A method for classifying water-soluble particles based on S-band dual-polarization radar

Abstract

This invention belongs to the technical fields of radar quantitative precipitation estimation, numerical weather prediction model data assimilation, severe convective weather early warning, and microphysical characteristic analysis of precipitation weather systems. Specifically, it relates to a method for identifying water condensate particle classification (HCA) based on S-band dual-polarization radar. By adding appropriate discrimination conditions, it solves the problem of HCA incorrectly identifying water condensate particle types in hail and three-body scattering regions. It utilizes the temperature analysis field of the half-hourly update and prediction system (HHUPS) short-term model from the Shandong Meteorological Research Institute to determine the melt layer height, thus solving the problem of HCA's inability to identify the melt layer under severe convective conditions. Based on the relationships between eight types of water condensate, it limits the vertical distribution of water condensate identified by HCA, solving the problem of unreasonable vertical distribution of water condensate particles identified by HCA.

Description

Technical Field

[0001] This invention belongs to the technical fields of radar quantitative estimation of precipitation, numerical weather prediction model data assimilation, severe convective weather early warning, and microphysical characteristic analysis of precipitation weather systems. Specifically, it relates to a method for classifying water condensate particles based on S-band dual polarization radar. Background Technology

[0002] A major operational breakthrough brought by dual-polarization radar is the ability to classify hydrometeors in precipitation systems with high spatiotemporal resolution. This is mainly due to the increased sensitivity of the newly added dual-polarization observations to information such as particle shape, size, and orientation of different hydrometeors. Fuzzy logic has gradually become the mainstream method for hydrometeor classification. Its principle is to generalize radar parameters using membership functions of different hydrometeors, eliminating the influence of overlapping radar parameter distributions for different hydrometeors, and thus more accurately identifying various types of hydrometeors. Park et al. (2009), based on previous research, added a confidence factor to the membership function and added discrimination conditions for stratiform clouds and convective clouds, establishing a hydrometeor classification algorithm (HCA).

[0003]

[0004] (1) is the calculation formula for HCA, where i represents the type of hydrogel. HCA can identify eight types of hydrogel: dry snow (DS), wet snow (WS), ice crystals (CR), graupel (GR), large drops (BD), moderate (light) rain (RA), heavy rain (HR), and rain-hail mixture (RH). It also identifies ground features or superrefractive ground features (GC / AP) and organisms (BS). A represents the integrated probability value. For each radar detection range library, HCA calculates the integrated probability value of all types based on the membership function and selects the type corresponding to the maximum value as the identification result. V represents the radar parameter, and j represents the parameter type, including the horizontal reflectivity factor (Z). H ), differential reflectance (Z) DR ), zero-lag correlation coefficient (CC), and logarithmic form of differential phase shift (LK) DP ), Z H Texture (SDZ) H ) and differential phase-shifted texture (SDΦ) DPP is the membership function, and an asymmetric trapezoid is selected to determine the probability distribution boundary of the parameters. W represents the weighting factor, which takes values ​​between 0 and 1. The higher the value, the greater the effect of identifying water condensate. Q represents the confidence factor, which eliminates the effects of radar calibration, attenuation, non-uniformity filling, beam blocking, observation error and noise (Park HS, Ryzhkov AV, Zrnic DS, et al, 2009. The Hydrometeor Classification Algorithm for the Polarimetric WSR-88D: Description and Application to an MCS[J]. Weather and Forecasting, 24(3): 730-748).

[0005] HCA integrates the Melting Layer Automatic Designation (MLDA) method designed by Giangrande et al., 2008, which divides the entire atmosphere into five altitude layers based on the location of the melting layer and limits the types of condensate that cannot appear in each layer; HCA adds the identification of stratiform clouds and convective clouds, and specifies the types of condensate that can appear in the two types of clouds; HCA also includes single radar parameter threshold discrimination conditions (Giangrande SE, Krause JM, Ryzhkov AV, 2008. Automatic Designation of the Melting Layer with a Polarimetric Prototype of the WSR-88D Radar[J]. JAppl Meteor Climatol, 47(5): 1354-1364).

[0006] However, the above methods have the following technical problems: (1) Since hail areas are usually accompanied by heavy precipitation, solid and liquid water condensates coexist, and CC is usually less than 0.9, HCA will identify the water condensates in the hail area as ground objects. The three-body scattering (TBSS) region is the false echo region that appears behind the radar antenna after the radar beam is scattered three times by large particles-ground-large particles. A large number of observations have found that when hail is accompanied by precipitation, precipitation particles may appear in the TBSS region, which are usually also incorrectly identified as ground objects by HCA. (2) HCA uses MLDA to identify the location of the melt layer, thereby distinguishing the water condensate types with more overlapping membership functions, such as dry snow and moderate (light) rain. The advantage of MLDA is that it can obtain the location of the melt layer with high spatiotemporal resolution. The disadvantage is that when the melt layer characteristics are not obvious, such as local severe convective weather, MLDA cannot identify the melt layer, which requires other data to assist in the judgment. (3) The vertical distribution of hydrogels identified by HCA is sometimes unreasonable, such as liquid particles (heavy rain) above solid particles (hail). Summary of the Invention

[0007] To address the issues of HCA's inaccurate identification of hydrophobic particle types in hail and TBSS regions, the inability of MLDA to identify the melt layer under severe convective weather, and the unreasonable vertical distribution of identified hydrophobic particles, this invention provides a method for classifying hydrophobic particles using S-band dual-polarization radar. This method incorporates appropriate discrimination conditions to identify hydrophobic particle types in hail and three-body scattering regions. It utilizes the temperature analysis field of the Half-Hourly Update and Prediction System (HHUPS) short-term mode to determine the melt layer height, thus resolving the HCA's inability to identify the melt layer under severe convective conditions. Furthermore, it limits the vertical distribution of identified hydrophobic particles based on the relationships between eight hydrophobic types, addressing the problem of unreasonable vertical distribution of hydrophobic particles.

[0008] The solution of the present invention is as follows:

[0009] A method for classifying water condensate particles based on S-band dual polarization radar is provided. The water condensate particle classification adopts the HCA-Opt method. By adding the identification of hail and three-body scattering regions in the HCA method using Equations 2 and 3, the water condensate identified by HCA in the two regions is corrected. The HCA method is shown in Equation 1.

[0010]

[0011] In Equation 1:

[0012] i represents the type of hydrogel. HCA can identify 8 types of hydrogel, namely dry snow (DS), wet snow (WS), ice crystals (CR), graupel (GR), large droplets (BD), light to moderate rain (RA), heavy rain (HR), and rain-hail mixture (RH).

[0013] A represents the integrated probability value; V represents the radar parameter; j represents the parameter type, including the horizontal reflectivity factor Z. H Differential reflectivity Z DR Zero-lag correlation coefficient CC, and logarithmic form of differential phase shift LK DP Z H Texture SDZ H With differential phase-shift texture SDΦ DP ;

[0014] P represents the membership function, and an asymmetric trapezoid is selected to determine the probability distribution boundary of the parameters;

[0015] W represents the weighting factor, which ranges from 0 to 1. The higher the value, the greater the effect of identifying water-coagulated substances.

[0016] Q represents the confidence factor, which eliminates the effects of radar calibration, attenuation, non-uniformity filling, beam blocking, observation errors, and noise.

[0017]

[0018] In Equation 2:

[0019] ETOP 18dBZ Z represents H The echo height is 18 dBZ;

[0020] In Equation 3:

[0021] ETOP 0dBZ Z represents H The echo height is 0 dBZ;

[0022] r stormcore The strong echo region indicates the radial direction Z. H Areas with a cumulative distance greater than 1 km and a range greater than 45 dBZ;

[0023] R is the radial distance from the strong echo zone to the radar station;

[0024] Equation 2 indicates that CC is less than 0.9, and Z H When the echo top height of 45dBZ or 18dBZ is greater than 8 km, it is considered a hail zone.

[0025] Equation 3 indicates that the area where TBSS may occur is determined behind the center of the strong echo. Then, the water condensate type in the TBSS area is determined by using the CC threshold and the 0dBZ echo top height. If the conditions of Equation 3 are met, it is a precipitation type; otherwise, it is a non-precipitation type.

[0026] When a certain distance library parameter satisfies Equation 2 or Equation 3, HCA-Opt sets the integration probability of land features and organisms to 0.

[0027] Furthermore, it also includes using the HHUPS short-term mode temperature analysis field to determine the height of the melt layer.

[0028] Preferably, the temperature analysis field of the HHUPS short-term model is constructed based on weather research and forecasting models, using 0.25° resolution data from the US National Environmental Prediction Center's global forecast model as background field data, and a data assimilation and gridded statistical interpolation module as the model assimilation system; two cold starts are performed daily at 08:00 and 20:00, and other times are hot starts. The forecast field one hour after integration of the previous time period is used as the background field for the next time period, and the observation data is assimilated every half hour; the assimilated data includes data from national-level ground meteorological observation stations, automatic ground meteorological observation stations, conventional radiosonde stations, L-band radar radiosonde stations and aircraft reports, radar reflectivity and Himawari-8 satellite data; 0-24 hour forecasts are produced hourly every day, with the forecast field at 0 time being the analysis field at the start time of the forecast.

[0029] Preferably, the 0°C temperature height is extracted from the temperature analysis field of the short-term model as the melting layer top and the 2°C wet-bulb temperature as the melting layer bottom. The latitude and longitude corresponding to the center of each distance library detected by radar are calculated in sequence. The data of the four closest model grid points are selected, and the heights corresponding to the melting layer top and bottom of the model are interpolated to the center coordinates of the calculated distance library using the bilinear method.

[0030] Furthermore, it also includes limiting the vertical distribution of the identified hydrogels based on vertical distribution constraints of hydrogel types, as shown in the table below:

[0031]

[0032] In the table, √ and × indicate whether the corresponding water condensate in the column can or cannot appear below the corresponding water condensate in the row. When a distance reservoir is identified as dry snow (DS), see row 1 in the table, any type of water condensate can appear below it. When a distance reservoir is identified as wet snow (WS), see row 2 in the table, dry snow (DS) and ice crystals (CR) cannot appear below it.

[0033] Preferably, since low elevation angles are easily contaminated by non-meteorological echoes, the water condensate type identified at the highest elevation angle is taken as the true value, and corrections are made sequentially from high elevation angle to low elevation angle according to the table's defined conditions; when the water condensate type identified at the lower elevation angle does not meet the conditions, the water condensate that meets the conditions is checked based on the classification results of the upper elevation angle, and the water condensate corresponding to the maximum integration probability value is selected as the corrected classification result.

[0034] Preferably, the HCA-Opt method is implemented using the Fortran language.

[0035] Another object of the present invention is to protect the application of the above method in the identification and classification of hydrogel particles.

[0036] Beneficial effects of the present invention

[0037] 1. Solved the problem of HCA's incorrect identification of water condensate type in the hail and three-body scattering regions.

[0038] By adding appropriate discrimination conditions, the reflectance threshold (Z) is used. H Determine whether it is a hail zone based on echo heights greater than 45dBZ and greater than or equal to 18dBZ; according to the reflectivity threshold (Z... H The location of the strong echo zone was determined by the distance (>45 dBZ) and length (cumulative distance greater than 1 km). The particle type behind the three-body scattering region was determined by the echo top height (>0 dBZ) behind the strong echo zone. Finally, the identification of water condensate particle types in the hail zone and the three-body scattering region was corrected.

[0039] 2. The height of the melt layer was determined by using the temperature analysis field of the HHUPS short-term mode, which solved the problem that HCA could not identify the melt layer under strong convection conditions.

[0040] This model is based on the Weather Research and Forecast (WRF) model, using 0.25° resolution data from the National Center for Environmental Prediction (NCEP) Global Forecast System (GFS) as background field data, and employing data assimilation (DA) and gridpoint statistical interpolation. The Interpolation (GSI) module, acting as the model assimilation system, performs two cold starts daily at 08:00 and 20:00, with other times being warm starts. The forecast field one hour after integration of the previous time period serves as the background field for the next time period. Observational data is assimilated every half hour, and 0-24 hour forecasts are produced hourly each day, with the forecast field at time 0 serving as the analysis field at the start of the forecast. The 0°C height is extracted from the temperature analysis field of the short-term model as the melting layer top and the 2°C wet-bulb temperature as the melting layer bottom. The latitude and longitude corresponding to the center of each range library detected by radar are calculated sequentially. The data of the four closest model grid points are selected, and the bilinear method is used to interpolate the corresponding heights of the melting layer top and bottom of the model to the center coordinates of the calculated range library.

[0041] 3. By combining the vertical distribution constraints of water-soluble material types, the vertical distribution of the identified water-soluble material is limited, thus solving the problem of unreasonable vertical distribution of water-soluble material particles identified by HCA.

[0042] The specific restrictions on the vertical distribution of the eight types of hydrogel particles are shown in the table above. Since lower elevation angles are easily contaminated by non-meteorological echoes, the hydrogel type identified at the highest elevation angle is taken as the true value, and corrections are made sequentially from high to low elevation angles according to the table's restrictions. If the hydrogel type identified at a lower elevation angle does not meet the conditions, the hydrogel that meets the conditions is checked against the classification results at the upper elevation angle, and the hydrogel corresponding to the highest integration probability value is selected as the corrected classification result.

[0043] 4. Using the Fortran language to implement the HCA-Opt method, it takes approximately 7 seconds for a single radar to complete one calculation in a single thread. Attached Figure Description

[0044] Figure 1 The Z-axis elevation at 0.5° on the Qingdao radar at 15:30 on August 16, 2019. H Hydrogel classification results (solid white ellipse lines represent hail areas, and dashed black ellipse lines represent three-body scattering areas);

[0045] Figure 2 The Z-axis elevation at 0.5° on the Qingdao radar at 15:30 on August 16, 2019. DRHydrogel classification results (solid white ellipse lines represent hail areas, and dashed black ellipse lines represent three-body scattering areas);

[0046] Figure 3 The results of CC hydrogel classification at 15:30 on August 16, 2019, at an elevation angle of 0.5° using the Qingdao radar (solid white ellipse represents the hail area, and dashed black ellipse represents the three-body scattering area).

[0047] Figure 4 The HCA hydrogel classification results from the Qingdao radar at an elevation angle of 0.5° at 15:30 on August 16, 2019 (the solid white ellipse represents the hail area, and the dashed black ellipse represents the three-body scattering area).

[0048] Figure 5 The HCA-Opt hydrogel classification results from the Qingdao radar at an elevation angle of 0.5° at 15:30 on August 16, 2019 (the solid white ellipse represents the hail area, and the dashed black ellipse represents the three-body scattering area).

[0049] Figure 6 The Z-axis elevation angle of the Jinan radar at 14:19 on July 9, 2021, at an angle of 1.5°. H Classification results of hydrogels (black dashed lines indicate three-body scattering regions);

[0050] Figure 7 The Z-axis elevation angle of the Jinan radar at 14:19 on July 9, 2021, at an angle of 1.5°. DR Classification results of hydrogels (black dashed lines indicate three-body scattering regions);

[0051] Figure 8 The results of CC hydrogel classification at 14:19 on July 9, 2021, at an elevation angle of 1.5° using the Jinan radar (black dashed lines indicate the three-body scattering region).

[0052] Figure 9 The HCA-Opt hydrogel classification results for Jinan radar at an elevation angle of 1.5° at 14:19 on July 9, 2021 (the black dashed line represents the three-body scattering region).

[0053] Figure 10 The data includes (a) observations of 0°C altitude at Zhangqiu radiosonde station at 08:00 and 20:00 on July 9, 2021, and 0°C altitude from 08:00 to 20:00 in HHUPS mode, and (b) the horizontal distribution of 0°C altitude in HHUPS mode at 14:00.

[0054] Figure 11 The HCA (Hydrogen Carrier Analysis) results for hydrogel classification at 14:19 on July 9, 2021, at an elevation angle of 0.5° using the Jinan radar (the black dashed lines indicate areas where HCA incorrectly identified hydrogel).

[0055] Figure 12 The HCA-Opt hydrogel classification results were obtained from the Jinan radar at an elevation angle of 0.5° at 14:19 on July 9, 2021.

[0056] Figure 13 The vertical profiles of the hydrogel classification results of Jinan radar (a) HCA and (b) HCA-Opt along the 87° azimuth angle at 14:19 on July 9, 2021 (the black dashed line is the area where HCA-Opt corrected the ice crystals as graupel, and the white dashed line is the area where HCA-Opt corrected the hail as heavy rain). Specific Implementation

[0058] A method for classifying water condensate particles based on S-band dual polarization radar is provided. The water condensate particle classification adopts the HCA-Opt method. By adding the identification of hail and three-body scattering regions in the HCA method using Equations 2 and 3, the water condensate identified by HCA in the two regions is corrected. The HCA method is shown in Equation 1.

[0059]

[0060] In Equation 1:

[0061] i represents the type of hydrogel. HCA can identify 8 types of hydrogel, namely dry snow (DS), wet snow (WS), ice crystals (CR), graupel (GR), large droplets (BD), light to moderate rain (RA), heavy rain (HR), and rain-hail mixture (RH).

[0062] A represents the integrated probability value; V represents the radar parameter; j represents the parameter type, including the horizontal reflectivity factor Z. H Differential reflectivity Z DR Zero-lag correlation coefficient CC, and logarithmic form of differential phase shift LK DP Z H Texture SDZ H With differential phase texture SDΦ DP ;

[0063] P represents the membership function, and an asymmetric trapezoid is selected to determine the probability distribution boundary of the parameters;

[0064] W represents the weighting factor, which ranges from 0 to 1. The higher the value, the greater the effect of identifying water-coagulated substances.

[0065] Q represents the confidence factor, which eliminates the effects of radar calibration, attenuation, non-uniformity filling, beam blocking, observation errors, and noise.

[0066]

[0067] In Equation 2:

[0068] ETOP18dBZ Z represents H The echo height is 18 dBZ;

[0069] In Equation 3:

[0070] ETOP 0dBZ Z represents H The echo height is 0 dBZ;

[0071] r stormcore The strong echo region indicates the radial direction Z. H Areas with a cumulative distance greater than 1 km and a range greater than 45 dBZ;

[0072] R is the radial distance from the strong echo zone to the radar station;

[0073] Equation 2 indicates that CC is less than 0.9, and Z H When the echo top height of 45dBZ or 18dBZ is greater than 8 km, it is considered a hail zone.

[0074] Equation 3 indicates that the area where TBSS may occur is determined behind the center of the strong echo. Then, the water condensate type in the TBSS area is determined by using the CC threshold and the 0dBZ echo top height. If the conditions of Equation 3 are met, it is a precipitation type; otherwise, it is a non-precipitation type.

[0075] When a certain distance library parameter satisfies Equation 2 or Equation 3, HCA-Opt sets the integration probability of land features and organisms to 0.

[0076] The height of the melt layer was then determined by combining the temperature analysis field using the HHUPS short-term mode.

[0077] The temperature analysis field of the HHUPS short-term model is as follows: This model is built upon the Weather Research and Forecast (WRF) model, using 0.25° resolution data from the National Center for Environmental Prediction (NCEP) Global Forecast System (GFS) as background field data. Data Assimilation (DA) and Gridpoint Statistical Interpolation (GSI) modules are used as the model assimilation system. It performs two cold starts daily at 08:00 and 20:00, with other times being warm starts. The forecast field one hour after integration of the previous time period serves as the background field for the next time period. Observational data is assimilated every half hour. Currently, the assimilated data includes data from national-level surface meteorological observation stations, automatic surface meteorological observation stations, conventional radiosonde stations, L-band radar radiosonde stations, aircraft reports, radar reflectivity, and data from the Himawari-8 satellite. The horizontal resolution is 3 km, and the temporal resolution is 1 hour. 0-24 hour forecasts are generated every day, with the 0 hour forecast field being the analysis field for the starting time of the forecast.

[0078] The 0°C height is extracted from the temperature analysis field of the short-term model as the top of the melting layer and the 2°C wet-bulb temperature as the bottom of the melting layer. The latitude and longitude corresponding to the center of each range library detected by radar are calculated in sequence. The data of the four closest model grid points are selected, and the heights corresponding to the top and bottom of the melting layer of the model are interpolated to the center coordinates of the calculated range library using the bilinear method.

[0079] Then, based on the vertical distribution constraints of the water-coagulate types, the vertical distribution of the identified water-coagulates is limited. The vertical distribution constraints of the water-coagulate types are shown in the table below:

[0080]

[0081] In the table, √ and × indicate whether the corresponding water condensate in the column can or cannot appear below the corresponding water condensate in the row. When a distance reservoir is identified as dry snow (DS), see row 1 in the table, any type of water condensate can appear below it. When a distance reservoir is identified as wet snow (WS), see row 2 in the table, dry snow (DS) and ice crystals (CR) cannot appear below it.

[0082] Since low elevation angles are easily contaminated by non-meteorological echoes, the water condensate type identified at the highest elevation angle is taken as the true value, and corrections are made sequentially from high elevation angle to low elevation angle according to the table's constraints. When the water condensate type identified at the lower elevation angle does not meet the conditions, the water condensate that meets the conditions is checked based on the classification results of the upper elevation angle, and the water condensate corresponding to the maximum integration probability value is selected as the corrected classification result.

[0083] The above method will be applied in specific ways as follows:

[0084] Example 1

[0085] On August 16, 2019, a severe hailstorm occurred in Zhucheng City, Weifang, Shandong Province. Figure 1-5 The data provides radar parameters and water condensate distribution at a 0.5° elevation angle in Qingdao at 15:30 on August 16th, showing a strong echo area (Z) within Zhucheng. H There is a warm and humid airflow inlet on the south side (>60 dBZ). Figure 1 (Black arrows) Echo top heights generally exceed 10 km, consistent with supercell characteristics;

[0086] Z corresponding to the strong echo region DR Mainly distributed in the range of -2.5 to 0 dB ( Figure 2 CC is mainly distributed in the range of 0.8-0.9 ( Figure 3 In addition to hail and heavy rain, the hydrophobic material in the strong echo areas identified by HCA was also identified as ground features. Figure 4 (white ellipse), this is due to non-spherical hailstones with relatively fixed orientation, Z DR Normally less than 0, but hail melting causes solid and liquid hydrophobic condensates to coexist, reducing CC and Z. DR The low membership degree of hail with CC results in a low integration probability value for hail, leading to misidentification as ground features. After adding the hail area discrimination condition (Equation 2), HCA-Opt no longer identifies some strong echo areas as ground features. Figure 5 Instead of identifying it as rain and hail, it was identified as rain and hail, which is consistent with the actual situation.

[0087] To the west of the strong echo center, within the territory of Yishui County, there are three radial Z-shaped lines. DR Distribution in -3 to 0 dB ( Figure 2 Black ellipse), CC distribution is between 0.65 and 0.85 ( Figure 3 (black ellipse), exhibiting TBSS characteristics, Z DR The negative value region and the low CC value region are caused by the attenuation of electromagnetic waves by strong hail and the non-uniform filling behind the hail, respectively, corresponding to Z. H The Z-axis is 20-40 dBZ, and the elevation angle of the upper layer is... H The distribution range is roughly the same, and the vertical gradient is small, indicating that it should be precipitation echo, based on the HCA classification results ( Figure 4 As can be seen from the black ellipse, this area includes not only moderate (light) rain, but also ground features. DR Negative values ​​and small CC values ​​result in a higher probability of ground features being integrated than hail. Therefore, a TBSS discrimination condition (Equation 3) is added. Figure 4 The ground features in the black ellipse area were identified by HCA-Opt as moderate (light) rain. Figure 5 ).same, Figure 4The area in the black ellipse in the middle, which was originally identified as a land feature, was corrected by HCA-Opt to be hail and moderate (light) rain.

[0088] Example 2

[0089] On the afternoon of July 9, 2021, a hailstorm occurred in Zhangqiu District, Jinan City. Figure 6-9 The data shows the radar parameter distribution at a 1.5° elevation angle in Jinan at 14:19 on July 9, 2021, indicating a strong echo center in eastern Zhangqiu. Figure 6 ), Z H The HCA-Opt classification result exceeds 65 dBZ. Figure 9 This area will be identified as a hail zone. The area behind the hail zone (Z) DR First increase then decrease to a negative value ( Figure 7 (black dashed line), CC is also low ( Figure 8 (Black dashed line) shows typical TBSS characteristics in this region Z. H It is mainly distributed in 5-15 dBZ, with a low echo top, which does not meet the conditions of Equation 3. HCA-Opt identifies this area as a super-refractive feature, which corresponds to TBSS.

[0090] Example 3

[0091] The height of the melt layer can provide some limitations for the classification results of water condensate. Below the melt layer, the classification results should not include dry (wet) snow or ice crystals, and above the melt layer, the classification results should not include moderate (light) rain or heavy rain. Figure 10 'a' is the hourly 0°C height in Zhangqiu on July 9, 2021, given by the HHUPS model analysis field, and the 0°C height observed by radiosonde at 08:00 and 20:00 on the same day. It can be seen that the 0°C height (melt layer top) from the model analysis is not significantly different from the radiosonde observations; therefore, using the 0°C height from the model analysis field to replace the radiosonde observations is feasible. At the same time, it can be observed that the 0°C height given by the model fluctuates significantly over time. From the HHUPS model's 14:00 0°C height analysis field (… Figure 10 (b) It can be seen that the 0°C elevation gradually increases by about 300 m from eastern Northwest Shandong to Central Shandong. Therefore, introducing hourly melting layer heights from model analysis can better match radar observations and improve the constraining effect of melting layer heights on hydrophobic material identification.

[0092] Figure 11 , Figure 12 The results presented are the HCA and HCA-Opt classification results from the Jinan radar at an elevation angle of 0.5° at 14:19 on July 9, 2021. The HCA hydrogel classification results show ( Figure 11In the Bohai Sea, eastern Weifang, and coastal areas of Qingdao and Rizhao (black dashed lines), there is a large amount of moderate to light rain, mixed with some dry snow and graupel. These areas are 250-350 km from the radar station, at a vertical altitude of 8-12 km, above the -20°C altitude (7.4 km), making it impossible for liquid water condensate to exist. Analysis 14:19 Jinan Radar Station Elevation Angle Z at Each Level H Z DR As shown in the CC distribution (figure omitted), the melt layer characteristics are not obvious and the MLDA threshold condition is not met. Therefore, HCA cannot stratify the height and thus cannot define the type of hydrophobic material. The membership functions of moderate (light) rain and dry snow are roughly the same, leading to an alternating distribution of liquid and solid hydrophobic materials in these areas, with moderate (light) rain being dominant. After incorporating the melt layer height information provided by the model analysis field, HCL-Opt identifies the hydrophobic material types in the above areas primarily as dry snow, graupel, and ice crystals, eliminating moderate (light) rain above the zero-degree layer. Figure 12 The distribution is more reasonable.

[0093] Example 4

[0094] Figure 13 A profile of HCA and HCA-Opt hydrogel classification along an 87° azimuth is presented at 14:19 on July 9, 2021. At an elevation angle of 3.3° and a distance of 69-73 km (… Figure 13 a) HCA identified ice crystals (CR) as part of the hydrophobic material below graupel (GR). Graupel is formed by dry snow or ice crystals adhering to frozen droplets. During its descent, graupel collides with the frozen droplets to form hail or melts into rain. Therefore, it is unreasonable for ice crystals to appear near the zero-degree layer and below graupel. At a distance of 0.5° elevation and 76km, two distances show rain and hail, located below the heavy rain at a 1.5° elevation angle and separated from the hail at a 2.4° elevation angle, which is also not entirely reasonable. After adding the constraint of vertical distribution of hydrophobic material, HCA-Opt identified ice crystals at 69-73km as graupel. Figure 13 (b. Black dashed line area), hail at 76km was identified as heavy rain. Figure 13 (b. White dashed line area) The vertical distribution of hydrogel is more reasonable.

Claims

1. A method for classifying water-soluble particles based on S-band dual-polarization radar, characterized in that, The classification of hydrophobic particles uses the HCA-Opt method. By adding equations 2 and 3 to the HCA method to identify hail and three-body scattering regions, the HCA method is corrected for its identification of hydrophobic particles in these two regions. The HCA method is shown in equation 1. In Equation 1: A represents the integrated probability value; i represents the type of hydrogel, HCA can identify 8 types of hydrogel: dry snow (DS), wet snow (WS), ice crystals (CR), graupel (GR), large droplets (BD), light to moderate rain (RA), heavy rain (HR), and mixed rain and hail (RH); V represents radar parameters, and j represents the type of radar parameter, including the horizontal reflectivity factor (Z). H Differential reflectivity Z DR Zero-lag correlation coefficient CC, and logarithmic form of differential phase shift LK DP Z H Texture SDZ H With differential phase-shift texture SDΦ DP P represents the membership function, which generalizes the radar parameters; W represents the weighting factor, which ranges from 0 to 1, with higher values ​​indicating greater effectiveness in identifying water-based contaminants; Q represents the confidence factor, which eliminates the effects of radar calibration, attenuation, non-uniformity filling, beam blocking, observation errors, and noise. In Equation 2: ETOP 18dBZ Z represents H The echo height is 18 dBZ; In Equation 3: ETOP 0dBZ Z represents H The echo height is 0 dBZ; r stormcore The strong echo region indicates the radial direction Z. H Areas with a cumulative distance greater than 1 km and a range greater than 45 dBZ; R is the radial distance from the strong echo zone to the radar station; Equation 2 indicates that CC is less than 0.9, and Z H When the echo top height of 45dBZ or 18dBZ is greater than 8 km, it is considered a hail zone. Equation 3 indicates that the area where TBSS may occur is determined behind the center of the strong echo. Then, the water condensate type in the TBSS area is determined by using the CC threshold and the 0dBZ echo top height. If the conditions of Equation 3 are met, it is a precipitation type; otherwise, it is a non-precipitation type. When a certain distance database parameter satisfies Equation 2 or Equation 3, HCA-Opt sets the integration probability of land features and organisms to 0; The height of the melt layer was determined by temperature analysis field using the HHUPS short-term mode.

2. The method for classifying water-soluble particles based on S-band dual-polarization radar according to claim 1, characterized in that, The temperature analysis field of the HHUPS short-term model is constructed based on weather research and forecasting models, using 0.25° resolution data from the US National Environmental Prediction Center's global forecast model as background field data, and a data assimilation and gridded statistical interpolation module as the model assimilation system. Two cold starts are performed daily at 08:00 and 20:00, with other times being hot starts. The forecast field one hour after integration of the previous time period serves as the background field for the next time period. Observational data is assimilated every half hour, including data from national-level surface meteorological stations, automatic surface meteorological stations, conventional radiosondes, L-band radar radiosondes, aircraft reports, radar reflectivity, and Himawari-8 satellite data. 0-24 hour forecasts are produced hourly each day, with the 0-hour forecast field serving as the analysis field at the start time.

3. The method for classifying water-soluble particles based on S-band dual-polarization radar according to claim 1 or 2, characterized in that, The 0°C temperature height is extracted from the temperature analysis field of the short-term model as the melting layer top and the 2°C wet-bulb temperature as the melting layer bottom. The latitude and longitude corresponding to the center of each range library detected by radar are calculated in sequence. The data of the four closest model grid points are selected, and the corresponding heights of the melting layer top and bottom of the model are interpolated to the center coordinates of the calculated range library using the bilinear method.

4. The method for classifying water-soluble particles based on S-band dual-polarization radar according to claim 1, characterized in that, It also includes limiting the vertical distribution of identified hydrogels based on vertical distribution constraints of hydrogel types, as shown in the table below: In the table, √ and × indicate whether the corresponding water condensate in the column can or cannot appear below the corresponding water condensate in the row. When a distance reservoir is identified as dry snow (DS), see row 1 in the table, any type of water condensate can appear below it. When a distance reservoir is identified as wet snow (WS), see row 2 in the table, dry snow (DS) and ice crystals (CR) cannot appear below it.

5. The method for classifying water-soluble particles based on S-band dual-polarization radar according to claim 4, characterized in that, Since low elevation angles are easily contaminated by non-meteorological echoes, the water condensate type identified at the highest elevation angle is taken as the true value, and corrections are made sequentially from high elevation angle to low elevation angle according to the table's constraints. When the water condensate type identified at the lower elevation angle does not meet the conditions, the water condensate that meets the conditions is checked based on the classification results of the upper elevation angle, and the water condensate corresponding to the maximum integration probability value is selected as the corrected classification result.

6. The method for classifying water-soluble particles based on S-band dual-polarization radar according to claim 1, characterized in that, Implement the HCA-Opt method using the Fortran language.

7. The application of the method according to any one of claims 1-6 in the identification and classification of hydrogel particles.

Images