Precipitation phase vertical detection and inversion method based on radio wave reflection

Abstract

The application discloses a precipitation phase state vertical detection and inversion method based on radio wave reflection and relates to the technical field of meteorological remote sensing detection. The application comprises S1, multi-frequency band radio wave synchronous transmission and echo signal collection: through a multi-frequency band radio transceiving system, multiple vertical detection radio waves of different frequency bands are synchronously transmitted, echo signals generated by precipitation particles in different height layers of the atmosphere are collected, and continuous collection of echo signals of full height layers is completed; through synchronous transmission of multi-frequency band radio waves and collection of echo signals of full height layers, the application obtains multi-dimensional scattering characteristic vertical profiles of different phase state particles, realizes decoupling division of phase state intervals of different height layers by constructing a multi-frequency band exclusive feature matrix and combining melting layer vertical gradient characteristics, corrects particle proportion weights in mixed phase state intervals by introducing adjacent height layer phase state continuity constraint conditions, and realizes accurate interpretation of the mixed phase state of the zero-degree layer melting zone.

Description

Technical Field

[0001] This invention relates to the field of meteorological remote sensing technology, specifically to a method for vertical detection and inversion of precipitation phase based on radio wave reflection. Background Technology

[0002] Vertical precipitation phase detection refers to the technical process of conducting continuous, high-resolution observations of the vertical spatial distribution of precipitation particles in the atmosphere, from the cloud precipitation generation area to the near-surface layer, to observe the solid, liquid, and solid-liquid mixed states. Vertical precipitation phase inversion refers to the technical process of inverting the phase types, phase proportions, and continuous vertical profile distribution of precipitation particles in different atmospheric layers based on the echo characteristics, atmospheric temperature and humidity profiles, and other relevant physical parameters obtained from detection, using a precipitation particle scattering physical model and supporting interpretation algorithms.

[0003] The vertical distribution and dynamic evolution of precipitation phases directly determine the final precipitation type near the ground, including rain, snowfall, freezing rain, and sleet. These are core input parameters for short-term nowcasting weather warnings, severe weather prevention and control, atmospheric water resource assessment, and numerical weather prediction model optimization. Accurate vertical detection and inversion of precipitation phases can clearly depict the entire phase transition process of precipitation particles during their descent, precisely identify the phase transition characteristics of the zero-degree melting zone, and provide crucial data support for early warning and forecasting of disaster-prone weather such as blizzards, freezing rain, and hail. This plays a vital role in ensuring the safety of transportation, agricultural production, urban infrastructure operation, and the lives and property of the public.

[0004] However, current precipitation phase detection and inversion technologies mostly rely on single-band radar detection as the core method, which makes it difficult to effectively distinguish the differences in scattering characteristics of solid-liquid mixed phase particles within the zero-degree melting zone. They also cannot accurately decouple the vertical transition intervals of mixed phases, resulting in insufficient identification accuracy of mixed phase regions and significant ambiguity in the inversion results. Furthermore, they struggle to achieve continuous and accurate phase inversion across all altitudes from near-surface to above zero-degree layers within clouds. This fails to meet the accuracy, continuity, and reliability requirements of precipitation phase vertical profile data for severe weather warnings and forecasts under complex meteorological conditions. Therefore, developing a precipitation phase vertical detection and inversion method based on radio wave reflection is of great significance. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method for vertical detection and inversion of precipitation phase states based on radio wave reflection. This method can solve the problems of insufficient identification accuracy of the vertical transition zone of the mixed phase state in the zero-degree layer melting zone and prominent multiple solutions in the existing single-band detection technology, which makes it impossible to achieve continuous and accurate inversion of the phase state of the entire layer.

[0006] To solve the above-mentioned technical problems, this invention provides the following technical solution: a method for vertical detection and inversion of precipitation phase state based on radio wave reflection, the method comprising the following steps: S1. Multi-band radio wave synchronous transmission and echo signal acquisition: Through a multi-band radio transceiver system, multiple vertical sounding radio waves of different frequency bands are transmitted synchronously to acquire echo signals generated by the reflection of precipitation particles at different altitudes in the atmosphere, thus completing the continuous acquisition of echo signals at all altitudes. S2. Echo signal preprocessing and feature profile extraction: The acquired echo signals of each frequency band are preprocessed to remove system noise and interference signals. The reflectivity, differential reflectivity, phase shift and attenuation features of each frequency band corresponding to different height layers are extracted to generate continuous vertical profiles of each frequency band feature. S3. Construction of Multi-Band Scattering Feature Matrix and Preliminary Phase Identification: Based on the differences in scattering characteristics of solid, solid-liquid mixture, and liquid precipitation particles in different frequency bands, a multi-band exclusive feature matrix corresponding to different phase particles is constructed. The multi-band features of each altitude layer are matched with the feature matrix to complete the preliminary identification of precipitation phase at each altitude layer. S4. Vertical gradient analysis and phase decoupling of the melting layer: Based on the feature vertical profile, the vertical gradient change characteristics of the melting zone of the zero-degree layer are extracted. Combined with the continuous change law of the features in the melting zone, the phase intervals that are initially identified are decoupled layer by layer according to height, and the vertical boundaries of the solid, solid-liquid mixture and liquid phase intervals are clarified. S5. Phase continuity constraint and mixed phase ratio correction: Phase continuity constraint conditions of adjacent height layers are introduced, and a weighted smoothing algorithm is used to correct the feature data in the solid-liquid mixed phase interval, and the proportion weight of different phase particles in the mixed phase interval is calculated. S6. Generation of vertical profiles of precipitation phases across all altitudes: Based on the decoupled phase intervals and the corrected phase proportion weights, the phase identification results of the entire altitude layer are integrated to generate continuous vertical profiles of precipitation phases across all altitudes.

[0007] Furthermore, in step S1, the vertically probed radio waves synchronously transmitted by the multi-band radio transceiver system include dual-polarized radio waves in the S-band, X-band, and Ka-band, that is, horizontally polarized and vertically polarized waves are transmitted simultaneously. The transceiver channels of each band of the multi-band radio transceiver system are synchronously controlled using the same reference clock. The transmission beams of each band of radio waves adopt vertically oriented pencil beams. The main lobe width of each band beam is kept consistent. The sampling frequency and the number of sampling points of the echo signals of each band are kept uniform.

[0008] Furthermore, during the continuous acquisition of echo signals across the entire altitude layer in step S1, the multi-band radio transceiver system continuously and repeatedly probes at preset time intervals. The transmission power, pulse width, and pulse repetition frequency of each probe remain fixed. Each echo signal acquired during a probe corresponds to a unique probe timestamp and altitude layer calibration data. The echo signals of each frequency band within the same probe cycle are matched using the same altitude layer calibration system.

[0009] Furthermore, step S2, when preprocessing the acquired echo signals of each frequency band, includes the following steps: Time-domain synchronization and alignment of echo signals in each frequency band are performed, and echo signals in different frequency bands are matched one-to-one according to sampling time and detection height to complete the spatiotemporal registration of multi-frequency band echo signals. Time-domain filtering and frequency-domain denoising are performed on the registered echo signal to filter out the inherent system noise and random interference signals in the echo signal. Clutter identification and removal processing is performed on the denoised echo signal to remove fixed clutter and multipath interference signals generated by non-precipitation targets in the echo signal. Beam pointing correction and propagation distance calibration are performed on the processed echo signal to establish the mapping relationship between the echo signal and the corresponding altitude layer.

[0010] Furthermore, step S3, when constructing multi-band specific feature matrices corresponding to particles of different phase states and performing feature matching, includes the following steps: Based on Mie scattering theory, the scattering characteristics of solid, solid-liquid mixture, and liquid precipitation particles under different frequency bands of radio waves were simulated to obtain a simulated dataset of scattering characteristics of particles of different phases for each frequency band. Historical sounding data and field calibration test data were collected, and the scattering characteristic simulation dataset was corrected to obtain the characteristic threshold ranges of each frequency band for particles of different phases. Establish the correlation between features of different frequency bands. The quantitative correlation is the ratio range of the same feature in different frequency bands and the Pearson correlation coefficient. Construct a three-dimensional feature matrix with phase type as the first dimension, detection frequency band as the second dimension, and feature type as the third dimension. Use the feature threshold range corresponding to different phase particles as the basic value of the matrix elements and the quantitative correlation between features of different frequency bands as the linkage constraint term of the matrix elements. Complete the integration and generate a multi-frequency band exclusive feature matrix for the corresponding phase particles. The linkage constraint term includes the ratio of features in different frequency bands and the Pearson correlation coefficient, which is used to constrain the consistency of the association of multi-frequency band features. During phase matching, the single-frequency band threshold and multi-frequency band association constraints are checked simultaneously, and the term directly participates in the matching judgment to ensure that the feature matrix construction and the recognition algorithm are consistent. A multi-band feature matching algorithm is used to calculate the matching degree between height layer features and phase features. The matching degree calculation formula is as follows: ,in Let f be the feature matching degree of the p-th phase state at height h, and f be the frequency band number. Here, δ represents the feature weighting coefficient of the f-th frequency band, and δ is the binary matching decision function. Let f be the measured characteristic value of the f-th frequency band at the h-th altitude level. The feature threshold interval for the p-th phase state in the f-th frequency band, and the feature weight coefficients. The binary matching judgment function δ is determined by field calibration data of the scattering sensitivity of precipitation particles in different frequency bands. When the measured characteristic value falls into the characteristic threshold range, the value is 1, and when it does not fall into the range, the value is 0.

[0011] Furthermore, step S5, when correcting the characteristic data within the solid-liquid mixed phase range and calculating the proportion weights, includes the following steps: Based on the physical process law of precipitation particle phase transformation, a reasonable range of phase ratio changes between adjacent height layers is set to form phase continuity constraints between adjacent height layers. Based on the phase continuity constraint, constraint verification is performed on the multi-band feature data of each height layer in the solid-liquid mixed phase range, and the effective feature data that meets the constraint conditions is selected. Define the solid-state particle characteristic components and liquid-state particle characteristic components of the h-th height layer and the f-th frequency band. Based on the characteristic threshold intervals of solid and liquid phases in the multi-band dedicated characteristic matrix, and combined with the prior scattering characteristic constraints of Mie scattering theory, a dual-band constraint decoupling method is used to separate the solid-state particle characteristic components of the h-th height layer and the f-th frequency band from the effective feature data. With liquid particle characteristic components A dual-frequency band constraint decoupling method is used to separate solid and liquid feature components from effective feature data. The specific implementation steps are as follows: Linear superposition constraint: The overall characteristics of the mixed phase state satisfy ,in The measured total characteristics of the f-th frequency band at the h-th altitude level Characteristic components of solid particles, Characteristic components of liquid particles; Dual-band ratio constraint: Select any two frequency bands ( , The ratio of solid / liquid particle dual-band scattering characteristics is a fixed constant, i.e. ,in , Obtained through field calibration tests; Solving the system of equations: Simultaneously solve the dual-band superposition equations and the proportional constraint equations, then use the least squares method iteratively to obtain a unique solution. and Complete the decoupling and separation of solid and liquid characteristic components; Based on the separated solid-state particle feature components and liquid-state particle feature components, combined with a multi-band specific feature matrix, the proportion weight of solid-state particles is calculated using a mixed-phase proportion calculation formula. The proportion calculation formula is as follows: ,in The weighting of solid precipitation particles at the h-th altitude layer is given by [the weighting]. Let f be the particle identification coefficient for the f-th frequency band. The solid particle characteristic components of the h-th height layer and the f-th frequency band are... The characteristic components of liquid particles in the f-th frequency band at the h-th height level, and the particle identification coefficient. Using synchronously measured data from a ground-based raindrop spectrometer and a radiosonde, a linear fitting statistical method was employed to determine: taking the radiosonde's measured phase state as the true value and the raindrop spectrometer's particle spectrum as the benchmark, scattering characteristics in the S / X / Ka bands were simultaneously acquired, and a fitting equation was constructed. ,in The scattering sensitivity of the f-th frequency band is... , These are the frequency band fitting coefficients; S-band X-band Ka band The sum of the coefficients for the three frequency bands is 1.

[0012] Furthermore, in step S3, when matching the multi-band features of each altitude layer with the multi-band exclusive feature matrix, a multi-band feature matching algorithm is used to compare the multi-band feature data extracted from each altitude layer with the feature threshold intervals corresponding to different phase particles in the feature matrix one by one, calculate the matching degree between the feature data of each altitude layer and the feature threshold intervals corresponding to different phase particles, and select the phase type with the highest matching degree as the preliminary identification result of the precipitation phase of that altitude layer.

[0013] Furthermore, in step S4, when extracting the vertical gradient variation features of the zero-degree melting zone, based on the continuous vertical profiles of each frequency band feature, the first and second derivatives of each detection feature as a function of height are calculated. The gradient feature value is determined using the melting layer gradient feature synthesis formula, which is as follows: ,in Let f be the gradient characteristic value of the h-th height layer and f-th frequency band. The first derivative of the characteristic of the f-th frequency band with altitude. Let be the second derivative of the characteristic of the f-th frequency band as a function of altitude, and α be the gradient synthesis coefficient. The gradient synthesis coefficient α is determined by historical statistical data of the vertical profile of atmospheric temperature, and the value of α ranges from [0.2, 0.8]. The fitting function is... ,in The vertical temperature gradient is expressed in °C / 100m. When the absolute value of the vertical temperature gradient is greater than 1°C / 100m, α takes the upper limit of 0.8. When the absolute value of the vertical temperature gradient is less than 0.2°C / 100m, α takes the lower limit of 0.2. Based on the calculation results, the abrupt change interval of the feature with altitude is determined, and the altitude range corresponding to the abrupt change interval is defined as the zero-degree layer melting zone. The vertical gradient change law of each detection feature within the melting zone is extracted.

[0014] Furthermore, the phase continuity constraint condition introduced in step S5 is based on the physical process of phase transformation of precipitation particles during their descent in the atmosphere. The constraint condition limits the variation range of the proportion weight of particles of the same phase between two adjacent altitude layers to within a preset reasonable range. The quantitative standard of the reasonable range is: when the height difference between two adjacent altitude layers does not exceed 100m, the variation range of the proportion weight of solid particles does not exceed 8%. The parameters of the constraint condition are adjusted accordingly based on the vertical profile data of atmospheric ambient temperature. When the absolute value of the vertical temperature gradient is greater than 1℃ / 100m, the upper limit of the variation range of the reasonable range is widened to 12%. When the absolute value of the vertical temperature gradient is less than 0.2℃ / 100m, the upper limit of the variation range of the reasonable range is tightened to 5%.

[0015] Furthermore, in step S6, when integrating the phase identification results of the entire height layer, the phase type and phase particle proportion weight of each height layer are integrated in order from high to low height. The phase data of adjacent height layers are smoothly connected. The generated continuous vertical profile of precipitation phase of the entire height layer includes three types of data: phase type, solid particle proportion weight, and liquid particle proportion weight corresponding to each height layer.

[0016] Compared with existing technologies, this method for vertical detection and inversion of precipitation phase based on radio wave reflection has the following advantages: This invention acquires multi-dimensional scattering characteristic vertical profiles of particles in different phases by simultaneously transmitting multi-band radio waves and collecting echo signals from all altitude layers. By constructing a multi-band exclusive feature matrix and combining it with the vertical gradient characteristics of the melting layer, it achieves decoupled division of phase intervals layer by layer. By introducing phase continuity constraints between adjacent altitude layers, it corrects the particle proportion weights within the mixed phase interval, achieving accurate interpretation of the mixed phases in the zero-degree melting zone. This effectively avoids the multi-solution problem of traditional single-band inversion methods, significantly improves the identification accuracy of the vertical distribution of precipitation phases, and completes continuous and accurate inversion of phases across all altitude layers, providing reliable core data support for the early warning and forecasting of severe weather.

[0017] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0019] Figure 1 This is a flowchart of a method for vertical detection and inversion of precipitation phase state based on radio wave reflection; Figure 2 This is a flowchart illustrating the steps of a method for vertical detection and inversion of precipitation phase state based on radio wave reflection. Figure 3 The flowchart shows the process of constructing multi-band specific feature matrices for particles of different phases and performing feature matching. Detailed Implementation

[0020] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0021] The present invention provides a method for vertical detection and inversion of precipitation phase state based on radio wave reflection, belonging to the field of meteorological remote sensing technology. The core technical solution involves simultaneously transmitting dual-polarized radio waves in S-band, X-band, and Ka-band using a multi-band radio transceiver system. Dual polarization includes horizontal and vertical polarization. This enables continuous acquisition of precipitation particle echo signals across the entire atmospheric altitude layer. The acquired echo signals are preprocessed, and vertical profiles of multi-dimensional scattering characteristics such as reflectivity, differential reflectivity, phase shift, and attenuation are extracted. Based on Mie scattering theory, multi-band specific feature matrices for solid, solid-liquid mixtures, and liquid precipitation particles are constructed. Feature matching is then used to retrieve the vertical profiles of precipitation at each altitude layer. Precipitation phase identification is performed, followed by extraction of vertical gradient variation features of the zero-degree melting zone based on characteristic vertical profiles. This achieves decoupling and layer-by-layer division of phase intervals, and introduces phase continuity constraints at adjacent height levels to correct the particle proportion weights within the solid-liquid mixed phase intervals. Finally, the phase identification results across all height levels are integrated to generate a continuous vertical profile of precipitation phases that includes phase type, solid particle proportion, and liquid particle proportion. This fundamentally solves the industry pain points of traditional single-band detection technology, such as insufficient accuracy in identifying mixed phases in the zero-degree melting zone, significant inversion ambiguity, and the inability to achieve continuous and accurate inversion across all height levels. This provides core data support for severe weather warnings and forecasts. A detailed description is provided below with specific implementation examples.

[0022] This embodiment is applied to meteorological detection scenarios of severe weather such as blizzards and freezing rain in northern winters. In such scenarios, precipitation particles from the upper atmosphere to the near ground undergo continuous phase transitions from solid to solid-liquid mixture to liquid. Identifying the mixed phases in the zero-degree melting zone is extremely difficult, and traditional detection methods cannot meet the requirements for accurate early warning. The technical solution of this invention can achieve refined detection and inversion of precipitation phases at all altitudes. The specific implementation process is as follows.

[0023] Synchronous transmission and echo signal acquisition of multi-band radio waves; such as Figure 1 , Figure 2 As shown, the first step is to activate a multi-band radio transceiver system adapted for meteorological observation scenarios. This system is the core hardware for achieving synchronous multi-band detection. Internally, the system configures the same reference clock for the S-band, X-band, and Ka-band transceiver channels. This unified clock source ensures precise synchronization of the three bands' transmission and reception timing, preventing timing deviations in signal acquisition across different bands. The system transmits radio waves using vertically oriented pencil beams, with the main lobe widths of the beams remaining completely consistent across all three bands. The sampling frequency and number of sampling points for the echo signals are also set to a unified standard, ensuring that the acquisition specifications of echo data from different bands are matched, laying the foundation for subsequent spatiotemporal registration.

[0024] The system conducts continuous and repeated detections at preset time intervals, strictly maintaining constant transmission power, pulse width, and pulse repetition frequency during each detection to prevent fluctuations in detection parameters from affecting the quality of the echo signal. After detection begins, dual-polarized radio waves in three frequency bands are simultaneously transmitted vertically upwards into the atmosphere. These radio waves penetrate different levels of the troposphere and, upon encountering high-altitude solid snowflakes, solid-liquid mixed-phase particles near the zero-degree layer, or near-surface liquid raindrops, generate corresponding reflected echo signals. The multi-band radio transceiver system receives these echo signals in real time, completing continuous acquisition of echo signals across the entire atmospheric altitude. Simultaneously, each acquired echo signal is matched with a unique detection timestamp and altitude calibration data. S-band, X-band, and Ka-band echo signals acquired within the same detection cycle are all matched using the same altitude calibration system, ensuring a precise correspondence between the echo signals of different frequency bands and the detection altitude, eliminating altitude misalignment issues.

[0025] This step, through multi-band synchronous transmission and continuous acquisition across all altitude layers, obtained raw scattered echo data of precipitation particles at different frequency bands, covering the entire detection altitude from cloud precipitation generation area to near-ground level, providing complete and stable raw data support for subsequent signal processing and feature extraction.

[0026] Echo signal preprocessing and feature profile extraction: After acquiring the raw echo signal, the process proceeds to echo signal preprocessing and feature profile extraction, such as... Figure 2 As shown, this step involves multi-level signal processing to eliminate interference and extract effective features, which is a key step in ensuring the accuracy of phase recognition.

[0027] First, the time-domain synchronization and alignment of the collected echo signals of each frequency band is performed. The echo signals of S-band, X-band, and Ka-band are matched one by one according to the sampling time and detection height to complete the spatiotemporal registration of the multi-frequency band echo signals. This eliminates the spatiotemporal deviations generated by different frequency band signals during transmission and acquisition, and ensures that the echo data of the three frequency bands correspond to the same detection height and the same detection time.

[0028] After spatiotemporal registration is completed, time-domain filtering and frequency-domain denoising are performed on the echo signal. Time-domain filtering is used to filter out the inherent noise of the system in the echo signal, while frequency-domain denoising is used to suppress random interference signals such as atmospheric turbulence and electromagnetic interference. After dual denoising, the signal-to-noise ratio of the echo signal is significantly improved, and the scattering characteristic information of precipitation particles is effectively preserved.

[0029] After the noise reduction process is completed, clutter identification and removal operations are carried out. By comparing signal features, non-precipitation target clutter in the echo signal is identified, including fixed clutter generated by ground buildings, mountains, vegetation, etc., as well as multipath interference signals caused by atmospheric refraction and ground reflection. These invalid clutter signals are completely removed to ensure that the pre-processed echo signal contains only effective scattering data of precipitation particles.

[0030] Finally, beam pointing correction and propagation distance calibration are performed on the processed echo signal. Beam pointing correction corrects the small angular deviations during beam propagation, while propagation distance calibration calculates the accurate detection altitude using radio wave propagation speed and propagation time, establishing a unique mapping relationship between the echo signal and the corresponding altitude layer.

[0031] After preprocessing, reflectivity, differential reflectivity, phase shift, and attenuation features of each frequency band at different altitudes are extracted. The feature data of each altitude are arranged in descending order of altitude to generate continuous vertical profiles for the S-band, X-band, and Ka-band. The three profiles fully present the variation of scattering characteristics of precipitation particles of different phases with altitude at each frequency band, providing multi-dimensional and high-resolution feature basis for subsequent phase identification.

[0032] The formula for calculating differential reflectance is as follows: ,in For horizontal polarization reflectivity, The vertical polarization reflectivity is given; the formula for calculating the differential phase shift is: ,in For horizontally polarized echo phase, This is the phase of the vertically polarized echo.

[0033] Construction of multi-band scattering feature matrix and preliminary phase state identification; such as Figure 3 As shown, this step relies on Mie scattering theory to construct a multi-band exclusive feature matrix, and completes the preliminary identification of precipitation phase at each altitude level through feature matching. This is the core step in distinguishing between solid, solid-liquid mixture, and liquid phases.

[0034] First, based on the Mie scattering theory, scattering characteristics simulations were carried out. Numerical simulations were performed on the scattering characteristics of solid precipitation particles, solid-liquid mixed precipitation particles, and liquid precipitation particles under S-band, X-band, and Ka-band radio wave illumination. A scattering characteristic simulation dataset for each frequency band corresponding to different phase particles was generated. This dataset covers the basic scattering characteristic laws of different phase particles in each frequency band.

[0035] Subsequently, historical radiosonde data and field calibration test data were collected to correct the scattering characteristic simulation dataset. The historical radiosonde data includes real observational information such as upper-air temperature and humidity, and precipitation particle phase states, while the field calibration test data consists of multi-frequency scattering response data collected in actual meteorological detection scenarios. By calibrating the simulation dataset using these two types of measured data, the deviation between theoretical simulation and actual detection was eliminated, and the characteristic threshold ranges of different phase particles in each frequency band were obtained.

[0036] Next, the inherent correlation between the features of the S-band, X-band, and Ka-band is analyzed. The feature threshold ranges of solid, solid-liquid mixture, and liquid phase particles are integrated with the feature correlation between frequency bands to generate multi-band feature matrices specific to the three phase particles. This matrix is ​​the core basis for phase matching and recognition. The linkage constraint term includes the ratio of features of different frequency bands and the Pearson correlation coefficient, which is used to constrain the consistency of the correlation of multi-band features. During phase matching, the single-band threshold and multi-band correlation constraints are checked simultaneously and directly participate in the matching judgment to ensure that the feature matrix construction and recognition algorithm are consistent.

[0037] In the specific implementation of this embodiment, a multi-band feature matching algorithm is used to calculate the matching degree between the measured features and phase features at each altitude layer. The matching degree calculation formula is as follows: ,in Let f be the feature matching degree of the p-th phase state at height h, and f be the frequency band number. Here, δ represents the feature weighting coefficient of the f-th frequency band, and δ is the binary matching decision function. Let f be the measured characteristic value of the f-th frequency band at the h-th altitude level. This represents the characteristic threshold interval of the f-th frequency band of the p-th phase state.

[0038] Feature weight coefficients The scattering sensitivity of precipitation particles at different frequency bands was determined through field calibration data. During the field calibration process, the scattering response intensity of solid, liquid, and solid-liquid mixed phase particles at three frequency bands was simultaneously collected. The proportion of scattering sensitivity at each frequency band was statistically analyzed to quantify the characteristic weight coefficients of the corresponding frequency bands, ensuring that the weight coefficients closely match the actual detected scattering response patterns. The binary matching judgment function δ takes a value of 1 when the measured characteristic value falls within the characteristic threshold interval, and a value of 0 when it does not.

[0039] After the calculation is completed, the matching degree of the three types of phase states at each height level is compared. The phase state type with the highest matching degree is selected as the preliminary identification result of the precipitation phase state at that height level. The preliminary determination of the precipitation phase state at all height levels is completed, providing a basic identification result for subsequent phase state decoupling and division.

[0040] Vertical gradient analysis and phase decoupling of the melting layer: During the descent of precipitation particles, the zero-degree melting zone is the core region of phase transition and a difficult area to accurately identify using traditional detection techniques. This step extracts the gradient features of the melting zone based on characteristic vertical profiles, completing the layer-by-layer decoupling of phase intervals, such as... Figure 1 , Figure 2 As shown.

[0041] First, based on the characteristic continuous vertical profiles of the S-band, X-band, and Ka-band, the first and second derivatives of each detection feature as a function of altitude are calculated. The first derivative reflects the linear rate of change of the feature with altitude, while the second derivative reflects the curvature and abrupt change of the feature. Both types of derivatives together characterize the vertical variation law of the feature.

[0042] In the specific implementation of this embodiment, the gradient feature value is determined using the melting layer gradient feature synthesis formula, which is as follows: ,in Let f be the gradient characteristic value of the h-th height layer and f-th frequency band. The first derivative of the characteristic of the f-th frequency band with altitude. Let f be the second derivative of the characteristic of the f-th frequency band as a function of altitude, and α be the gradient synthesis coefficient.

[0043] The gradient synthesis coefficient α is determined by historical statistical data of the vertical profile of atmospheric temperature. The variation law of atmospheric temperature with altitude under different meteorological conditions is statistically analyzed, with a focus on analyzing the temperature gradient characteristics near the zero-degree layer. Combining the correlation between temperature change and scattering characteristic change, the gradient synthesis coefficient is quantitatively fitted to obtain the basis for its value, ensuring that the gradient characteristic value can accurately reflect the abrupt change law of the melting zone.

[0044] The value of α ranges from [0.2, 0.8], and the fitting function is... ,in The vertical temperature gradient is expressed in °C / 100m. When the absolute value of the vertical temperature gradient is greater than 1 °C / 100m, α takes the upper limit of 0.8. When the absolute value of the vertical temperature gradient is less than 0.2 °C / 100m, α takes the lower limit of 0.2. The abrupt change range of the feature with altitude is determined based on the calculation results.

[0045] Based on the gradient eigenvalue calculation results, the abrupt change intervals of the features with altitude are located. These abrupt change intervals correspond to the altitude range of the zero-degree melting zone. The continuous change pattern of the gradient eigenvalues ​​within the abrupt change intervals clearly presents the characteristic change trend of precipitation particle phase transformation within the melting zone. Combining the continuous change pattern of features within the melting zone, the initially identified phase intervals are decoupled and divided layer by layer at different altitudes. Ambiguous judgment results in the initial identification are eliminated, and the vertical boundaries of the solid phase interval, solid-liquid mixed phase interval, and liquid phase interval are accurately defined, solving the technical problems of ambiguous division of mixed phase intervals and confusion of phase types.

[0046] Phase continuity constraints and correction of mixed phase proportions: Within the zero-degree melting zone, a solid-liquid mixed phase is dominant. Accurate calculation of particle proportions is crucial for improving inversion accuracy. This step introduces phase continuity constraints between adjacent height layers to correct the particle proportion weights within the mixed phase range. Figure 1 , Figure 2 As shown.

[0047] First, based on the physical laws governing the phase transition of precipitation particles as they fall through the atmosphere, a reasonable range for the change in the proportion of phases between adjacent altitude layers is established, forming a phase continuity constraint condition for adjacent altitude layers. This constraint condition is set based on the physical characteristics of precipitation particle phase transition, limiting the change in the proportion weight of the same phase particles between two adjacent altitude layers to a preset reasonable range, avoiding abnormal results such as abrupt changes in proportion. The parameters of the constraint condition are adjusted in real time according to the vertical profile data of atmospheric ambient temperature. When the temperature gradient is large, the range of proportion change is appropriately widened, and when the temperature gradient is small, the range of change is tightened to ensure that the constraint condition adapts to the phase transition characteristics under different meteorological environments.

[0048] Subsequently, based on this continuity constraint, the multi-band characteristic data of each height layer in the solid-liquid mixed phase range were constrained and verified. The characteristic data were verified one by one to see if they met the reasonable range requirements of the proportion change. Valid characteristic data that met the constraint conditions were selected, and abnormal characteristic data caused by noise and clutter residues were removed to ensure the data quality of the proportion calculation.

[0049] In the specific implementation of this embodiment, firstly, based on the physical laws of phase transition of falling precipitation particles, a continuity constraint between adjacent height layers is constructed to screen effective mixed phase characteristic data; then, based on the pure phase characteristic threshold range of the multi-band specific feature matrix and the prior physical constraint of Mie scattering, the solid particle characteristic component of the h-th height layer in the f-th frequency band is directly separated and extracted from the measured characteristics of the mixed phase synthetic echo using a dual-band constraint decoupling method. liquid particle characteristic components This separation and decoupling process relies solely on the prior characteristic matrix and the laws of scattering physics, without depending on the phase proportion weights to be solved, thus completely breaking the original logical loop problem. After separation, the mixture is substituted into the formula for calculating the proportion of mixed phases. ,in The weighting of solid precipitation particles at the h-th altitude layer is given by [the weighting]. Let f be the particle identification coefficient for the f-th frequency band. The solid particle characteristic components of the h-th height layer and the f-th frequency band are... The characteristic components of liquid particles in the f-th frequency band of the h-th height layer.

[0050] The solid and liquid feature components are separated from the effective feature data using a dual-band constraint decoupling method. The specific implementation steps are as follows: Linear superposition constraint: The overall characteristics of the mixed phase state satisfy ,in The measured total characteristics of the f-th frequency band at the h-th altitude level Characteristic components of solid particles, Characteristic components of liquid particles; Dual-band ratio constraint: Select any two frequency bands ( , The ratio of solid / liquid particle dual-band scattering characteristics is a fixed constant, i.e. ,in , Obtained through field calibration tests; Solving the system of equations: Simultaneously solve the dual-band superposition equations and the proportional constraint equations, then use the least squares method iteratively to obtain a unique solution. and The solid-liquid characteristic components are decoupled and separated.

[0051] Particle recognition coefficient Using synchronously measured data from a ground-based raindrop spectrometer and a radiosonde, a linear fitting statistical method was employed to determine: taking the radiosonde's measured phase state as the true value and the raindrop spectrometer's particle spectrum as the benchmark, scattering characteristics in the S / X / Ka bands were simultaneously acquired, and a fitting equation was constructed. ,in The scattering sensitivity of the f-th frequency band is... , These are the frequency band fitting coefficients; S-band X-band Ka band The sum of the coefficients of the three frequency bands is 1. During the actual measurement, the ground raindrop spectrometer collects data such as the phase state, particle size, and number concentration of near-ground precipitation particles in real time, while the radiosonde collects data such as the temperature and humidity, air pressure, and phase state of precipitation particles in the upper atmosphere simultaneously. By using the synchronous observation data of the two types of equipment, the differences in the recognition response of solid particles and liquid particles in each frequency band are statistically analyzed, and the particle recognition coefficient of each frequency band is quantified to ensure that the coefficients conform to the recognition rules of actual detection.

[0052] The solid particle proportion weight obtained by solving this formula, combined with the constraint conditions, corrects the characteristic deviation in the mixed phase range, and the liquid particle proportion weight can be derived simultaneously, thus realizing the accurate calculation of the particle proportion in the mixed phase range.

[0053] The vertical profiles of precipitation phases across all altitude levels are generated. After completing the decoupling of phase intervals and the correction of the weighting of mixed phase proportions, the process proceeds to the generation of vertical profiles of precipitation phases across all altitude levels. Figure 2 As shown.

[0054] Following the order of height hierarchy from high to low, the three core data categories of phase type, solid particle proportion weight, and liquid particle proportion weight of each height hierarchy are integrated sequentially. During the integration process, the phase data of adjacent height hierarchy are smoothly connected to eliminate abrupt deviations at the connection points of height hierarchy data and ensure the continuity and smoothness of the profile.

[0055] The final generated continuous vertical profile of precipitation phases across all altitudes fully covers the entire detection altitude from high-altitude clouds to near-ground levels. It clearly presents the complete transformation process of precipitation particles from solid to solid-liquid mixture and then to liquid during their descent, accurately depicting the mixed phase transition characteristics of the zero-degree melting zone. Each altitude level corresponds to a clear phase type and particle proportion data, possessing the characteristics of high resolution, high continuity, and high reliability.

[0056] In summary, this embodiment, through a complete technical process, successfully achieved accurate detection and inversion of precipitation phases across the entire atmospheric altitude in complex meteorological detection scenarios such as winter blizzards and freezing rain. Leveraging the differentiated advantages of multi-band scattering characteristics, it overcomes the technical bottleneck of traditional single-band techniques being unable to interpret mixed phases. By employing melt layer gradient analysis and phase continuity constraints, the ambiguity of the inversion results is significantly reduced, improving the identification accuracy of mixed phases in the zero-degree melt zone. The generated precipitation phase vertical profiles can be directly applied to scenarios such as short-term nowcasting weather warnings, severe weather prevention and control, transportation safety assurance, and agricultural production protection, providing accurate core data support for related work. This fully verifies the feasibility, practicality, and advancement of the technical solution of this invention, effectively compensating for the shortcomings of existing precipitation phase detection technologies and promoting the refined development of meteorological remote sensing technology.

[0057] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for vertical detection and inversion of precipitation phase state based on radio wave reflection, characterized in that, The method includes the following steps: S1. Multi-band radio wave synchronous transmission and echo signal acquisition: Through a multi-band radio transceiver system, multiple vertical sounding radio waves of different frequency bands are transmitted synchronously to acquire echo signals generated by the reflection of precipitation particles at different altitudes in the atmosphere, thus completing the continuous acquisition of echo signals at all altitudes. S2. Echo signal preprocessing and feature profile extraction: The acquired echo signals of each frequency band are preprocessed to remove system noise and interference signals. The reflectivity, differential reflectivity, phase shift and attenuation features of each frequency band corresponding to different height layers are extracted to generate continuous vertical profiles of each frequency band feature. S3. Construction of Multi-Band Scattering Feature Matrix and Preliminary Phase Identification: Based on the differences in scattering characteristics of solid, solid-liquid mixture, and liquid precipitation particles in different frequency bands, a multi-band exclusive feature matrix corresponding to different phase particles is constructed. The multi-band features of each altitude layer are matched with the feature matrix to complete the preliminary identification of precipitation phase at each altitude layer. S4. Vertical gradient analysis and phase decoupling of the melting layer: Based on the feature vertical profile, the vertical gradient change characteristics of the melting zone of the zero-degree layer are extracted. Combined with the continuous change law of the features in the melting zone, the phase intervals that are initially identified are decoupled layer by layer according to height, and the vertical boundaries of the solid, solid-liquid mixture and liquid phase intervals are clarified. S5. Phase continuity constraint and mixed phase ratio correction: The phase continuity constraint condition of adjacent height layers is introduced to correct the characteristic data in the solid-liquid mixed phase interval and calculate the proportion weight of different phase particles in the mixed phase interval. S6. Generation of vertical profiles of precipitation phases across all altitudes: Based on the decoupled phase intervals and the corrected phase proportion weights, the phase identification results of the entire altitude layer are integrated to generate continuous vertical profiles of precipitation phases across all altitudes.

2. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 1, characterized in that, In step S1, the vertically probed radio waves synchronously transmitted by the multi-band radio transceiver system include dual-polarized radio waves in the S-band, X-band, and Ka-band. The transceiver channels of each band of the multi-band radio transceiver system are synchronized using the same reference clock. The transmission beams of each band of radio waves are vertically oriented pencil beams. The main lobe width of each band beam is kept consistent. The sampling frequency and the number of sampling points of the echo signals of each band are kept uniform.

3. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 1, characterized in that, In step S1, when the echo signal of the entire altitude layer is continuously acquired, the multi-band radio transceiver system continuously and repeatedly detects at a preset time interval. The transmission power, pulse width, and pulse repetition frequency of each detection remain fixed. Each echo signal acquired by each detection corresponds to a unique detection timestamp and altitude layer calibration data. The echo signals of each frequency band within the same detection period are matched using the same altitude layer calibration system.

4. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 1, characterized in that, Step S2, when preprocessing the acquired echo signals of each frequency band, includes the following steps: Time-domain synchronization and alignment of echo signals in each frequency band are performed, and echo signals in different frequency bands are matched one-to-one according to sampling time and detection height to complete the spatiotemporal registration of multi-frequency band echo signals. Time-domain filtering and frequency-domain denoising are performed on the registered echo signal to filter out the inherent system noise and random interference signals in the echo signal. Clutter identification and removal processing is performed on the denoised echo signal to remove fixed clutter and multipath interference signals generated by non-precipitation targets in the echo signal. Beam pointing correction and propagation distance calibration are performed on the processed echo signal to establish the mapping relationship between the echo signal and the corresponding altitude layer.

5. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 1, characterized in that, Step S3, when constructing multi-band specific feature matrices corresponding to particles of different phase states and performing feature matching, includes the following steps: Based on Mie scattering theory, the scattering characteristics of solid, solid-liquid mixture, and liquid precipitation particles under different frequency bands of radio waves were simulated to obtain a simulated dataset of scattering characteristics of particles of different phases for each frequency band. Historical sounding data and field calibration test data were collected, and the scattering characteristic simulation dataset was corrected to obtain the characteristic threshold ranges of each frequency band for particles of different phases. A quantitative correlation relationship is established between features of different frequency bands. The quantitative correlation relationship is the ratio range of the same feature in different frequency bands and the Pearson correlation coefficient. A three-dimensional feature matrix is ​​constructed with phase type as the first dimension, detection frequency band as the second dimension, and feature type as the third dimension. The feature threshold range corresponding to different phase particles is used as the basic value of the matrix element, and the quantitative correlation relationship between features of different frequency bands is used as the linkage constraint term of the matrix element. The integration is completed to generate a multi-frequency band exclusive feature matrix for corresponding phase particles. A multi-band feature matching algorithm is used to calculate the matching degree between height layer features and phase features. The matching degree calculation formula is as follows: ,in Let f be the feature matching degree of the p-th phase state at height h, and f be the frequency band number. Here, δ represents the feature weighting coefficient of the f-th frequency band, and δ is the binary matching decision function. Let f be the measured characteristic value of the f-th frequency band at the h-th altitude level. This represents the characteristic threshold interval of the f-th frequency band of the p-th phase state.

6. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 1, characterized in that, Step S5, when correcting the characteristic data within the solid-liquid mixed phase range and calculating the proportion weights, includes the following steps: Based on the physical process law of precipitation particle phase transformation, a reasonable range of phase ratio changes between adjacent height layers is set to form phase continuity constraints between adjacent height layers. Based on the phase continuity constraint, constraint verification is performed on the multi-band feature data of each height layer in the solid-liquid mixed phase range, and the effective feature data that meets the constraint conditions is selected. Define the solid-state particle characteristic components and liquid-state particle characteristic components of the h-th height layer and the f-th frequency band. Based on the characteristic threshold intervals of solid and liquid phases in the multi-band dedicated characteristic matrix, and combined with the prior scattering characteristic constraints of Mie scattering theory, a dual-band constraint decoupling method is used to separate the solid-state particle characteristic components of the h-th height layer and the f-th frequency band from the effective feature data. With liquid particle characteristic components ; Based on the separated solid-state particle feature components and liquid-state particle feature components, combined with a multi-band specific feature matrix, the proportion weight of solid-state particles is calculated using a mixed-phase proportion calculation formula. The proportion calculation formula is as follows: ,in The weighting of solid precipitation particles at the h-th altitude layer is given by [the weighting]. Let f be the particle identification coefficient for the f-th frequency band. The solid particle characteristic components of the h-th height layer and the f-th frequency band are... The characteristic components of liquid particles in the f-th frequency band of the h-th height layer.

7. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 5, characterized in that, In step S3, when matching the multi-band features of each altitude layer with the multi-band exclusive feature matrix, a multi-band feature matching algorithm is used to compare the multi-band feature data extracted from each altitude layer with the feature threshold intervals corresponding to different phase particles in the feature matrix one by one, calculate the matching degree between the feature data of each altitude layer and the feature threshold intervals corresponding to different phase particles, and select the phase type with the highest matching degree as the preliminary identification result of the precipitation phase of that altitude layer.

8. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 1, characterized in that, When extracting the vertical gradient variation features of the zero-degree melting zone in step S4, based on the continuous vertical profiles of each frequency band feature, the first and second derivatives of each detection feature as a function of height are calculated. The gradient feature value is determined using the melting layer gradient feature synthesis formula, which is as follows: ,in Let f be the gradient characteristic value of the h-th height layer and f-th frequency band. The first derivative of the characteristic of the f-th frequency band with altitude. Let α be the second derivative of the feature of the f-th frequency band as a function of altitude, and let α be the gradient synthesis coefficient. The gradient synthesis coefficient α is determined by historical statistical data of the vertical profile of atmospheric temperature. Based on the calculation results, the abrupt change interval of the feature as a function of altitude is determined, and the altitude range corresponding to the abrupt change interval is defined as the zero-degree melting zone. The vertical gradient change law of each detected feature within the melting zone is extracted.

9. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 1, characterized in that, The phase continuity constraint introduced in step S5 is based on the phase transition physical process of precipitation particles falling in the atmosphere. The constraint limits the change in the proportion of particles of the same phase between two adjacent altitude layers to a preset reasonable range. The parameters of the constraint are adjusted accordingly based on the vertical profile data of atmospheric ambient temperature.

10. The method for vertical detection and inversion of precipitation phase state based on radio wave reflection according to claim 1, characterized in that, In step S6, when integrating the phase identification results of the entire altitude layer, the phase type and phase particle proportion weight of each altitude layer are integrated in order from high to low altitude. The phase data of adjacent altitude layers are smoothly connected. The generated continuous vertical profile of precipitation phase of the entire altitude layer includes three types of data: phase type, solid particle proportion weight, and liquid particle proportion weight corresponding to each altitude layer.

Images