KR20150059864A - method of hydrometeor classification using raw data of X band dual polarization radar - Google Patents

Abstract

The present invention relates to a method to classify types of hydrometeors by using a correlation coefficient (ρhv), measured propagation phase differential (Ψ_dp(r)) or characteristic of measured propagation phase differential (φ_dp), and a signal to noise ratio (SNR) as input variables input feature vector, shortly a radar measurement vector for fuzzy logic method, disclosing a method to classify hydrometers using X band dual polarization radar composed by having object determining steps, comprising steps of: obtaining a signal to noise ratio (SNR), a correlation coefficient (ρhv), and a measured propagation phase differential (Ψ_dp(r)); classifying a melting layer; calculating a melting layer height and a distribution characteristic (melting depth) after classifying the melting layer, determining a two-dimensional fuzzy function in σ(Ψ_dp)-SNR relation and 10^ρhv-SNR relation, and calculating two fuzzy values (fuzzy function values) by using them; obtaining inference values for each object by using a fuzzy value of height and the prior two fuzzy values (fuzzy function values according to σ(Ψ_dp)-SNR, 10^ρhv-SNR relations); and managing quality and classifying hydrometers by selecting a maximum inference value (Max(RS)) among inference values of each object.

Description

A method of classifying a precipitation body using x-band dual polarized radar source data using raw data of an X-band dual polarization radar

The present invention relates to a method for acquiring weather information, and more particularly, to a method for identifying precipitation-related weather information, particularly, a precipitation body (hydrometeor).

The meteorological phenomenon is closely related to life, and it is very important to accurately predict the meteorological phenomenon. Numerical forecasting systems that reflect various parameters are required to improve the prediction ability of meteorological phenomena, and various techniques for such numerical forecasting are being developed variously.

The most important source of information for numerical forecasting is radar data. The results of rainfall estimation, wind field estimation, and first time precipitation estimation using radar data are based on real-time radar data and computer To help improve forecast accuracy.

Dual polarized radar, which has recently been attracting much attention as a weather forecasting system, has many advantages in improving the accuracy of precipitation measurement, classification of precipitation type, and data quality control in comparison with conventional single polarized radar. Therefore, many research and development related thereto have been carried out in order to take advantage of these advantages.

Dual polarized radar typically transmit horizontally polarized electromagnetic waves and vertically polarized electromagnetic waves, receive back scattered signals, and analyze this signal to obtain much information related to meteorological phenomena.

 Since the precipitation particles are not exact spheres, horizontal and vertical back scattering vertical sections are not the same. Electromagnetic waves propagating through precipitation cause scattering, differential attenuation, differential phase shift, and depolarization. These signals have information to continuously estimate the size, shape, orientation, and thermodynamic phase of a steel importer during electromagnetic wave propagation.

Therefore, it is possible to improve the accuracy of precipitation estimation, radar classification, quality control by meteororogical echo and non meteororogical echo classification through radar observation. In particular, in the case of the polarization radar reflectivity observed in the Doppler radar (Z _{h, v),} the Doppler velocity, and differential reflectivity as well as the spectral width (Z _dr), linear non-polarization ratio (LDR), the differential propagation phase (φ _dp), cross The correlation coefficient (ρ _hv ), and the type (characteristics) of the object (the object is usually the precipitation body: the atmospheric water body) and the cloud physics research are possible.

The object classification is more accurate as the atmospheric body types are compared, and the polarization parameters describing the different atmospheric body types are clearly different from the observed parameter values even if they are not clearly defined or overlapped. Considering the various waiting bodies, the distinction of the waiting body becomes unclear and the logic is liable to be inaccurate. For this reason, a fuzzy logic method was used to distinguish the atmospheric bodies (Mandel, 1995)

Airborne particle identification (PID) algorithms have been developed by Vivekanandan et al. In 1999 and research on airborne body classification (PID) algorithms in X-band dual polarization radar has been studied in recent years. (Z _h ), differential reflectivity (Z _dr ), and non-uniformity (D) were measured in 2009, The study begins by dividing the precipitation type into 17 by seven input variables: equilibrium phase (K _dp ), cross correlation coefficient (ρ _hv ), temperature (T), differential reflectivity standard deviation, and differential phase standard deviation.

Here, fuzzy logic is used to classify the atmospheric bodies as mentioned above. According to the PID algorithm using fuzzy logic, each membership function and its value according to the precipitation particle or the atmospheric body are obtained for each dual polarization (radar) variable (fuzzification). Usually, this membership function becomes a fuzzy function, and the fuzzy function is obtained through experience or statistical data, and the trapezoidal function or the one-dimensional beta function . Also, the function value has a value between 0 and 1.

In addition, the membership function exists according to each double polarization variable (the threshold value of the variable) and the precipitation type in the inference step, and the weighted sum for each particle classification (Q) Can be determined by multiplying the function value (interest: P: interest value) by the variable weight (W: weight) for all the variables.

The maximum value of the weighted sum (Max (Q)) is found and this maximum value is determined as the precipitation form of the particle (defuzzyfication).

Membership functions (function values) can be defined as empirical values of the conditional probability for the value of a particular variable to which each echo input variable belongs. For example, in the example where the horizontal reflectivity (Z _h ), the differential reflectivity (Z _dr ) and the linear non-polarization ratio (LDR) observed in the radar are input variables and rainfall and hail are classified, the membership function value P Can be set to a value between 0 and 1 for six, depending on the precipitation type and each input variable.

In the next step, weight is set to the membership function value determined according to the observation value. For example, the weight of the horizontal reflectivity and the differential reflectivity may be 1 and the weight of the linear non-polarization ratio may be 0.8.

The aggregation value (Q) for the classification of rainfall and hail is obtained by multiplying the membership function values of three different precipitation rates by the respective weights, for all three variables. Thus, through these rules and reasoning processes, we can obtain a weighted sum for rainfall and hail.

If the weighted sum for rainfall is 1.7 and the weighted sum for hail is 0.5, the largest value is 1.7, so the deflation of the atmospheric water body for this area is determined by rainfall.

The membership function, which is a function applied to the dual polarization parameter, depends on the threshold value of the variable. The determination of the threshold value of this membership function is an important part of the study of queuing modeling.

For fuzzy logic that determines specific variables and membership functions and combines them to calculate a specific function value, we refer to the concept of "airborne classifications based on polarization radar observations: development of fuzzy logic and neurofuzzy systems And its verification based on the results of the proposed method, and finally, the final form of the radar measurements is presented in the form of a polynomial function of the fuzzy logic. ).

This paper also discloses a two-dimensional beta membership function (2D-MBF) that forms a product of these by using a one-dimensional beta membership function as shown in Equation 1 below.

For example, if the one-dimensional membership function for rainfall with respect to the horizontal polarization reflectivity Z _h is f _{rain - zh} , then it can be expressed by the following Equation 2,

When Z _h and Z _dr are treated as independent variables for rainfall, the two-dimensional membership function in the two-dimensional plane of Z _h - Z _dr can be expressed as Equation 3, Can be.

On the other hand, the differential polarization phase (φ _dp ) and the cross correlation coefficient (ρ _hv ), which are dual polarization parameters or parameters, can be obtained by using the following series of equations.

First, the microphysical properties of the rain medium can be described by a drop size distribution (DSD).

To study the shape of DSD in varying rainfall rates, the natural variation of DSD can be expressed by a normalized (normalized) gamma function such as:

Where D ₀ is the equivolumetric median volume diameter in mm, μ is the shape parameter, N _W is the normalized parameter of the exponential distribution with the same water content and D ⁰ (normalized intercept parameter of the exponential distribution).

The radar observations in the precipitation body can be expressed in terms of DSD, and the reflectivity Z _{h , v} in horizontal and vertical polarization can be defined by the following equation.

Where λ is the radar wavelength, σ _{h, v} is the radar reflection cross section in horizontal or vertical polarizations, Kw is the dielectric factor of water and ε _r is the complex dielectric constant of water Kw = (? _R- 1) / (? _R + 2).

The differential reflectivity (Z _dr ) is defined as the ratio of reflectivity factors in the horizontal and vertical polarizations as:

And, _(to K, unit deg / km) non-phase difference and so on can be defined by the following equation.

Where R is the real part of the complex number, and f _h and f _v are the forward scattering magnitudes in the horizontal and vertical polarizations.

Room measurement of dual polarized radar rainfall level is the reflectivity ^{_{(Z h: mm 6 / m}} 3), differential reflectivity (Z _dr:: dB), and so on odds phase (deg / km _in K).

The differential propagation phase? _Dp between the target positions r1 and r2 is defined using non-equal phase in the following equation.

The cross correlation coefficient (rho _hv ) represents the correlation between signals at the horizontal and vertical polarizations and is defined by the following equations.

In this case, S _hh and S _vv mean the relationship between the horizontally received electric field when horizontally transmitted and the vertically received electric field when vertically transmitting, in a backscattering matrix (Zrnic, 1991) The struisk symbol is a complex conjugate, meaning that there is a 90 ° difference in phase, and the bracket symbol means the expectation element. The size of these variables is determined by the radar propagation frequency and the atmospheric particle size, shape, and configuration (state).

The differential propagation phase is proportional to the forward scattering characteristics of the vortex due to the difference between the horizontal and vertical propagation phases. In horizontal deflection aids such as raindrops, horizontal propagation phase shift is larger than vertical propagation phase shift. Moreover, in the non-magnetic echo, the fluctuation of the differential propagation phase is clearly larger than the fluctuation in the precipitation due to a weak correlation between the signals of the horizontal and vertical polarizations.

The cross correlation coefficient is affected by the variation of the horizontal to vertical ratio of individual bodies. The value of the cross correlation coefficient is close to 1 for rain or initial ice crystal. In the case of melting snow or mixed state, the cross correlation coefficient is less than 1. A low value of the cross correlation coefficient can be used to detect contamination by hail or mixed state rainfall, terrestrial clutter, and non-tropic scattering.

Dual polarized radar can be divided into S band, C band, and X band according to the electromagnetic wave to be used. In the case of X band radar, the damping due to precipitation is severe and the accurate attenuation correction for attenuation can be used as the polarization radar data .

Korean Patent No. 10-1221773: Method of Classifying Weather and Non-Gaseous Echoes Using Dual Polarized Radar. Korean Patent No. 10-0931950: Method of Classifying Precipitation Type of Weather Radar. Korean Registered Patent No. 10-1131194: Local Precipitation Radar System Using Ship X-Band Radar Network.

Radar Utilization Technology Research (Ⅰ): 2007, National Institute of Meteorological Research. al. Estimation of Rainfall in X - Band Dual - Polarized Radar: The Korean Institute of Civil Engineers. 2009 Intensive observations of dual polarizers: atmospheric aerosol classification; Korea Meteorological Society Autumn Conference, National Meteorological Research Institute Suk Mi Kyung, Nam Kyung Yeop, Cho Chun-ho Airborne Classification Based on Polarized Radar Observation: Development of Fuzzy Logic and Neuro-Fuzzy System and Its Verification (JURANAL OF ATMOSPHERIC) AND OCEANIC TECHNOLOGY Volume 17, Hongping Liu and V. Chandrasekar, final form 27 April 1999.)

The present invention relates to an X-band dual-polarized radar system for use with an integrated rainfall information processing system including an attenuated X-band dual-polarized radar to distinguish an atmospheric body by using raw observation data of an X- And a method thereof.

An object of the present invention is to distinguish an atmospheric body (precipitation body) using an X-band dual-polarized radar, and to provide necessary weather information such as precipitation and the like based thereon.

X-band of the present invention for achieving the above object, a dual-polarization radar rainfall body nine minutes way, that especially works well in the high frequency region as in the X band, the correlation coefficient (cross-correlation coefficient; ρ _hv), the measured differential propagation The characteristics of the phase (Ψ _dp (r)) or the differential propagation phase (φ _dp ) and the SNR are used as input variables (input feature vector, ie radar observation vector for the fuzzy logic technique) In order to be able to distinguish kinds,

(SNR), correlation coefficient (ρ _hv ) and measured differential propagation phase (Ψ _dp (r)) as initial input data (variables) through radar electromagnetic wave transmission and reception data, (Ψ _dp ) -SNR relationship and a 10 ^{ρ hv-} SNR relationship, and then, by using this, the melting depth, A fuzzyfication step of calculating two fuzzy values (a fuzzy function value), a step of calculating the fuzzy values of the heights and the two preceding two-dimensional fuzzy values (? ( _Dp ) -SNR, 10 ^? Hv- (RS)) of the inference values for each object, and a step of determining the object to classify the quality control and the atmospheric body by selecting the maximum inference value (Max (RS)) among the inference values for each object Respectively.

At this time, in obtaining the signal-to-noise ratio (SNR), the correlation coefficient (ρ _hv ) and the measured differential propagation phase (Ψ _dp ) as initial input data through the radar electromagnetic wave transmission and reception, the correlation coefficient (ρ _hv ) 10 ^{ρ hv} , and the standard deviation σ (Ψ _dp (r)) of the measured differential propagation phase is used as the input data of the fuzzy system. The deviation of the measured differential propagation phase is obtained by the following general formula as well known in the art.

Where N is the number of samples or the number of gates of the target section for which the deviation is sought.

The measured differential propagation phase (Ψ _dp ) has a relationship with the differential propagation phase (φ _dp ) as shown in the following equation.

and δ is the backscatter phase, which is the arguments of the complex backscattering amplitudes of the horizontal and vertical polarizations' complex backscattering amplitudes. Therefore, when the back scattering is large, it is preferable to use the measured differential propagation phase.

In the step of dividing the melting layer, for example, using the distinction of the boundary regions of the atmospheric water bodies in the σ (Ψ _dp ) -SNR plane and the 10 ^{ρ hv-} SNR plane as shown in FIGS. 1 and 2, By identifying the region of the upper plane, the melting layer can be identified, and the average height of the melting layer can be determined using the measurement data belonging to the melting layer region. That is, it is determined in consideration of the signal-to-noise ratio (SNR) to identify the melting layer.

After dividing the melting layer, two fuzzy function values are calculated by using the same σ (Ψ _dp ) -SNR relation and the 10 ^{ρ hv-} SNR relation as the boundary sections in the above figures. For the determination of the two-dimensional fuzzy function, we can use the boundaries of the zone separation by the atmospheric body in the σ (Ψ _dp ) -SNR plane and the 10 ^{ρ hv-} SNR plane.

Height (Height) fuzzy value f (H) and two two-dimensional spread function value _{(σ (Ψ dp) -SNR,} 10 beta membership function value according to the ^hv -SNR ^ρ) the number of each object or each air by using the upper body of the For reason, the inference value is obtained by the inference rule as the following expression.

Finally, in the step of selecting the maximum speculation value (Max (RS)) and classifying the quality control and the precipitation object, the maximum speculation value for the atmospheric water body is checked and the corresponding atmospheric water body (precipitation object) is selected .

According to the present invention, in the transmission and reception data of a dual polarized radar, raw data are directly used without attenuation correction to obtain necessary double polarized parameters. The signal-to-noise ratio is included in these dual polarized parameters, The membership function of the form is selected, the membership function value is obtained based on these variables and the membership function, the function value is processed by the predetermined processing method or rule, and the result is sorted and compared, (Precipitation body) can be classified.

FIGS. 1 and 2 are graphs showing SNR, correlation coefficient, SNR and standard deviation of measured differential phase by HYDROX radar data,
Figs. 3 and 4 are data based on Figs. 1 and 2, showing the SNR and correlation coefficient of each radar measurement data, the SNR and the standard deviation of the measured differential propagation phase,
FIGS. 5A to 5D are radar observation data obtained at a specific point in a specific region in which the research is performed. The radar data are transmitted by raising the elevation angle at a predetermined azimuth angle around the radar, A graph showing the horizontal reflectance at the position, the measured differential propagation phase, the cross correlation coefficient, and the atmospheric water class classified by these bases in different colors,
6 is a flow chart showing the flow of the method of the present invention,
Figure 7 is a graph illustrating an example of a two-dimensional membership beta function that may be used in the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

The present invention is based on data obtained from the dual-polarized X-band radar (HYDROX) for hydro-meteorological studies during the 2005-2006 season of the hydrological meteorological test bench (HMT) in the NOAA Earth System Research Laboratory (ESRL) data).

In the present invention, a new signal-to-noise ratio (hereinafter referred to as SNR) is used as an input variable in comparison with the conventional method using variables affected by attenuation such as horizontal reflectivity and differential reflectance for separating a water body (atmospheric water body). The reason for using the SNR is to discriminate the variability in the dual polarization parameters due to the change in the SNR. For example, in a rainfall region having a high SNR value, the correlation coefficient is higher than the correlation coefficient in the mixed-state region such as the snowfall region, and is generally a large value close to unity.

However, as can be seen from the data in FIG. 3, at low SNR, the correlation coefficient in the rainfall can be actually smaller than the correlation coefficient in the melting precipitation (mixed precipitation), so that the correlation coefficient at, for example, 10 decibels 0.93, while the correlation coefficient in the melting precipitation region with an SNR of 35 decibels may be 0.95.

Fig. 3 shows the distribution of each observation result in the correlation coefficient plane and the SNR in a period in which laminar flow and convective rainfall having a bright band are mixed. The blue dot means all data above 2.5 km above sea level, and includes rain, snow and mixed precipitation. The red dot means only the data for the precipitation area (between 0.8km and 2.5km above sea level). The data in FIG. 3 show that there is a significant overlap in the distribution of the categories of precipitation form for all types of precipitation currecies, indicating that searching for the melting bed using a constant correlation coefficient is not appropriate for low SNR it means.

Therefore, by combining the SNR with the correlation coefficient, it is possible to distinguish rainfall from mixed precipitation over the entire SNR region, and it is necessary for accurate melting layer judgment.

In this way the S-band (2 ~ 4GHz) in the X band (8 ~ 12GHz) relatively, but can be operated at high frequencies, the correlation coefficient in the X band (ρ _hv) and the differential propagation phase (φ _dp) to the more Sensitive, so it is more useful at relatively high frequency bands such as the X band. For example, _(in K: deg / km) phase, such as odds that these values are reflected in the value of the X-band is largely up to three times the value at S-band.

Over the past decade, a number of atmospheric aerodynamic segregation methods based on dual-polar radar parameters have been developed, mainly developed for S-band or C-band frequencies. Recently, a method using dual polarization observation characteristic specific to X band has been proposed. However, the present invention proposes a more robust airborne object classification method that can be applied in relatively high frequency bands such as X-bands.

6 is a flow chart schematically illustrating the sequence in which an embodiment of the method of the present invention is performed.

According to Fig. 6, this classification method obtains the signal-to-noise ratio (SNR), the correlation coefficient (rho _hv ) and the differential propagation phase ( _dp ) through radar measurement and inputs it as an input element, That is, non meteororogical objects, rain, rain / ice mix, and snow particles.

After the input element is inserted, a melting layer (ML) is detected and a fuzzyfication process is performed. In the fuzzification process, a 2D beta membership function based on the relationship between σ (φ _dp ) and SNR, a 2D beta membership function based on the relationship between 10 ^{ρ hv} and SNR, a 1D beta membership function for altitude (H) And the function value for the input element is derived.

If we derive the function values for these three beta membership functions, we derive the respective inference values for rainfall, rainfall / ice mixing, and ice according to the predetermined rules, (Non-meteorological object discrimination) by comparing the measured value with a preset constant value, and a defuzzification process of confirming the atmospheric water body having the maximum value is performed.

In the context of the present invention, the reason for using the signal-to-noise ratio as an input element has already been mentioned above.

The characteristics of the measured differential propagation phase (Ψ _dp ) are efficient in distinguishing between precipitation, terrestrial clutter, abnormal propagation, and biological scattering. This property value can also be extended to detect dissolved particles. The characteristic information can be obtained by using the root mean square value (RMS) of the difference of the radar observations at the azimuth and distance as shown in the following equation.

Where r represents the location in the target area and j varies here over the range of the gate length, N is the number of resolution gates in the range, The number of measured samples for acquisition is indicated. The gate is a unit that becomes a basic unit when processing the radar signals received from arbitrary azimuths by unit distance. Since the radar signals are present one at a certain distance starting from the radar, Of the pixel of interest. The more gates in the same distance range for the same target range, the more accurate the data with higher resolution can be accumulated, but there is a limit to increase the processing capacity. For example, seven gates with a length of 150 meters (meters) are used.

The proposed method also has four basic respects: melting layer detection, fuzzification, inference, defuzzification, and fuzzy logic, The process of deriving a quantifiable result within the membership degrees.

With respect to melting layer detection, the correlation coefficient drops sharply as the radar beam passes from the lower rainfall zone to the upper melting zone, and this information can be used to determine the altitude and depth of the melting zone. Usually, the upper part of the 'light layer' is judged to be ice, the bright layer area is mixed with the snow, and the bright layer is judged by the rain. This approach works well for high SNR.

As already mentioned, correlation coefficients at low SNR values are sensitive to SNR values. Even in the rainfall region, there may be a correlation coefficient value such as 0.9, which is smaller than the correlation coefficient value of 0.93 in mixed precipitation with respect to a high SNR value. The display of the melting layer using a constant correlation coefficient criterion does not work properly at low SNRs.

In order to overcome this difficulty, the present invention proposes the following improved melting layer detection method. That is, the melting layer confirms which boundary region each observation data belongs to in the 10 ^{ρ hv-} SNR relationship in FIG. 1 and in the σ (Ψ _dp ) -SNR plane in FIG. The average altitude and depth of the melting layer (mixed precipitation area) can be obtained by examining the altitude of these at the observation data belonging to the mixed precipitation area and taking an average value thereof. 1 and 2 to be used here will be established through an investigation on a plurality of measurement results in advance.

The average height of the detected melting layer is used as the environmental element (temperature) in the cruciform classification system in the subsequent process.

With respect to fuzzification, another improvement of the airborne body classification of the present invention is to use two two-dimensional beta membership functions (see Liu and Chandrasekar's paper) by using σ (Ψ _dp ), SNR, 10 ^{ρ hv} variables, Lt; / RTI >

FIGS. 1 and 2 show a two-dimensional beta membership function derived from the above-mentioned NOAA dual polarized X-band radar data (HYDROX data). In particular, Fig. 2 shows sigma ([Delta] _dp ) as a function of SNR and Fig. 1 shows the corresponding relationship between ^{10rho hv} and SNR. The classification boundary is obtained with HYDROX radar data. The process involves estimating the mean value of σ (Ψ _dp ) and 10 ^{ρ hv} according to the SNR and obtaining the optimum relationship according to the data.

It should be noted, however, that these two-dimensional beta membership functions depend on the characteristics of the radar system, such as signal variations. Therefore, the adjustment of the two-dimensional beta membership function can be made for each radar system, and it is necessary to adjust it. That is, as in the conventional fuzzy logic, it should be considered that the fuzzy function does not consist of a single, commonly used case.

The subsequent rule strength step is determined by the following equation in this embodiment.

Where f is the fuzzy function and H is the measured altitude. As the fuzzification function, a one-dimensional membership function using the beta function as in Equation (1) and the altitude as the variable (x) can be used. If a one-dimensional membership function is created according to the height of the melting layer, a fuzzy function value can be obtained by applying a, b, and m according to the specific altitude and precipitation classification of the individual data.

f (2D-MBF1) and f (2D-MBF2) are two-dimensional beta membership functions for the σ (Ψ _dp ) -SNR relationship and the 10 ^{ρ hv-} SNR relation respectively.

In the defuzzification phase, if the maximum estimate is lower than 0.2, the object at that location is determined to be a non-meteorological or low signal object. When the maximum estimated value is higher than 0.2, it is determined that one of the three atmospheric circulations of rainfall, ice / rain mixture, and ice has the maximum estimated value. The reference value of 0.2 is obtained empirically.

It can be verified through the existing data whether the method of classifying the atmospheric body is valid.

Figures 5A-5C depict the horizontal reflectivity (Zh), the correlation coefficient (rho _hv ) and the measured differential propagation phase (Ψ _dp ) obtained from one meteorological observation. This data was taken during periods such as the convective and stratified coupled precipitation periods shown in FIGS. 3 and 4, which include rainfall below 2.8 km above sea level and snowfall above this altitude. HYDROX data and radio wave observations near the S band, which are oriented vertically, indicate the presence of a significant melting layer of approximately 2.8 km above sea level.

This fact provides a perfect opportunity to test the method of sculpture of the present invention. The result of the classification from the proposed method is shown in Fig. 5D. The melting layer detected by the black lines of each panel was compared with observation data obtained from the X-band radar in the range of approximately 18 km and S band propagation observation. It can be seen that the melting regions observed in the S band and the X band agree well. Based on the properties of the melting layer, the classification results shown in FIG. 5D are appropriate.

While the present invention has been described with reference to exemplary embodiments, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. That is, those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims.

Claims (7)

(SNR), correlation coefficient (ρ _hv ) and measured differential propagation phase (Ψ _dp (r)) as the initial input variables through the electromagnetic wave transmission and reception observation data of the X-band dual polarization radar,
(Melting precipitation layer) and calculating altitude and distribution characteristics (melting depth) of the melting layer in the atmosphere,
A standard deviation σ (Ψ _dp ) and 10 ^{ρ hv} of the differential propagation phase measured using the input variable are obtained, a 2-dimensional fuzzy function indicating a σ (Ψ _dp ) -SNR relationship and a 10 ^{ρ hv-} SNR relation is determined, A fuzzyfication step of calculating two fuzzy function values using a two-dimensional fuzzy function,
An inference step of obtaining an inference value (RS) using the fuzzy function value of the melting layer height and the two two-dimensional fuzzy function values for each atmospheric water body,
The quality control that selects non-meteorological objects by selecting the maximum reasoning value (Max (RS)) among the reasoning values obtained in the above reasoning step, and the defuzzification method which classifies precipitation by three kinds of rainfall, The method comprising the steps of: (a) inputting at least one of a plurality of radar signals to a radar system; The method according to claim 1,
Wherein the inference step uses the following equation as an inference rule to obtain an inference value.

F (2D-MBF1) and f (2D-MBF2) are two-dimensional beta memberships for the σ (Ψ _dp ) -SNR relationship and the 10 ^{ρ hv-} SNR relationship. Function. The method according to claim 1 or 2, wherein
The SNR, the correlation coefficient (ρ _hv ), and the correlation coefficient (ρ _hv ) are used as input parameters in the step of calculating the altitude and distribution characteristics (melting depth) of the melting layer The individual observed data obtained from the measured differential propagation phase (Ψ _dp (r)) are located in the boundary regions determined for the separation of atmospheric bodies in the σ (Ψ _dp ) -SNR plane and the 10 ^{ρ hv-} SNR plane Wherein the average height and depth of the melting layer are calculated from the observation data belonging to the melting layer region and the average height and depth of the melting layer are calculated from the observation data belonging to the melting layer region. (SNR), correlation coefficient (ρ _hv ) and differential propagation phase (φ _dp ) as initial input variables through the electromagnetic wave transmission and reception data of the X-band dual polarization radar,
(Melting precipitation layer) and calculating altitude and distribution characteristics (melting depth) of the melting layer in the atmosphere,
The standard deviations σ (Ψ _dp ) and 10 ^{ρ hv} of the differential propagation phase measured using the general relationship between the input variable, the differential propagation phase and the measured differential propagation phase are obtained, and the σ (Ψ _dp ) -SNR relation and the 10 ^a fuzzyfication step of determining a two-dimensional fuzzy function expressing the relation ^{ρ hv-} SNR and calculating two fuzzy function values using the two-dimensional fuzzy function,
An inference step of obtaining a reasoning value using the fuzzy value of the melting layer height after the purging step and the two two-dimensional fuzzy function values,
And selecting the maximum speculation value (Max (RS)) among the speculation values obtained in the speculation step to classify the quality control and the precipitation object. 5. The method of claim 4,
Wherein the inference step uses the following equation as an inference rule to obtain an inference value.

F (2D-MBF1) and f (2D-MBF2) are two-dimensional beta memberships for the σ (Ψ _dp ) -SNR relationship and the 10 ^{ρ hv-} SNR relationship. Function. The method according to claim 4 or 5,
The SNR, the correlation coefficient (ρ _hv ), and the correlation coefficient (ρ _hv ) are used as input parameters in the step of calculating the altitude and distribution characteristics (melting depth) of the melting layer The individual observed data obtained from the measured differential propagation phase (Ψ _dp (r)) are located in the boundary regions determined for the separation of atmospheric bodies in the σ (Ψ _dp ) -SNR plane and the 10 ^{ρ hv-} SNR plane Wherein the average height and depth of the melting layer are calculated from the observation data belonging to the melting layer region and the average height and depth of the melting layer are calculated from the observation data belonging to the melting layer region. A method for distinguishing objects using initial input variables and fuzzy logic obtained from electromagnetic wave transmission and reception data of an X-band dual polarized radar,
(SNR), a correlation coefficient (rho _hv ), a differential propagation phase (? _Dp ) or a measured differential propagation phase (? _Dp (r)) is used as the initial input parameter,
Wherein the object is classified into rain, rain, ice, and ice as non-meteorological objects and atmospheric water bodies, using the x-band dual polarized radar source data.

Images