can effectively integrate multi-source meteorological data.

[0031] In one embodiment, the process first normalizes the temperature gradient, water vapor condensation rate, and updraft velocity to make data of different dimensions comparable. Then, weights are determined based on the correlation coefficients of each altitude layer, with higher weights assigned to altitude layers with stronger correlations. The fused eigenvalues ​​are obtained through weighted summation, reflecting the combined influence of multiple meteorological elements. The corrected ice crystal nucleus formation rate reflects the coupling effect between dynamic and microphysical processes.

[0032] Preferably, when the originally calculated ice crystal nucleus formation rate is 1.2 g / m³, and the correlation coefficient for that altitude layer is 0.9, it can be corrected to 1.35 g / m³ through fusion calculation. This correction takes into account the enhancing effect of updrafts, making the estimation of the ice crystal nucleus formation rate more accurate and providing a reliable basis for calculating the catalyst seeding amount for weather modification operations.

[0033] S104. Based on the correlation coefficient between convection intensity and the corrected ice crystal nucleus formation rate, the simulation parameters of the ice crystal nucleus formation rate are dynamically adjusted to obtain the adjusted ice crystal nucleus formation rate distribution. Combined with the spatiotemporal characteristics of convection intensity changes, the evolution path of the hail formation process is simulated to obtain the dynamic simulation results of the hail formation process.

[0034] Based on the correlation coefficient between convection intensity and the corrected ice crystal nucleus formation rate, a parameter adjustment factor is calculated using a linear interpolation method. The adjustment factor is equal to the ratio of the correlation coefficient to a preset baseline value. When the correlation coefficient is higher than a preset threshold, the adjustment factor is greater than 1; when the correlation coefficient is lower than the preset threshold, the adjustment factor is less than 1. By multiplying the original ice crystal nucleus formation rate by the adjustment factor, the adjusted ice crystal nucleus formation rate distribution data is obtained. Based on the adjusted ice crystal nucleus formation rate distribution data, combined with the spatiotemporal characteristics of convection intensity changes, the Lagrange particle tracking method is used to track the movement path of ice crystal particles in the vertical wind field. By calculating the water vapor sublimation growth and collision-coalescence growth of particles at each altitude layer, particles with diameters reaching the millimeter level are recorded as graupel, and those reaching the centimeter level are recorded as hail. The cumulative recording of particle size changes yields the evolution path of hail formation. Based on the evolution path of hail formation, the time and location when the particle diameter exceeds the preset hail size threshold are determined by combining particle size changes, vertical movement trajectory and number of cycles at each height layer. The height range and time series of hail formation are recorded to obtain the dynamic simulation results of the hail formation process.

[0035] Specifically, the application of linear interpolation methods in parameter tuning embodies the idea of ​​dynamic optimization.

[0036] In one possible implementation, when the correlation coefficient between convection intensity and ice crystal nucleation rate is 0.8, and the preset baseline value is 0.6, the adjustment factor is calculated as 0.8 divided by 0.6, resulting in 1.33. This means that under strong convection conditions, the ice crystal nucleation process is more active, requiring a corresponding increase in simulation parameters. Conversely, when the correlation coefficient drops to 0.4, the adjustment factor becomes 0.67, indicating a decrease in ice crystal formation efficiency under weak convection conditions. The Lagrange particle tracking method is an important tool for studying particle motion in clouds.

[0037] Specifically, this method treats each ice crystal as an independent point mass and calculates its trajectory based on the wind conditions at its location. When an ice crystal particle encounters an updraft of 2 m / s at an altitude of 5000 meters, it is carried upwards. During its ascent, the particle grows through water vapor sublimation and simultaneously collides and merges with other particles. This tracking method can accurately record the growth history of each particle. The phased changes in particle size reflect the physical processes of hail formation.

[0038] It should be noted that when ice crystals grow from the micrometer level to 0.5 millimeters in diameter, their falling velocity begins to increase, but they can still be supported by updrafts. As they continue to grow to 5 millimeters, the particles transform into graupel, their density increases, and they begin to move up and down within the cloud. When the diameter reaches 10 millimeters or more, true hail embryos form, their mass sufficient to overcome updrafts of moderate intensity and fall. Water vapor sublimation growth and collision-coalescence growth are the two main mechanisms of particle growth.

[0039] In one embodiment, at -10°C, supersaturated water vapor directly sublimates on the ice crystal surface, causing particles to grow slowly. As the particle size increases, the collision cross-section with the supercooled water droplets increases, capturing the droplets and rapidly freezing them on the surface, achieving rapid growth. This collision-coalescence growth efficiency is highest, particularly in the temperature range of -5°C to -15°C. The number of vertical cycles determines the final size of the hailstone.

[0040] For example, a grain of graupel is lifted to an altitude of 8,000 meters by a strong updraft, then falls to 3,000 meters due to gravity, only to be captured and lifted again by the updraft. With each such cycle, an ice shell forms on the particle's surface. After five cycles, a large hailstone with a diameter of 3 centimeters may form. The more cycles, the more pronounced the layered structure inside the hailstone. The spatiotemporal characteristics of the dynamic simulation results reveal key information about hailstone formation.

[0041] Preferably, simulations show that hail mainly forms at altitudes between 4,000 and 7,000 meters, with the entire incubation process lasting 20 to 40 minutes. Horizontally, hail formation areas are typically located at the leading edge of clouds with the strongest updrafts. These simulation results provide a scientific basis for determining the optimal timing and location for artificial hail suppression operations, enabling effective intervention in the early stages of hail formation.

[0042] S105. Analyze the temporal characteristics of hail from its formation to its fall by using dynamic simulation results of the hail formation process. Record the moment when the hail particle size growth rate reaches its maximum value as the key turning point. When the hail particle size exceeds the critical size and the updraft can no longer support its weight, it is determined to be the maturity moment. Compare the time difference between the peak moment of convection intensity and the hail maturity moment. If the change in convection intensity during a certain period reaches the preset critical condition, mark that period as the potential fall time window and determine the preliminary range of hail fall time.

[0043] By analyzing the dynamic simulation results of the hail formation process, a data sequence of hail particle size changes over time is extracted. The ratio of the difference in particle size between adjacent moments to the time interval is calculated to obtain the particle size growth rate. The growth rate sequence is then differentially calculated (the growth rate at the current moment minus the growth rate at the previous moment). When the calculation result changes from a positive value to a negative value, this moment is recorded as a critical turning point, indicating the end of the rapid growth phase of the hail. Based on this critical turning point, the relationship between hail particle size and updraft velocity is continuously monitored. By calculating the balance between air resistance and gravity on the hail, where air resistance is proportional to the square of the particle size and the square of the airflow velocity, and gravity is proportional to the cube of the particle size, the moment when the hail particle size increases to the point where gravity exceeds the resistance generated by the maximum updraft is determined as the hail maturity moment. The timestamp of the maturity moment and the corresponding hail particle size value are recorded. For the hail maturity time, data on the change of convection intensity over time are extracted from the dynamic simulation results. The peak time when the convection intensity reaches its maximum value is identified, and the time difference between the peak time and the maturity time is calculated. If the rate of decrease of convection intensity after the maturity time exceeds a preset critical condition, the period from the maturity time to the time when the convection intensity drops to half of its initial value is marked as the potential landing time window, and the preliminary hail landing time range is determined.

[0044] Specifically, time-series analysis of grain size growth rate revealed key stages in hail development.

[0045] In one possible implementation, by continuously recording hail particle size data, a typical pattern of slow initial growth, rapid mid-term growth, and eventual stabilization can be observed. When it takes only 3 minutes for the particle size to grow from 5 mm to 10 mm, but 8 minutes to grow from 15 mm to 20 mm, it indicates a slowing growth rate. This change reflects the mismatch between the depletion of available water vapor in the cloud and the rate of particle surface area growth. Differential computation plays a central role in identifying key inflection points.

[0046] Specifically, assuming a growth rate of 2 mm per minute at a certain moment, and decreasing to 1.5 mm per minute at the next moment, the difference is -0.5, marking the beginning of a decline in the growth rate. A series of negative differences indicates that the hailstones have passed their rapid growth phase. Identifying this turning point is crucial for predicting the final size of the hailstones and assessing their destructive potential. The balance between air resistance and gravity determines the vertical motion of the hailstones.

[0047] It should be noted that air resistance primarily depends on the windward area of ​​the hailstone and the relative velocity of the airflow; as the hailstone diameter increases, the windward area increases quadratically. The gravity of the hailstone, however, is proportional to its volume and increases cubically. When the hailstone diameter reaches 25 millimeters, even in a strong updraft of 10 meters per second, its gravity begins to exceed air resistance, causing the hailstone to fall. The criterion for determining maturity reflects the critical state of mechanical equilibrium.

[0048] For example, in a specific case, when hailstones reach a diameter of 30 mm and a density of approximately 0.8 g / cm³, their weight is about 11 grams. At this point, the strongest updraft velocity in the cloud is 15 m / s, generating a maximum drag of only 9 grams. This imbalance indicates that the hailstones have matured and can no longer remain suspended and grow within the cloud. The temporal evolution of convection intensity provides crucial information for predicting landing times.

[0049] In one embodiment, convection intensity typically peaks 20 minutes before hail maturity, with updraft speeds reaching up to 20 m / s at the peak. The convection intensity then gradually weakens, and most mature hailstones begin to fall when it drops to 10 m / s. This time span from peak to critical value is typically 15 to 25 minutes, constituting the potential fall window. The determination of the fall window takes into account several dynamic factors.

[0050] Preferably, by monitoring the rate of decrease in convection intensity, a decrease exceeding 3 meters per second every 5 minutes indicates that the convective system is rapidly decaying. Combining the hail maturity time and the rate of convection decay, it is possible to predict that hail will reach the ground within 10 to 20 minutes after maturity.

[0051] S106. Based on the preliminary hailfall time range, real-time data on cloud environment changes are integrated to extract the temperature gradient values ​​at various cloud heights at the current moment and calculate the difference between them and the average temperature gradient values ​​at the same height in historical hailfall cases. When the difference exceeds a set threshold, the boundary of the potential fall time window is dynamically corrected by combining time series characteristics and key turning points to obtain the corrected hailfall time prediction result.

[0052] Based on the preliminary hailfall time range, real-time data on cloud environment changes are acquired, and the temperature gradient values ​​at each altitude are extracted at the current moment. Hailfall records for the same season and geographical area are retrieved from the historical hailfall case database, and the average temperature gradient at the corresponding altitude in the historical cases is calculated. The current temperature gradient value is then compared with the historical average layer by layer to obtain the temperature gradient deviation data for each altitude level. For the temperature gradient deviation data, if the deviation value at any altitude level exceeds a set threshold, the time window correction calculation process is activated. The correction direction and magnitude are determined based on the sign and magnitude of the deviation value. A positive deviation indicates that the current environment is more conducive to rapid hailfall, while a negative deviation indicates a delayed descent. The time correction amount is calculated by multiplying the deviation value by a preset unit deviation corresponding to the time change. Based on the time correction amount, and combined with the start and end times of the preliminary fall time range, the start and end boundaries of the potential fall time window are adjusted respectively. The original boundary times are added to the corresponding time correction amount to determine the adjusted time window boundaries, obtaining the corrected hailfall time prediction result.

[0053] Specifically, the construction of a database of historical hail cases demonstrates the application value of meteorological big data.

[0054] In one possible implementation, the database records hail events in the same geographical area over the past 10 years. Each record includes key meteorological parameters such as temperature gradient, humidity distribution, wind speed, and wind direction at different altitudes during the hail event. Statistical analysis of this historical data yields typical temperature gradient values ​​for different seasons and weather systems. For example, during severe afternoon convective weather in summer, the average temperature gradient at 3000 meters is -8°C / km, while during a cold front in spring, this value is -6°C / km. The calculation of the temperature gradient deviation reveals the difference between the current atmospheric conditions and historically typical conditions.

[0055] Specifically, when the real-time temperature gradient at an altitude of 4000 meters is -10℃ / km, while the historical average for the same period is -7℃ / km, the deviation is -3℃ / km. This large negative deviation indicates that the current atmospheric stratification is more unstable, with a greater vertical temperature lapse rate, which is conducive to the development and maintenance of strong convection. This deviation analysis method can quickly identify abnormal weather conditions. The threshold triggering mechanism is designed based on statistical principles and practical experience.

[0056] It should be noted that the threshold is usually set using the standard deviation of historical data as a reference. When the temperature gradient deviation exceeds 1.5 times the standard deviation, it indicates that the current atmospheric conditions have deviated significantly from normal, and the prediction results need to be corrected.

[0057] For example, if the standard deviation of historical data is 2°C / km, then the threshold is set to 3°C / km. Exceeding this threshold indicates a significant change in the atmospheric environment, and the original prediction model may no longer be applicable. The calculation of the time correction reflects the combination of physical processes and statistical laws.

[0058] In one embodiment, the time variation corresponding to a unit temperature gradient deviation is derived through historical case regression analysis. When the temperature gradient deviation is positive 1°C / km, the hailfall time is advanced by an average of 2 minutes; when the deviation is negative 1°C / km, the fall time is delayed by an average of 3 minutes. This asymmetry reflects the different mechanisms by which changes in atmospheric stability affect the hailfall process. Dynamic adjustment of the time window boundaries enables adaptive optimization of the prediction results.

[0059] For example, if the initial predicted landing time window is 14:30 to 15:00, and multiple temperature gradient deviations at different altitudes are detected to be positive and exceed a threshold, the calculated time correction is 5 minutes earlier. Accordingly, the corrected landing time window is adjusted to 14:25 to 14:55. This dynamic correction mechanism continuously optimizes prediction accuracy based on real-time changes in atmospheric conditions. The corrected prediction combines the advantages of historical experience and real-time observations.

[0060] Preferably, by combining historical statistical patterns with current atmospheric conditions, the accuracy of hailfall time prediction can be significantly improved. Practice shows that using this correction method, the prediction time error can be reduced from ±15 minutes to within ±8 minutes, providing a more accurate time reference for hail suppression operations and disaster warnings.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A hail fall time prediction evaluation method based on multi-element data, characterized by, The method comprises: Obtain temperature distribution data and vertical wind speed data at different heights of the cloud layer, record temperature values and updraft speed at each height layer, calculate temperature gradient between adjacent height layers, determine water vapor condensation rate under the influence of temperature gradient, and obtain water vapor condensation rate distribution; According to the water vapor condensation rate distribution and the temperature gradient, calculate the ice crystal nucleus generation rate, mark the height where the water vapor condensation rate exceeds the preset threshold as the key area, and determine the distribution range of the key area; Extract the updraft speed at the corresponding height in the key area, calculate the fluctuation amplitude and frequency of the updraft speed, determine the convection intensity distribution, calculate the correlation coefficient of the updraft speed and the ice crystal nucleus generation rate, and output the corrected ice crystal nucleus generation rate according to the correlation coefficient; According to the corrected ice crystal nucleus generation rate and the convection intensity distribution, simulate the hail breeding evolution path, determine the key turning point when the hail particle size growth rate is maximum, mark the mature time when the hail particle size exceeds the critical size and the updraft cannot support, determine the time difference between the convection intensity peak value and the mature time, and mark the time window as the potential landing time window when the convection intensity change meets the preset condition; Differentially calculate the temperature gradient value of each height of the cloud layer at the current time and the average temperature gradient value of the same height in the historical same period hail case, adjust the boundary of the potential landing time window, and output the corrected hail landing time prediction.

2. The hail fall time prediction evaluation method based on multi-element data according to claim 1, characterized by, The method comprises: Obtain temperature distribution data and vertical wind speed data at different heights of the cloud layer, process the temperature data in layers according to the preset height interval, calculate the ratio of the temperature difference value and the height difference value between adjacent height layers, obtain the temperature gradient value of each height layer; according to the temperature gradient value and the vertical wind speed at the corresponding height, obtain the water vapor pressure data of each height layer, calculate the difference between the actual water vapor pressure and the saturated water vapor pressure, and determine the water vapor supersaturation degree of each height layer in combination with the temperature gradient; according to the water vapor supersaturation degree and the vertical wind speed, calculate the water vapor flux per unit time, determine the ratio of the water vapor flux difference value and the height difference value between adjacent height layers, multiply the supersaturation degree of the corresponding layer, and obtain the vertical distribution of the water vapor condensation rate of each height layer.

3. The hail fall time prediction evaluation method based on multi-element data according to claim 1, characterized by, The method comprises: According to the water vapor condensation rate distribution and the temperature gradient, calculate the ice crystal nucleus generation rate of each height layer, multiply the absolute value of the temperature gradient and the supercooling degree coefficient, obtain the distribution of the ice crystal nucleus generation rate with time and height; according to the ice crystal nucleus generation rate distribution, compare the ice crystal nucleus generation rate of each height layer with the preset threshold, mark the height where the ice crystal nucleus generation rate exceeds the preset threshold as the key area, and determine the distribution range of the key area by summarizing the heights of the key area.

4. The hail fall time prediction evaluation method based on multi-element data according to claim 1, characterized by, The ascending airflow velocity corresponding to the height in the key area is extracted, the fluctuation amplitude and frequency of the ascending airflow velocity are calculated, and the convection intensity distribution is determined, including: The ascending airflow velocity corresponding to the height in the key area is extracted, the standard deviation of the ascending airflow velocity at each height layer at consecutive time points is calculated to obtain the fluctuation amplitude, the number of changes in the velocity direction per unit time is counted to obtain the frequency value, and the convection intensity value of each height layer is determined according to the product of the fluctuation amplitude and the frequency value to obtain the distribution of the convection intensity with time and height.

5. The multi-nodal data based hail fall time prediction evaluation method of claim 1, wherein, The simulated hail generation and evolution path includes: According to the corrected ice crystal nucleus generation rate and the convection intensity distribution, the movement path of the ice crystal particles in the vertical wind field is tracked to obtain the evolution path of the hail generation.

6. The multi-nodal data-based hail fall time prediction evaluation method of claim 1, wherein, The maximum moment of the hail particle size growth rate is determined as the key turning point, and the mature moment when the hail particle size exceeds the critical size and the ascending airflow cannot support is marked, including: The data of the change of the hail particle size with time in the hail generation and evolution path is extracted, the ratio of the particle size difference value and the time interval at adjacent time points is calculated to obtain the particle size growth rate, and the moment when the growth rate changes from a positive value to a negative value is recorded as the key turning point; according to the key turning point, the resistance and gravity balance of the hail particle size and the ascending airflow velocity are calculated, and the moment when the gravity exceeds the resistance is marked as the mature moment.

7. The multi-nodal data based hail fall time prediction evaluation method of claim 1, wherein, The temperature gradient value of each height of the cloud layer at the current moment is calculated by difference with the average temperature gradient value of the same height in the historical hail case, and the boundary of the potential landing time window is adjusted, including: The real-time temperature gradient value of the cloud layer is obtained, the average temperature gradient value of the corresponding height in the historical hail case is retrieved, the difference between the real-time temperature gradient value and the average value is calculated to obtain the temperature gradient deviation, the product of the deviation value and the corresponding time change amount per unit deviation is calculated according to the temperature gradient deviation to obtain the time correction amount, and the starting and ending boundaries of the potential landing time window are adjusted according to the time correction amount to determine the corrected time window boundary.

Images