Method and system for identifying ice crystal morphology of precipitation-induced stratified mixed clouds in plateau areas

Abstract

The present invention discloses a method and system for identifying the morphology of ice crystals in mixed stratified clouds with precipitation in plateau areas, which belongs to the technical fields of visual recognition, meteorology and cloud physics, and solves the problem that the prior art is difficult to accurately identify the morphology of ice crystals in mixed stratified clouds with precipitation in plateaus, resulting in insufficient understanding of cloud microphysical processes and difficulty in effectively performing artificial weather modification operations. The present invention is based on obtaining multi-source observation data of mixed stratified clouds with precipitation in plateaus; identifying the morphology of ice crystals based on airborne particle detection data; analyzing the macroscopic structural characteristics of clouds based on X-band radar data, and matching them with aircraft detection trajectories to determine the altitude layer where the ice crystals are located; extracting the environmental conditions for the formation and growth of ice crystals based on sounding data, satellite data and reanalysis data; and establishing an ice crystal morphology recognition model for training to achieve automatic recognition and classification of ice crystal morphology. The present invention is used for a method for identifying the morphology of ice crystals in mixed stratified clouds with precipitation in plateau areas.

Description

Technical Field

[0001] A method and system for identifying ice crystal morphology of precipitation-accumulating mixed clouds in plateau areas are used for identifying ice crystal morphology of precipitation-accumulating mixed clouds in plateau areas, and belong to the technical fields of visual recognition, meteorology and cloud physics. Background Art

[0002] The shape of ice crystals in clouds and their growth mechanism are important contents in the study of cloud microphysical structural characteristics, and are also an important link in the formation of cold cloud precipitation. They play an important role in establishing or improving cloud microphysical schemes in numerical models. However, the formation and development process of ice crystals is very complicated, and often varies depending on the type of cloud and the position of ice crystals in the cloud, temperature, and humidity conditions. The microphysical structural characteristics of clouds are affected by many factors, including the type of cloud, the characteristics of the underlying surface, the development and changes of the corresponding weather system, etc. Cloud physics is the study of the physical and dynamic processes followed by the formation, evolution, structure and precipitation (such as rain, snow, hail, etc.) of clouds in the atmosphere on the basis of atmospheric thermodynamics and atmospheric dynamics. At present, the main means of research include radar microphysical parameter inversion, space-borne cloud satellites, numerical models, and aircraft cloud penetration detection. Aircraft cloud penetration detection is an important means of in-situ detection of cloud microphysical parameters, and its data is more reliable and intuitive for the study of the formation and evolution process of cloud particles and precipitation particles in the cloud. However, the existing technology has the following technical problems:

[0003] For plateaus with high altitudes and thin air, due to harsh natural conditions and scarce observation stations, existing technologies make it difficult to accurately identify the morphology of ice crystals in plateau precipitation stratified mixed clouds, resulting in insufficient understanding of cloud microphysical processes and difficulty in effectively carrying out artificial weather modification operations. Summary of the invention

[0004] In response to the above-mentioned research problems, the purpose of the present invention is to provide a method and system for identifying the morphology of ice crystals in precipitation-stratified mixed clouds in plateau areas, so as to solve the problem that the existing technology is difficult to accurately identify the morphology of ice crystals in precipitation-stratified mixed clouds in plateau areas, resulting in insufficient understanding of cloud microphysical processes and difficulty in effectively carrying out artificial weather modification operations.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] A method for identifying ice crystal morphology of precipitation stratified mixed clouds in plateau areas comprises the following steps:

[0007] Step 1: Obtain multi-source observation data of precipitation-stratus mixed clouds over the plateau, including airborne particle detection data, X-band radar data, sounding data, satellite data, and reanalysis data;

[0008] Step 2: Identify the morphology of ice crystals based on airborne particle detection data, including the spectral distribution characteristics of cloud and precipitation particles, and the liquid water content of cloud and precipitation particles;

[0009] Step 3: Analyze the macroscopic structural characteristics of the cloud based on the X-band radar data and match it with the aircraft detection trajectory to determine the altitude layer where the ice crystals are located;

[0010] Step 4: Extract the environmental conditions for ice crystal formation and growth based on sounding data, satellite data and reanalysis data;

[0011] Step 5: Based on the results obtained in steps 2 to 4, an ice crystal morphology recognition model is established and trained to realize automatic recognition and classification of ice crystal morphology.

[0012] Further, the specific steps of step 1 are:

[0013] Step 1.1, using the airborne particle detection system to obtain airborne particle detection data of the plateau, including two-dimensional images of clouds and precipitation particles on the plateau obtained by CIP and PIP probes, particle spectrum distribution data and cold water content data. The particle spectrum distribution data includes the average diameter of cloud particles, cloud particle number concentration, total cloud particle concentration and cloud particle sampling volume. The cold water content data includes the number, size distribution, volume and temperature of supercooled water droplets.

[0014] Step 1.2, obtaining X-band radar data of the plateau through an X-band radar system, including radar reflectivity factor, echo intensity and altitude;

[0015] Step 1.3, using the sounding system to obtain sounding data of the plateau, including water surface saturated vapor pressure, ice surface saturated vapor pressure, ice crystal surface temperature, water surface supersaturation, ice surface supersaturation, water vapor density difference when the water surface and ice surface are saturated, and vapor pressure;

[0016] Step 1.4: Use satellite systems to obtain satellite data of the plateau, including infrared satellite images and visible light satellite images;

[0017] Step 1.5: Obtain reanalysis data from a reanalysis data system, wherein the reanalysis data system is a reanalysis data service or database.

[0018] Furthermore, the calculation formulas for the average cloud particle diameter, cloud particle number concentration, total cloud particle concentration and cloud particle sampling volume in step 1.1 are:

[0019] Average diameter of cloud particles D m for:

[0020]

[0021] Where m is the total number of channels of the airborne particle detection system to measure the cloud particle spectrum, Di is the median diameter of the cloud particles in the ith channel, N(D i ) is the cloud particle number concentration in the i-th channel, and N is the total cloud particle concentration;

[0022] The cloud particle number concentration N(D i )for:

[0023] N(D i )=n(D i ) / V×ΔD i

[0024] Among them, n(D i ) is the number of cloud particles in the i-th channel, V is the sampling volume, ΔD i is the particle size interval of each channel;

[0025] The total cloud particle concentration N is:

[0026]

[0027] The sampling volume V is:

[0028] V=S E ×T AS ×T

[0029] Among them, S E is the effective sampling area of ​​the airborne particle detection system, T AS is the true airspeed of the aircraft, T is the sampling period of the airborne particle detection system;

[0030] The specific steps of step 1.2 are:

[0031] Step 1.21: Data acquisition and processing. The specific steps are as follows:

[0032] The X-band radar system reads volume scan data at 12 levels of elevation;

[0033] Calculate the offset angle of the starting azimuth of the 12-layer PPI relative to the true north based on the volume scan data;

[0034] Based on the volume scan data, the distances of each distance library in each radial direction of each layer of the 12-layer PPI mapped to the combined reflectivity coordinates are calculated, and the number of offset libraries is calculated;

[0035] Based on the offset angle and the offset library number, the maximum value of the combined reflectivity in the volume scan data corresponding to the radial and azimuth angles of each layer is obtained and stored in the form of an array as the combined reflectivity value;

[0036] Step 1.22: Based on the combined reflectivity value and the vertical profile algorithm, the height, radar reflectivity factor and echo intensity of each layer of elevation data of the vertical profile are obtained. The specific steps are as follows:

[0037] Define the longitude and latitude values ​​of any two points on the baseline, and convert the longitude and latitude coordinates of the two points into polar coordinates with the longitude and latitude of the radar site as the center;

[0038] According to the longitude and latitude values ​​of the two points, a linear equation is created. According to the linear equation, 200 corresponding longitude and latitude values ​​on the straight line are constructed, and the longitude and latitude values ​​are converted into polar coordinates with the radar as the center.

[0039] The distance and azimuth parameters from the library to the center of the radar are calculated by using the polar coordinate distance and angle parameters of the 200 points constructed, as well as the distance library with 12 layers of elevation angles. The distance coordinates from the library to the center of the radar are converted into longitude and latitude coordinates. Then, based on the combined reflectivity value, two-dimensional linear interpolation is performed on each point to obtain the basic reflectivity value of the profile on each layer of elevation baseline.

[0040] The obtained profile data on the baseline are further calculated to obtain vertical profile information, that is, the height, radar reflectivity factor and echo intensity of each layer of elevation angle data of the vertical profile information;

[0041] The calculation formulas for the saturated water vapor pressure on the water surface, the saturated water vapor pressure on the ice surface, the surface temperature of the ice crystal, the supersaturation of the water surface, the supersaturation of the ice surface, the water vapor density difference when the water surface and the ice surface are saturated, and the water vapor pressure in step 1.3 are:

[0042] Saturated vapor pressure at water surface s for:

[0043]

[0044] Among them, T air is the ambient temperature;

[0045] Ice surface saturated water vapor pressure e i for:

[0046]

[0047] Among them, T x The surface temperature of the ice crystal is:

[0048] T x =T air +ΔT

[0049] Where ΔT is the temperature difference between the ambient temperature and the ice crystal surface temperature;

[0050] Water surface supersaturation S W for:

[0051]

[0052] Where e is the water vapor pressure, when S WWhen >0, it means that the environment is in a state of oversaturation or saturation of the water surface. At this time, the water vapor condenses and grows into water droplets or is in a state of equilibrium. On the contrary, the environment is in a state of unsaturated water surface, and the water droplets will evaporate.

[0053] Ice surface supersaturation S i for:

[0054]

[0055] When S i When S is ≥0, it means that the environment is in an ice surface supersaturated or saturated state, at which time the water vapor condensation growth process occurs or the environment is in a balanced state. On the contrary, the environment is in an ice surface unsaturated state, and the ice crystal sublimation process occurs. i >S W When ≥0, it means that the environment is in a supersaturated state of ice surface and a saturated or supersaturated state of water surface. When the ice-water coexistence conditions are met, ice crystals grow through the Bergeron process.

[0056] The difference in water vapor density between the saturated water surface and the saturated ice surface is:

[0057]

[0058] Among them, ρ s (T air ) is the saturated water vapor density on the water surface, ρ i (T x ) is the saturated water vapor density of the ice surface, R v is the specific gas constant of water vapor, which is 461.5 J kg -1 K -1 .

[0059] The water vapor pressure e is:

[0060] e=e s (T air )×RH

[0061] Here, RH is the relative humidity, which is obtained by the onboard AIMMS-20 measurement system.

[0062] Further, the specific steps of step 2 are:

[0063] Step 2.1: Process the two-dimensional images obtained by the CIP and PIP probes to extract the characteristic information of clouds and precipitation particles; the specific steps are as follows:

[0064] Preprocess the two-dimensional images acquired by CIP and PIP probes, including noise removal and contrast enhancement;

[0065] Image processing technology is used to perform threshold segmentation, edge detection and morphological operations on the processed images to extract features, and machine learning algorithms are used to classify and identify the extracted features to obtain different types of clouds and precipitation particles. The feature information of clouds and precipitation particles is extracted through convolutional neural networks, where the extracted features include shape features, texture features and size features, and the machine learning algorithms include support vector machines and random forests;

[0066] Step 2.2, the characteristic information of cloud and precipitation particles is combined with the particle spectrum distribution data, and the spectrum distribution characteristics of cloud and precipitation particles are obtained by using statistical methods;

[0067] Step 2.3: Extract the features related to liquid water content based on the supercooled water content data, and use statistical methods to obtain the liquid water content features of cloud and precipitation particles, and obtain the particle number concentration that satisfies the CAS value greater than 10 cm -3 And the hot wire moisture meter measured value is greater than 0.1gm -3 The liquid water content of the liquid water content includes the concentration and volume distribution of cold water droplets.

[0068] According to the aforementioned method for identifying ice crystal morphology of precipitation-induced stratified mixed clouds in plateau areas, the specific steps of step 3 are:

[0069] Step 3.1: Extract the macroscopic structural features of clouds based on X-band radar data; the specific steps are as follows:

[0070] The density and size of the particles in the cloud obtain the radar reflectivity factor distribution at different altitudes, and the thickness and height distribution of the cloud are obtained;

[0071] Based on the fact that echo intensity is proportional to the radar reflectivity factor, the echo intensity distribution at different altitudes is obtained to obtain the intensity and distribution of precipitation in the cloud;

[0072] Based on the propagation time of radar signals at different distances from the center of the radar, the height distribution of clouds and precipitation is obtained, and the vertical structure characteristics of clouds, namely the macroscopic structure characteristics, are obtained by combining the results obtained by analyzing the radar reflectivity factor and the echo intensity;

[0073] Step 3.2: Match the aircraft detection trajectory with the macroscopic structural features to determine the altitude layer where the ice crystals are located.

[0074] Furthermore, the ice crystal morphology recognition model in step 5 includes an input layer, a convolutional neural network CNN, a recurrent neural network RNN, a fully connected layer and an output layer in sequence.

[0075] A system for identifying ice crystal morphology of precipitation-accumulated mixed clouds in plateau areas comprises a method for identifying ice crystal morphology of precipitation-accumulated mixed clouds in plateau areas.

[0076] Compared with the prior art, the present invention has the following beneficial effects:

[0077] The present invention integrates airborne particle detection data, X-band radar data, sounding data, satellite data and reanalysis data to comprehensively obtain the microphysical characteristics of plateau clouds. The airborne particle detection data is used to identify and classify the morphology of ice crystals through image processing and machine learning algorithms, and the spectral distribution characteristics and liquid water content characteristics of clouds and precipitation particles are extracted. Based on X-band radar data, the macroscopic structural characteristics of clouds are analyzed and matched with the aircraft detection trajectory to determine the altitude layer where the ice crystals are located. The sounding data, satellite data and reanalysis data are used to extract the environmental conditions for the formation and growth of ice crystals and then establish an ice crystal morphology recognition model. The automatic recognition and classification of ice crystal morphology is achieved through training. Combined with a variety of observation data on the plateau, more comprehensive cloud microphysical information is provided, while considering the environmental factors of ice crystal formation and growth, deepening the understanding of cloud microphysical processes, and using image processing and machine learning algorithms to improve the accuracy of ice crystal morphology recognition, which is conducive to the precise implementation of artificial weather modification operations, which is specifically embodied in the following aspects:

[0078] 1. Using airborne particle detection data, image processing and machine learning algorithms to identify and classify ice crystal morphology can remove noise, enhance contrast, and effectively classify and identify the extracted features, thereby more accurately obtaining the spectral distribution characteristics and liquid water content characteristics of cloud and precipitation particles, and improving the accuracy and efficiency of ice crystal morphology identification;

[0079] 2. Based on X-band radar data, the macroscopic structural characteristics of clouds are analyzed and matched with the aircraft detection track to determine the altitude layer where ice crystals are located, providing detailed information on the three-dimensional structure of clouds and the distribution of ice crystals, which is beneficial to the spatial distribution and growth environment of ice crystals in clouds, and provides an important reference for the study of cloud microphysical processes;

[0080] 3. Use sounding data, satellite data and reanalysis data to extract the environmental conditions for the formation and growth of ice crystals. By analyzing the environmental conditions, the physical process of ice crystal formation can be revealed, providing a scientific basis for the improvement of cloud microphysics models and artificial weather modification operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 It is a schematic diagram of the process flow of the framework of the present invention;

[0082] Figure 2 This is a schematic diagram of verifying the reliability of quality control of liquid water content values, total particle concentration values ​​of CIP and PIP, taking the aircraft detection data on September 26, 2021 as an example in the present invention;

[0083] Figure 3 It is the volume scan data of 12 layers of elevation angle of the radar in the present invention;

[0084] Figure 4 It is a schematic diagram of radar combination reflectivity in the present invention;

[0085] Figure 5 It is a combined reflectivity diagram of the rectangular coordinates converted into longitude and latitude coordinates in the present invention;

[0086] Figure 6 It is a vertical cross-sectional diagram of the radar combined reflectivity and the blue solid line in the present invention. DETAILED DESCRIPTION

[0087] The present invention will be further described below in conjunction with the accompanying drawings and specific implementation methods.

[0088] A method for identifying ice crystal morphology of precipitation stratified mixed clouds in plateau areas comprises the following steps:

[0089] Step 1: Obtain multi-source observation data of precipitation-containing stratocumulus mixed clouds over the plateau, including airborne particle detection data, X-band radar data, sounding data, satellite data, and reanalysis data; the specific steps are as follows:

[0090] Step 1.1, using the airborne particle detection system to obtain airborne particle detection data of the plateau, including two-dimensional images of clouds and precipitation particles on the plateau obtained by CIP and PIP probes, particle spectrum distribution data and cold water content data. The particle spectrum distribution data includes the average diameter of cloud particles, cloud particle number concentration, total cloud particle concentration and cloud particle sampling volume. The cold water content data includes the number, size distribution, volume and temperature of supercooled water droplets.

[0091] The DMT airborne particle measurement system (airborne particle detection system) includes multiple probes, such as the Cloud Imaging Probe (CIP), the Precipitation Imaging Probe (PIP), the Cloud and Aerosol Spectrometer (CAS), and the King Liquid Water Content Sensor (King-LWC). Compared with the earlier PMS series detection instruments, these probes have been greatly improved and can measure particles of different sizes more quickly and accurately. Detailed measurement elements, measurement range and resolution parameters are shown in the table below.

[0092] Table 2-1 Main performance parameters of airborne instruments

[0093]

[0094]

[0095] The calculation formulas for the average diameter of cloud particles, cloud particle number concentration, total cloud particle concentration and cloud particle sampling volume are:

[0096] Average diameter of cloud particles D m for:

[0097]

[0098] Where m is the total number of channels of the airborne particle detection system to measure the cloud particle spectrum, D i is the median diameter of the cloud particles in the ith channel, N(D i ) is the cloud particle number concentration in the i-th channel, and N is the total cloud particle concentration;

[0099] The cloud particle number concentration N(D i )for:

[0100] N(D i )=n(D i ) / V×ΔD i

[0101] Among them, n(D i ) is the number of cloud particles in the i-th channel, V is the sampling volume, ΔD i is the particle size interval of each channel;

[0102] The total cloud particle concentration N is:

[0103]

[0104] The sampling volume V is:

[0105] V=S E ×T AS ×T

[0106] Among them, S E is the effective sampling area of ​​the airborne particle detection system, T AS is the true airspeed of the aircraft, T is the sampling period of the airborne particle detection system;

[0107] Step 1.2: Obtain X-band radar data of the plateau through the X-band radar system, including radar reflectivity factor, echo intensity and altitude; the main parameters of the X-band radar are as follows:

[0108] Main parameters of X-band radar

[0109]

[0110] The specific steps are:

[0111] Step 1.21, data acquisition and processing. Combined reflectivity refers to the ratio of radar waves reflected by clouds at different altitudes within a certain range around the meteorological radar. It is a product that projects the maximum reflectivity factor found in a constant elevation azimuth scan onto a Cartesian grid in a volume scan. It can show the maximum basic reflectivity factor distribution of the entire detectable atmospheric space, and can provide accurate key information such as atmospheric cloud bodies and precipitation locations, which is conducive to a better understanding of cloud system structure. The X-band radar of the Zeku Station used in this case only contains 12 layers of volume scan data at elevation angles, and does not provide combined reflectivity data. Therefore, it is obtained in the following way. The specific steps are:

[0112] The X-band radar system reads volume scan data at 12 levels of elevation;

[0113] Calculate the offset angle of the starting azimuth of the 12-layer PPI relative to the true north based on the volume scan data;

[0114] Based on the volume scan data, the distances of each distance library in each radial direction of each layer of the 12-layer PPI mapped to the combined reflectivity coordinates are calculated, and the number of offset libraries is calculated;

[0115] Based on the offset angle and the offset library number, the maximum value of the combined reflectivity in the volume scan data corresponding to the radial and azimuth angles of each layer is obtained and stored in the form of an array as the combined reflectivity value;

[0116] Step 1.22, the vertical profile of radar reflectivity factor can reflect the different characteristics of precipitation clouds in the three-dimensional structure of weather radar reflectivity factor. According to the difference in reflectivity factor intensity, reflectivity factor gradient, vertical thickness and other factors, it can distinguish between layered clouds and convective stratified mixed clouds, and grasp the intensity characteristics of different heights in the cloud. The height, radar reflectivity factor and echo intensity of each layer of elevation angle data of the vertical profile can be obtained based on the combined reflectivity value and vertical profile algorithm. The specific steps are:

[0117] Define the longitude and latitude values ​​of any two points on the baseline, and convert the longitude and latitude coordinates of the two points into polar coordinates with the longitude and latitude of the radar site as the center;

[0118] According to the longitude and latitude values ​​of the two points, a linear equation is created. According to the linear equation, 200 corresponding longitude and latitude values ​​on the straight line are constructed, and the longitude and latitude values ​​are converted into polar coordinates with the radar as the center.

[0119] The distance and angle parameters of the polar coordinates of the 200 points and the distance library with 12 layers of elevation angles are used to calculate the distance from the library to the center of the radar and the corresponding azimuth parameters. The distance coordinates from the library to the center of the radar are converted into longitude and latitude coordinates. Then, based on the combined reflectivity value, two-dimensional linear interpolation is performed on each point to obtain the basic reflectivity value of the profile on each layer of elevation baseline. The formula for converting the distance coordinates from the library to the center of the radar into longitude and latitude coordinates is as follows:

[0120] long2=long1+d*sinα / [ARC*cos(lat1)*2π / 360]

[0121] lat2=lat1+d*cosα / (ARC*2π / 360)

[0122] Among them, ARC is the average radius of the earth, and its parameter ARC = 6371.393*1000; long1 and lat1 are the longitude and latitude values ​​of the radar center, d is the distance from each radial library of the radar to the radar center; α is the azimuth angle with 0° as the standard, and the radian system is used for calculation; long2 and lat2 represent the longitude and latitude values ​​of the radar center after conversion;

[0123] The obtained profile data on the baseline are further calculated to obtain vertical profile information, that is, the height, radar reflectivity factor and echo intensity of each layer of elevation angle data of the vertical profile information;

[0124] Step 1.3, use the sounding system to obtain the sounding data of the plateau, including the saturated water vapor pressure of the water surface, the saturated water vapor pressure of the ice surface, the surface temperature of the ice crystal, the supersaturation of the water surface, the supersaturation of the ice surface, the water vapor density difference between the water surface and the saturated ice surface, and the water vapor pressure; the calculation formulas for the saturated water vapor pressure of the water surface, the saturated water vapor pressure of the ice surface, the surface temperature of the ice crystal, the supersaturation of the water surface, the supersaturation of the ice surface, the water vapor density difference between the water surface and the saturated ice surface, and the water vapor pressure are:

[0125] Saturated vapor pressure at water surface s for:

[0126]

[0127] Among them, T air is the ambient temperature;

[0128] Ice surface saturated water vapor pressure e i for:

[0129]

[0130] Among them, T x The surface temperature of the ice crystal is:

[0131] T x =T air +ΔT

[0132] Where ΔT is the temperature difference between the ambient temperature and the ice crystal surface temperature;

[0133] Water surface supersaturation S w for:

[0134]

[0135] Where e is the water vapor pressure, when S w When >0, it means that the environment is in a state of oversaturation or saturation of the water surface. At this time, the water vapor condenses and grows into water droplets or is in a state of equilibrium. On the contrary, the environment is in a state of unsaturated water surface, and the water droplets will evaporate.

[0136] Ice surface supersaturation S i for:

[0137]

[0138] When S i When S is ≥0, it means that the environment is in an ice surface supersaturated or saturated state, at which time the water vapor condensation growth process occurs or the environment is in a balanced state. On the contrary, the environment is in an ice surface unsaturated state, and the ice crystal sublimation process occurs. i >S W When ≥0, it means that the environment is in a supersaturated state of ice surface and a saturated or supersaturated state of water surface. When the ice-water coexistence conditions are met, ice crystals grow through the Bergeron process.

[0139] The difference in water vapor density between the saturated water surface and the saturated ice surface is:

[0140]

[0141] Among them, ρ s (T air ) is the saturated water vapor density on the water surface, ρ i (T x ) is the saturated water vapor density of the ice surface, R v is the specific gas constant of water vapor, which is 461.5 J kg -1 K -1 .

[0142] The water vapor pressure e is:

[0143] e=e s (T air )×RH

[0144] Here, RH is the relative humidity, which is obtained by the onboard AIMMS-20 measurement system.

[0145] Step 1.4: Use satellite systems to obtain satellite data of the plateau, including infrared satellite images and visible light satellite images;

[0146] Step 1.5: Obtain reanalysis data from a reanalysis data system, wherein the reanalysis data system is a reanalysis data service or database.

[0147] Step 2: Identify the morphology of ice crystals based on airborne particle detection data, including the spectral distribution characteristics of cloud and precipitation particles, and the liquid water content of cloud and precipitation particles; the specific steps are:

[0148] Step 2.1: Process the two-dimensional images obtained by the CIP and PIP probes to extract the characteristic information of clouds and precipitation particles; the specific steps are as follows:

[0149] Preprocess the two-dimensional images acquired by CIP and PIP probes, including noise removal and contrast enhancement;

[0150] Image processing technology is used to perform threshold segmentation, edge detection and morphological operations on the processed images to extract features, and machine learning algorithms are used to classify and identify the extracted features to obtain different types of clouds and precipitation particles. The feature information of clouds and precipitation particles is extracted through convolutional neural networks, where the extracted features include shape features, texture features and size features, and the machine learning algorithms include support vector machines and random forests;

[0151] Step 2.2, the characteristic information of cloud and precipitation particles is combined with the particle spectrum distribution data, and the spectrum distribution characteristics of cloud and precipitation particles are obtained by using statistical methods;

[0152] Step 2.3: Extract the features related to liquid water content based on the supercooled water content data, and use statistical methods to obtain the liquid water content features of cloud and precipitation particles, and obtain the particle number concentration that satisfies the CAS value greater than 10 cm -3 And the hot wire moisture meter measured value is greater than 0.1gm -3 The liquid water content of the liquid water content includes the concentration and volume distribution of cold water droplets.

[0153] In this case, the liquid water content corrected by dry power is obtained, which has good accuracy. Taking the aircraft detection data on September 26, 2021 as an example, the liquid water content value, the total particle concentration value of CIP and PIP are used to verify the reliability of quality control. Figure 2 The following table shows the liquid water content values ​​obtained by CAS and hot wire water content meter. The horizontal axis is time and the vertical axis is the liquid water content value in gm. 3, (a) is the liquid water content value measured by CAS, (b) is the liquid water content value calculated from the hot-wire water content instrument data and corrected by dry power, (c) and (d) are the observation value of the hot-wire water content instrument and the original value of the liquid water content corresponding to the total power measured by the instrument, respectively. Comparing these four liquid water content values, CAS mainly measures small cloud droplets in the range of 0.54-50μm. Its measurement value will be affected by small particles, and the value is relatively small, which has a certain reference value. The three liquid water content values ​​obtained by the hot-wire water content instrument are compared with the liquid water content values ​​of cloud microphysics research in the Qilian Mountains of the Qinghai-Tibet Plateau, Nagqu area, and the eastern part of the Qinghai-Tibet Plateau and the Sanjiangyuan area. The peak values ​​do not exceed 0.5gm 3 The liquid water content value calculated after dry power correction is 0-0.2gm 3 The other two types of liquid water content values ​​obtained from the liquid water content are 0.8-1g m 3 , its value is too large, with a large error. Referring to the research value and processing method of liquid water content in the Qinghai-Tibet Plateau, the comparative study shows that the liquid water content value calculated by the hot wire water content meter and corrected by dry power has good accuracy. Therefore, the value after quality control is used as the standard in the analysis, supplemented by the liquid water content value measured by CAS.

[0154] Step 3: Analyze the macroscopic structural characteristics of the cloud based on X-band radar data and match it with the aircraft detection trajectory to determine the altitude layer where the ice crystals are located; the specific steps are as follows:

[0155] Step 3.1: Extract the macroscopic structural features of clouds based on X-band radar data; the specific steps are as follows:

[0156] The density and size of the particles in the cloud obtain the radar reflectivity factor distribution at different altitudes, and the thickness and height distribution of the cloud are obtained;

[0157] Based on the fact that echo intensity is proportional to the radar reflectivity factor, the echo intensity distribution at different altitudes is obtained to obtain the intensity and distribution of precipitation in the cloud;

[0158] Based on the propagation time of radar signals at different distances from the center of the radar, the height distribution of clouds and precipitation is obtained, and the vertical structure characteristics of clouds, namely the macroscopic structure characteristics, are obtained by combining the results obtained by analyzing the radar reflectivity factor and the echo intensity;

[0159] Step 3.2: Match the aircraft detection trajectory with the macroscopic structural features to determine the altitude layer where the ice crystals are located.

[0160] Take the X-band radar volume scan data at 9:46:41 on September 26, 2021 as an example. Figure 3It is the volume scan data of the radar at 12 elevation levels. As can be seen from the figure, the detection range is wider at low elevation levels, and more radar echo information is detected. Most of the radar echoes are distributed to the east of the radar center. There is a small convection area with an echo center intensity of 35dBZ, and the echo intensity is 0-40dBZ. The radar combined reflectivity algorithm is used to map the radar echo information of the 12-layer volume scan to a plane, and the combined reflectivity is used to reflect the information of the entire volume scan. Figure 4 This is the result obtained by the combined reflectivity algorithm. As can be seen from the figure, the data maps the 12-layer volume scan data onto a plane very well. Most of the echoes with larger intensity are distributed to the east of the radar center. Figure 3 The elevation angle diagram of the 10th layer basically realizes the integration of radar echo data, indicating that the designed radar combination reflectivity algorithm is suitable for this X-band radar and achieves good results.

[0161] Figure 5 It is a combined reflection map after converting the radar rectangular coordinates into longitude and latitude coordinates. The map is also superimposed with terrain height color map, maps, rivers and other elements. The horizontal coordinate is longitude and the vertical coordinate is latitude. From the comprehensive view of these elements, the conversion effect between rectangular coordinates and longitude and latitude coordinates is better, and basically a reasonable conversion can be achieved.

[0162] From the above research, it is found that the radar combined reflectivity algorithm and the radar echo profile algorithm of arbitrary baseline in this case can basically realize the corresponding functions and achieve good results. The conversion of rectangular coordinates and longitude and latitude coordinates also greatly increases the practicality of the data, and can be comprehensively analyzed and processed with regional terrain and other forms, such as Figure 6 As shown, a is the radar combined reflectivity (the blue line is the radar echo profile position), and b is the vertical profile.

[0163] Step 4: Extract the environmental conditions for ice crystal formation and growth based on sounding data, satellite data and reanalysis data;

[0164] Step 5: Based on the results obtained in steps 2 to 4, an ice crystal morphology recognition model is established and trained to realize automatic recognition and classification of ice crystal morphology.

[0165] The ice crystal morphology recognition model includes an input layer, a convolutional neural network (CNN), a recurrent neural network (RNN), a fully connected layer, and an output layer.

[0166] The above are only representative embodiments of the present invention in many specific application scopes, and do not constitute any limitation on the protection scope of the present invention. Any technical solutions formed by transformation or equivalent replacement fall within the protection scope of the present invention.

Claims

1. A method for identifying ice crystal morphology of precipitation-induced stratified mixed clouds in plateau areas, characterized in that: The steps include: Step 1: Obtain multi-source observation data of precipitation-stratus mixed clouds over the plateau, including airborne particle detection data, X-band radar data, sounding data, satellite data, and reanalysis data; Step 2: Identify the morphology of ice crystals based on airborne particle detection data, including the spectral distribution characteristics of cloud and precipitation particles, and the liquid water content of cloud and precipitation particles; Step 3: Analyze the macroscopic structural characteristics of the cloud based on the X-band radar data and match it with the aircraft detection trajectory to determine the altitude layer where the ice crystals are located; Step 4: Extract the environmental conditions for ice crystal formation and growth based on sounding data, satellite data and reanalysis data; Step 5: Based on the results obtained in Steps 2 to 4, an ice crystal morphology recognition model is established and trained to realize automatic recognition and classification of ice crystal morphology; Step 1.1, using the airborne particle detection system to obtain airborne particle detection data of the plateau, including two-dimensional images of clouds and precipitation particles on the plateau obtained by CIP and PIP probes, particle spectrum distribution data and cold water content data. The particle spectrum distribution data includes the average diameter of cloud particles, cloud particle number concentration, total cloud particle concentration and cloud particle sampling volume. The cold water content data includes the number, size distribution, volume and temperature of supercooled water droplets. Step 1.2, obtaining X-band radar data of the plateau through an X-band radar system, including radar reflectivity factor, echo intensity and altitude; Step 1.3, using the sounding system to obtain sounding data of the plateau, including water surface saturated vapor pressure, ice surface saturated vapor pressure, ice crystal surface temperature, water surface supersaturation, ice surface supersaturation, water vapor density difference when the water surface and ice surface are saturated, and vapor pressure; Step 1.4: Use satellite systems to obtain satellite data of the plateau, including infrared satellite images and visible light satellite images; Step 1.5, obtaining reanalysis data from a reanalysis data system, wherein the reanalysis data system is a reanalysis data service or database; The specific steps of step 2 are: Step 2.1: Process the two-dimensional images obtained by the CIP and PIP probes to extract the characteristic information of clouds and precipitation particles; the specific steps are as follows: Preprocess the two-dimensional images acquired by CIP and PIP probes, including noise removal and contrast enhancement; Image processing technology is used to perform threshold segmentation, edge detection and morphological operations on the processed images to extract features, and machine learning algorithms are used to classify and identify the extracted features to obtain different types of clouds and precipitation particles. The feature information of clouds and precipitation particles is extracted through convolutional neural networks, where the extracted features include shape features, texture features and size features, and the machine learning algorithms include support vector machines and random forests; Step 2.2, the characteristic information of cloud and precipitation particles is combined with the particle spectrum distribution data, and the spectrum distribution characteristics of cloud and precipitation particles are obtained by using statistical methods; Step 2.3: Extract the features related to liquid water content based on the supercooled water content data, and use statistical methods to obtain the liquid water content features of cloud and precipitation particles, and obtain the particle number concentration that satisfies the CAS value greater than 10 cm -3 And the hot wire moisture meter measured value is greater than 0.1 gm -3 liquid water content, wherein characteristics related to the liquid water content include a concentration and volume distribution of cold water droplets; The specific steps of step 3 are: Step 3.1: Extract the macroscopic structural features of clouds based on X-band radar data; the specific steps are as follows: The density and size of the particles in the cloud obtain the radar reflectivity factor distribution at different altitudes, and the thickness and height distribution of the cloud are obtained; Based on the fact that echo intensity is proportional to the radar reflectivity factor, the echo intensity distribution at different altitudes is obtained to obtain the intensity and distribution of precipitation in the cloud; Based on the propagation time of radar signals at different distances from the center of the radar, the height distribution of clouds and precipitation is obtained, and the vertical structure characteristics of clouds, namely the macroscopic structure characteristics, are obtained by combining the results obtained by analyzing the radar reflectivity factor and the echo intensity; Step 3.2: Match the aircraft detection trajectory with the macroscopic structural features to determine the altitude layer where the ice crystals are located.

2. The method for identifying ice crystal morphology of precipitation-induced stratified mixed clouds in plateau areas according to claim 1, characterized in that: The calculation formulas for the average cloud particle diameter, cloud particle number concentration, total cloud particle concentration and cloud particle sampling volume in step 1.1 are: Average diameter of cloud particles for: in, The total number of channels for measuring cloud particle spectra for the airborne particle detection system, For the The median diameter of the channel cloud particles, For the Channel cloud particle number concentration, is the total concentration of cloud particles; The cloud particle number concentration in the i-th channel for: in, For the Channel cloud particle number, is the sampling volume, is the particle size interval of each channel; Total cloud particle concentration for: Sampling volume for: in, is the effective sampling area of ​​the airborne particle detection system, is the true airspeed of the aircraft, T is the sampling period of the airborne particle detection system; The specific steps of step 1.2 are: Step 1.21: Data acquisition and processing. The specific steps are as follows: The X-band radar system reads volume scan data at 12 levels of elevation; Calculate the offset angle of the starting azimuth of the 12-layer PPI relative to the true north based on the volume scan data; Based on the volume scan data, the distances of each distance library in each radial direction of each layer of the 12-layer PPI mapped to the combined reflectivity coordinates are calculated, and the number of offset libraries is calculated; Based on the offset angle and the offset library number, the maximum value of the combined reflectivity in the volume scan data corresponding to the radial and azimuth angles of each layer is obtained and stored in the form of an array as the combined reflectivity value; Step 1.22: Based on the combined reflectivity value and the vertical profile algorithm, the height, radar reflectivity factor and echo intensity of each layer of elevation data of the vertical profile are obtained. The specific steps are as follows: Define the longitude and latitude values ​​of any two points on the baseline, and convert the longitude and latitude coordinates of the two points into polar coordinates with the longitude and latitude of the radar site as the center; According to the longitude and latitude values ​​of the two points, a linear equation is created. According to the linear equation, 200 corresponding longitude and latitude values ​​on the straight line are constructed, and the longitude and latitude values ​​are converted into polar coordinates with the radar as the center. The distance and azimuth parameters from the library to the center of the radar are calculated by using the polar coordinate distance and angle parameters of the 200 points constructed, as well as the distance library with 12 layers of elevation angles. The distance coordinates from the library to the center of the radar are converted into longitude and latitude coordinates. Then, based on the combined reflectivity value, two-dimensional linear interpolation is performed on each point to obtain the basic reflectivity value of the profile on each layer of elevation baseline. The obtained profile data on the baseline are further calculated to obtain vertical profile information, that is, the height, radar reflectivity factor and echo intensity of each layer of elevation angle data of the vertical profile information; The calculation formulas for the saturated water vapor pressure on the water surface, the saturated water vapor pressure on the ice surface, the surface temperature of the ice crystal, the supersaturation of the water surface, the supersaturation of the ice surface, the water vapor density difference when the water surface and the ice surface are saturated, and the water vapor pressure in step 1.3 are: Saturated vapor pressure on water surface for: in, is the ambient temperature; Ice surface saturation water pressure for: in, The surface temperature of the ice crystal is: in, is the temperature difference between the ambient temperature and the surface temperature of the ice crystal; Water surface supersaturation for: in, is the water vapor pressure, when When >0, it means that the environment is in a state of oversaturation or saturation of the water surface. At this time, the water vapor condenses and grows into water droplets or is in a state of equilibrium. On the contrary, the environment is in a state of unsaturated water surface, and the water droplets will evaporate. Ice surface supersaturation for: when When , it means that the environment is in an ice surface supersaturated or saturated state, at which time the water vapor condensation growth process occurs or the environment is in a balanced state. On the contrary, the environment is in an ice surface unsaturated state, and the ice crystal sublimation process occurs; when When , it means that the environment is in a supersaturated state of ice surface and a saturated or supersaturated state of water surface. When the ice-water coexistence conditions are met, ice crystals grow through the Bergeron process. The difference in water vapor density between the saturated water surface and the ice surface is: in, is the saturated water vapor density on the water surface, is the saturated water vapor density of the ice surface, is the specific gas constant of water vapor, which is 461.5 J kg -1 K -1 ; Vapor Pressure for: in, is the relative humidity, obtained by the onboard AIMMS-20 measurement system.

3. The method for identifying ice crystal morphology of precipitation-induced stratified mixed clouds in plateau areas according to claim 1, characterized in that: The ice crystal morphology recognition model in step 5 includes an input layer, a convolutional neural network CNN, a recurrent neural network RNN, a fully connected layer and an output layer in sequence.

4. A system for identifying ice crystal morphology of precipitation-induced stratified mixed clouds in plateau areas, characterized by: A method for identifying ice crystal morphology of precipitation-induced stratified mixed clouds in plateau areas as described in any one of claims 1-3.

Images