A raindrop spectrum parameter profile inversion method and system

Abstract

The application discloses a raindrop spectrum parameter profile inversion method and system, and aims at the problem that an existing precipitation microphysical parameterization scheme is insufficient in representing regional differences and vertical evolution characteristics of a raindrop spectrum, and the method is characterized in that the inversion process of the raindrop spectrum parameter profile is optimized by fusing precipitation particle phase classification, and the precision and reliability of the inversion result are improved. According to the method, multi-source vertical detection data are used, a small-particle detection micro rain radar and a laser raindrop spectrometer are used for cooperative inversion, the vertical gradient change rule of the raindrop spectrum parameter is quantified, the mu-Lambda relationship at different heights is constructed, and the raindrop spectrum parameter profile inversion of the precipitation particle phase classification is realized.

Description

Technical Field

[0001] This invention belongs to the field of precipitation inversion technology, and particularly relates to a method and system for inverting raindrop spectral parameter profiles. Background Technology

[0002] As a core parameter describing the microphysical processes of precipitation, the raindrop spectrum directly determines the intensity and evolution of precipitation, and is key to improving cloud microphysical parameterization schemes and enhancing precipitation forecast accuracy. Currently, raindrop spectrum distribution models are generally used to describe the distribution of the raindrop spectrum.

[0003] Influenced by factors such as climate, topography, and atmospheric environment, raindrop spectra exhibit significant regional variability. Most parameterization schemes employ fixed parameter ranges, making it difficult to accurately reflect precipitation characteristics across different regions. Furthermore, existing parameterization schemes are largely based on theoretical assumptions or horizontal observation data, failing to adequately consider the microphysical processes of precipitation particles in the vertical direction, such as collision, fragmentation, and evaporation. Differences in raindrop spectrum distribution at different altitudes lead to variations in the μ–Λ relationship (the relationship between the shape and slope parameters of the distribution model) with altitude, and this variation is more pronounced during heavy precipitation. However, existing schemes, based on horizontal observation data, do not quantify the microphysical processes of precipitation particles in the vertical direction, and generally adopt a uniform μ–Λ relationship based on the ground, resulting in significant model bias in precipitation simulation. Additionally, traditional schemes rely on weather radar or single-source ground raindrop spectrum data, lacking the ability for multi-source vertical observation collaborative inversion, further increasing errors. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention proposes a method and system for inverting raindrop spectral parameter profiles.

[0005] The technical solution of the present invention is as follows:

[0006] A method for inverting raindrop spectral parameter profiles, wherein the raindrop spectral parameters include shape parameters and slope parameters of a distribution model characterizing the raindrop spectral distribution, including:

[0007] The micro-rain radar base data and laser raindrop spectral base data after quality control are spatiotemporally stitched together to form a raindrop spectral fusion profile with several height layers. In the raindrop spectral fusion profile, one height layer near the ground is composed of laser raindrop spectral base data, and multiple height layers in the air are composed of micro-rain radar base data.

[0008] The raindrop spectrum fusion profile was subjected to ground raindrop spectrum inversion considering the phase of ground hydrophobic material and airborne raindrop spectrum inversion considering the phase of airborne hydrophobic material, respectively.

[0009] Raindrop spectral parameters are inverted at different altitudes from the ground raindrop spectrum and the air raindrop spectrum to obtain raindrop spectral parameter profiles.

[0010] Furthermore, the quality control of the light rain radar base data includes: preprocessing, first light rain radar data removal, and second light rain radar data removal.

[0011] The preprocessing specifically includes noise removal, microwave attenuation correction, Mie scattering correction, air density correction, hydrophobic deformation correction, and hydrophobic terminal velocity correction based on the micro-rain radar standard inversion algorithm.

[0012] The first light rain radar data removal includes removing light rain radar base data corresponding to the first altitude layer near the ground and the highest altitude layer in the air;

[0013] The second light rain radar data removal includes removing Doppler velocity ambiguity data and data with a precipitation rate lower than a set precipitation rate threshold per unit time.

[0014] Furthermore, the quality control of the laser raindrop spectral base data includes a first laser raindrop spectral base data removal and a second laser raindrop spectral base data removal.

[0015] The first laser raindrop spectral base data removal includes removing data where the total number of raindrops per unit time is less than a set total raindrop number threshold or the precipitation rate is less than a set precipitation rate threshold;

[0016] The second laser raindrop spectral base data elimination includes eliminating the number of particles in the first two diameter levels of the laser raindrop spectral base data and eliminating data that does not satisfy the ±40% theoretical velocity-diameter relationship.

[0017] Furthermore, the specific method for spatiotemporally stitching the quality-controlled micro-rain radar base data and laser raindrop spectral base data includes:

[0018] The temporal stitching includes: aligning the timestamps of the micro-rain radar base data and the laser raindrop spectral base data. If the timestamps of the micro-rain radar base data and the laser raindrop spectral base data are not synchronized, a linear interpolation method is used to match the micro-rain radar base data to the sampling time point of the laser raindrop spectral base data.

[0019] Spatial stitching includes: a near-ground altitude layer using laser raindrop spectral data, multiple altitude layers in the air using micro-rain radar-based data, and missing data at different altitude layers supplemented using cubic spline interpolation.

[0020] Furthermore, the specific method for inverting the ground raindrop spectrum considering the phase state of ground hydrophobic substances includes:

[0021] 1) Ground water condensate is classified into five phases: drizzle, rain, snow, hail, and mixed phases.

[0022] Based on the observed diameter and terminal velocity data of water condensate from the laser raindrop spectral base data, the measured discrete curve of diameter-terminal velocity was obtained.

[0023] The measured discrete curves are divided into several diameter ranges according to a pre-set droplet diameter-velocity classification standard. The measured velocities of the five types of ground water condensate phases corresponding to each diameter range are further calculated.

[0024] Based on the empirical relationship between diameter and final velocity corresponding to the five types of ground hydrate phases, the theoretical velocity corresponding to each diameter range for the five types of ground hydrate phases is calculated.

[0025] Calculate the discrete Friesian distance between the measured velocity and the theoretical velocity corresponding to the same ground hydrogel phase for each diameter range, and take the hydrogel phase corresponding to the minimum discrete Friesian distance as the hydrogel phase for that diameter range.

[0026] 2) Ground raindrop spectrum inversion based on ground hydrophobic phase:

[0027] Based on the empirical diameter-terminal velocity curves for particles of different phases, the deviation between the observed diameter-terminal velocity and the corresponding empirical curves for particles of different phases is calculated. Precipitation particles with a deviation exceeding 60% are removed. The formula for calculating the particle number concentration per unit volume is as follows:

[0028]

[0029] In the formula, For the diameter at the th Number concentration of hydrogel in the range; This represents the total number of diameter ranges. This refers to the serial number of the diameter range; For the diameter at the th Within the range and the final velocity of the fall is at the 1st The number of particles within the range is The sampling area of ​​the instrument; Sampling time; Indicates the first The velocity of the particle at the end of its fall corresponding to the gear. This refers to the width of the diameter gauge.

[0030] Furthermore, the specific method for inverting the aerial raindrop spectrum considering the phase state of aerial hydrophobic condensates includes:

[0031] 1) First, read the raw reflectivity spectrum data observed in the micro-rain radar base data. Introducing speed The reflectance spectral density of the function:

[0032]

[0033] in, Distance-based grading ; Divide the Doppler velocity into gears. ; For intermediate parameters, , The frequency resolution of the Doppler spectrum; This is the wavelength for light rain radar;

[0034] Then analyze the original reflectance spectrum data Perform signal enhancement processing, peak signal determination, and Doppler velocity deblurring processing;

[0035] Then calculate the equivalent radar reflectivity factor. The average falling velocity of particles after deblurring Doppler spectral width after deblurring skewness after deblurring :

[0036]

[0037]

[0038]

[0039]

[0040] In the formula, Where is the dielectric constant. , The complex refractive index of water, For the i-th Doppler velocity detected by the micro-rain radar, ;

[0041] Then calculate the equivalent radar reflectivity factor. The vertical gradient, combined with the skewness after deblurring. If a bright band exists, calculate its top height. Bright and high bottom ;

[0042] Then calculate the average velocity when the phase is rain. and the average velocity when the phase is snow , , ;

[0043] The following discrimination method was used to classify the phase states of aerial hydrogels at different altitudes:

[0044] Will & As the first judgment condition, and & As the second judgment condition, and will & As a third condition for judgment;

[0045] Bright band bottom height Existence serves as the fourth condition for judgment.

[0046] Will As the fifth judgment condition, and will & As the sixth condition for judgment, and will As the seventh condition for judgment;

[0047] Will As the eighth judgment condition, and will & As the ninth judgment condition; This represents the difference in equivalent radar reflectivity factor between different altitude levels;

[0048] For atmospheric condensates at altitudes that simultaneously meet the first, fourth, and fifth criteria, the phase is drizzle, rain, or hail.

[0049] For atmospheric condensates at altitudes that simultaneously meet the second, fourth, and fifth judgment conditions but do not meet the first judgment condition, the phase of the condensate is drizzle, rain, or hail.

[0050] For airborne hydrophobic substances at altitudes that simultaneously meet the third and fourth criteria but do not simultaneously meet the first, second, and seventh criteria, the phase of the hydrophobic substance is drizzle, rain, or hail.

[0051] For airborne hydrogel phases at altitudes that meet the third judgment condition but do not meet the first, second, and fourth judgment conditions, the phase is drizzle, rain, or hail.

[0052] Further classification methods for phases such as drizzle, rain, or hail include:

[0053] The atmospheric condensate phase at the altitude level that meets the eighth judgment condition is hail;

[0054] For atmospheric water condensate at altitudes that do not meet the eighth judgment condition but meet the ninth judgment condition, the phase is drizzle.

[0055] For atmospheric water condensate at altitudes that do not simultaneously meet the eighth and ninth criteria, the phase state is rain.

[0056] The phase state of airborne hydrogel at altitudes that do not simultaneously meet the first, second, and third criteria is uncertain.

[0057] For airborne hydrogel phases at altitudes that simultaneously meet the first and sixth judgment conditions but do not meet the fourth judgment condition, the phase state is a mixed phase.

[0058] The airborne hydrogel phase at a height level that simultaneously meets the second and sixth judgment conditions but does not simultaneously meet the first and fourth judgment conditions is a mixed phase.

[0059] The airborne hydrophobic phase of a height layer that simultaneously meets the third, fourth, seventh, and sixth judgment conditions but does not simultaneously meet the first and second judgment conditions is a mixed phase.

[0060] For airborne hydrogel phases at altitudes that satisfy the first judgment condition but do not simultaneously satisfy the fourth and sixth judgment conditions, the phase of the hydrogel is snow.

[0061] For airborne hydrogel phases at altitudes that satisfy the second judgment condition but do not simultaneously satisfy the first, fourth, and sixth judgment conditions, the phase is snow.

[0062] The atmospheric water condensate phase at a height level that simultaneously meets the third, fourth, and seventh judgment conditions but does not simultaneously meet the first, second, and sixth judgment conditions is snow;

[0063] 2) Acquire historical ground layer data at one near-surface altitude and historical vertical layer data at multiple airborne altitudes; both the ground layer data and the vertical layer data include several characteristic variables and PSDs (Particle Spectrum Distributions) of different phase states; the characteristic variables include radar reflectivity factor. The final velocity of the falling particles Doppler spectral width Precipitation rate R, precipitation particle phase Ph; the precipitation particle phase includes drizzle, rain, snow, hail and mixed phase;

[0064] The historical ground layer data and historical vertical layer data are stitched together according to height layer to form a feature variable matrix and a raindrop spectrum matrix;

[0065] The feature variable matrix is ​​represented as follows:

[0066]

[0067] In the formula, For the first Radar reflectivity factor at higher altitudes; For the first The terminal velocity of particles falling at higher altitudes; For the first Doppler spectral width at higher altitudes; For the first Precipitation rate at higher altitudes; For the first Phase of precipitation particles at higher altitudes;

[0068] The raindrop spectral matrix is ​​represented as follows: ;

[0069] In the formula, For the first Distribution of different phase particle spectra in the height layer:

[0070] A neural network model for aerial raindrop spectrum inversion is constructed, using the feature variable matrix as input data and the raindrop spectrum matrix as output data to train the neural network model.

[0071] The feature variable matrix obtained from the target ground layer data and vertical layer data is input into the trained neural network model to obtain the corresponding raindrop spectrum matrix. From this, the shape parameters and slope parameters of multiple altitude layers in the air, as well as the distribution of particle spectrum in different phases, are obtained from the raindrop spectrum matrix.

[0072] Furthermore, the specific method for inverting the raindrop spectral parameters at different altitudes from the ground raindrop spectrum and the airborne raindrop spectrum to obtain the raindrop spectral parameter profile is as follows:

[0073] Based on the ground raindrop spectrum and the airborne raindrop spectrum, through The step-moment method is used to calculate the raindrop spectral parameter profiles at different altitudes, including the slope parameter Λ and the shape parameter of the raindrop spectrum. ,in, Step moment is defined as

[0074]

[0075]

[0076] In the formula, and These are the lower and upper limits of the raindrop diameter, respectively; For concentration parameters, For the slope parameter, For shape parameters;

[0077] Then, using the second, third, and fourth moments, calculate the slope parameter Λ and shape parameter of the raindrop spectrum as follows. :

[0078]

[0079]

[0080] ;

[0081] In the formula, This is an intermediate parameter.

[0082] Furthermore, the method also includes: classifying precipitation periods into stratiform cloud precipitation, convective precipitation, and weak precipitation based on the laser raindrop spectral base data;

[0083] Raindrop spectrum parameters are inverted at different altitudes for the ground raindrop spectrum and the air raindrop spectrum according to the precipitation category corresponding to the precipitation period, to obtain raindrop spectrum parameter profiles; raindrop spectrum parameters are inverted at different altitudes for the ground raindrop spectrum and the air raindrop spectrum according to the precipitation category corresponding to the precipitation period, to obtain raindrop spectrum parameter profiles.

[0084] Furthermore, the precipitation category classification includes: classifying precipitation categories based on the surface precipitation rate R and its time series retrieved from laser raindrop spectra.

[0085] For precipitation rate observation data with a 1-minute time resolution, if the precipitation rate is greater than 0.5 mm / h for a period of more than 10 minutes and the standard deviation of the precipitation rate is less than 1.5 mm / h during that period, it is identified as stratiform cloud precipitation.

[0086] If the precipitation rate is greater than 5 mm / h and the standard deviation of the precipitation rate is greater than 1.5 mm / h, it is identified as convective precipitation.

[0087] Precipitation samples with a precipitation rate below 0.5 mm / h are identified as weak precipitation.

[0088] A raindrop spectral parameter profile inversion system, wherein the raindrop spectral parameters include shape parameters and slope parameters of a distribution model used to characterize the raindrop spectral distribution, and includes a spatiotemporal stitching module, a raindrop spectral inversion module, and a raindrop spectral parameter profile inversion module;

[0089] The spatiotemporal stitching module is used to spatiotemporally stitch together the quality-controlled micro-rain radar base data and the laser raindrop spectrum base data to form a raindrop spectrum fusion profile with several height layers. In the raindrop spectrum fusion profile, one height layer near the ground is composed of laser raindrop spectrum base data, and multiple height layers in the air are composed of micro-rain radar base data.

[0090] The raindrop spectrum inversion module is used to perform ground raindrop spectrum inversion considering the phase of ground hydrophobic substances and airborne raindrop spectrum inversion considering the phase of airborne hydrophobic substances on the raindrop spectrum fusion profile, respectively.

[0091] The raindrop spectrum parameter profile inversion module is used to invert the raindrop spectrum parameters at different altitudes of the ground raindrop spectrum and the air raindrop spectrum to obtain the raindrop spectrum parameter profile.

[0092] Compared with the prior art, the present invention has the following beneficial effects:

[0093] This invention proposes a method for inverting raindrop spectral parameter profiles. This method addresses the shortcomings of existing precipitation microphysical parameterization schemes in characterizing regional differences and vertical evolution features of raindrop spectra. By integrating precipitation particle phase classification, it improves the method and accuracy of raindrop spectral parameter profile inversion, enhancing the rationality and regional adaptability of the precipitation microphysical process description. The method utilizes multi-source vertical sounding data, employing micro-rain radar (high-sensitivity small particle detection) and laser raindrop spectrometer (ground-based droplet spectrum verification) for collaborative inversion. It quantifies the vertical gradient variation of raindrop spectral parameters, constructs μ–Λ relationships at different altitudes, and achieves raindrop spectral parameter profile inversion based on precipitation particle phase classification.

[0094] The method of this invention reduces the limitations of a single sensor by fusing multi-source data, thereby reducing the error in raindrop spectrum inversion. The improved parameter profile can more accurately indicate the vertical evolution of the raindrop spectrum. Attached Figure Description

[0095] Figure 1 This is a flowchart of the raindrop spectral parameter profile inversion method of the present invention;

[0096] Figure 2(a) is a schematic diagram of the velocity-diameter relationship corresponding to the raw data of raindrop spectrum observation;

[0097] Figure 2(b) is a schematic diagram of the velocity-diameter relationship of raindrop spectrum observation data after removing data that do not meet the ±40% theoretical velocity-diameter relationship;

[0098] Figure 3 This is a schematic diagram of a fully connected neural network model structure;

[0099] Figure 4 A logical diagram illustrating the phase division of atmospheric condensates at different altitudes;

[0100] Figure 5 This is a schematic diagram showing the relationship between the Mie scattering cross section and the Rayleigh scattering cross section. Detailed Implementation

[0101] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0102] Example 1:

[0103] This invention provides a method for inverting raindrop spectral parameter profiles. The raindrop spectral parameters include shape and slope parameters of a distribution model characterizing the raindrop spectral distribution. Generally, this distribution model is a three-parameter Gamma distribution model, which describes the raindrop particle size per unit space. (Unit: mm) corresponding number concentration (Unit: mm⁻¹·m⁻³) Distribution: In the formula, For concentration parameters, For the slope parameter, For shape parameters, such as Figure 1 As shown, the inversion method specifically includes:

[0104] The micro-rain radar base data and laser raindrop spectral base data after quality control are spatiotemporally stitched together to form a raindrop spectral fusion profile with several height layers. In the raindrop spectral fusion profile, one height layer near the ground is composed of laser raindrop spectral base data, and multiple height layers in the air are composed of micro-rain radar base data.

[0105] Ground raindrop spectrum inversion considering the phase of ground hydrophobic deposits and airborne raindrop spectrum inversion considering the phase of airborne hydrophobic deposits were performed on the raindrop spectrum fusion profile respectively.

[0106] Raindrop spectral parameters at different altitudes were inverted from the ground and air raindrop spectra to obtain raindrop spectral parameter profiles.

[0107] Example 2:

[0108] This embodiment is further designed based on Embodiment 1, in that the quality control of the light rain radar base data in this example includes: preprocessing, first light rain radar data removal, and second light rain radar data removal;

[0109] Preprocessing specifically includes noise removal, microwave attenuation correction, Mie scattering correction, air density correction, hydrophobic deformation correction, and hydrophobic terminal velocity correction based on the standard inversion algorithm of micro-rain radar.

[0110] The first light rain radar data removal includes removing the light rain radar base data corresponding to the first altitude layer near the ground and the highest altitude layer in the air, in order to remove data of light rain radar that are easily affected by clutter.

[0111] The second step in filtering micro-rain radar data includes removing Doppler velocity ambiguity data and data with a precipitation rate lower than a set precipitation rate threshold per unit time, such as removing data with a precipitation rate lower than 0.1 mm / h within 1 minute.

[0112] Example 3:

[0113] This embodiment is further designed based on embodiment two. In this example, the quality control of the laser raindrop spectral base data includes: first laser raindrop spectral base data removal and second laser raindrop spectral base data removal; the first laser raindrop spectral base data removal removes data in which the total number of raindrops per unit time is less than a set total number of raindrops threshold or the precipitation rate is lower than a set precipitation rate threshold, such as data in which the total number of raindrops is less than 10 or the precipitation rate is 0.1 mm / h within 1 minute.

[0114] The second laser raindrop spectral basis data elimination includes eliminating the number of particles in the first two diameter levels of the laser raindrop spectral basis data and eliminating data that do not satisfy the ±40% theoretical velocity-diameter relationship;

[0115] Generally, raindrop spectrum observation data are divided into 32 levels according to particle diameter and velocity. The number of particles in the first two diameter levels (the first two levels) are directly removed from all data, which is the data with a diameter of less than 0.25 mm. The smallest detectable particle diameter is 0.25 mm.

[0116] Furthermore, the above-mentioned data that do not satisfy the ±40% theoretical velocity-diameter relationship are eliminated using the following method.

[0117] The theoretical falling velocity of raindrops is calculated using the following formula. :

[0118]

[0119] In the formula, The theoretical falling speed, The equivalent diameter of the particle.

[0120] Calculate the measured velocity of the observed raindrops. Compared with theoretical speed Data with a deviation greater than 40% is removed. The calculation formula is as follows: .

[0121] The following is an explanation with reference to the figures. Figure 2(a) shows the velocity-diameter relationship of the original raindrop spectrum observation data, that is, the data that does not meet the ±40% theoretical velocity-diameter relationship before the data is removed. Figure 2(b) shows the velocity-diameter relationship of the raindrop spectrum observation data after removing the data that does not meet the ±40% theoretical velocity-diameter relationship.

[0122] Example 4:

[0123] This embodiment, based on Embodiment 1, further designs the following method for spatiotemporally stitching together the quality-controlled micro-rain radar base data and the laser raindrop spectral base data:

[0124] The temporal stitching includes: aligning the timestamps of the micro-rain radar base data and the laser raindrop spectral base data. If the timestamps of the micro-rain radar base data and the laser raindrop spectral base data are not synchronized, a linear interpolation method is used to match the micro-rain radar base data to the sampling time point of the laser raindrop spectral base data.

[0125] The vertical splicing in space includes: a near-ground altitude layer using laser raindrop spectral data, multiple altitude layers in the air using micro-rain radar-based data, and missing data at different altitude layers supplemented using cubic spline interpolation.

[0126] Example 5:

[0127] This embodiment, based on Embodiment 1, further designs the following: The specific method for inverting the ground raindrop spectrum considering the phases of ground hydrophobic substances includes:

[0128] 1) Ground water condensate is classified into five phases: drizzle, rain, snow, hail, and mixed phases.

[0129] First, based on the observed diameter and terminal velocity data of the water condensate in the laser raindrop spectral base data, the measured discrete curve of diameter-terminal velocity is obtained.

[0130] The measured discrete curves were divided into several diameter ranges according to the pre-set droplet diameter-velocity classification standard. The measured velocities of the five types of ground water condensate phases corresponding to each diameter range were further calculated.

[0131] Based on the empirical relationship between diameter and final velocity corresponding to the five types of ground hydrated phases, the theoretical velocity corresponding to each diameter range of the five types of ground hydrated phases is calculated.

[0132] Calculate the discrete Friesian distance between the measured velocity and the theoretical velocity corresponding to the same ground hydrogel phase for each diameter range, and take the hydrogel phase corresponding to the minimum discrete Friesian distance as the hydrogel phase for that diameter range.

[0133] 2) Ground raindrop spectrum inversion based on ground hydrophobic phase:

[0134] For velocity-diameter spectra of different phases, precipitation particles with a deviation of more than ±60% from the empirical terminal velocity of the particles are removed. The specific removal method is similar to the steps described above for removing data that do not meet the ±40% theoretical velocity-diameter relationship, in order to reduce the impact of fragmented particles on the observation results. The formula for calculating the particle number concentration per unit volume is as follows:

[0135]

[0136] In the formula, For the diameter at the th Number concentration of hydrogel in the range; This represents the total number of diameter ranges. This refers to the serial number of the diameter range; For the diameter at the th Within the range and the final velocity of the fall is at the 1st The number of particles within the range is The sampling area of ​​the instrument; Sampling time; Indicates the first The velocity of the particle at the end of its fall corresponding to the gear. This refers to the width of the diameter gauge.

[0137] Furthermore, the aforementioned empirical deviation in the final velocity of the falling particle The calculation formula is as follows:

[0138]

[0139] In the formula, This is the actual measured speed; This is the theoretical speed.

[0140] Example 6:

[0141] This embodiment, based on Embodiment 1, further incorporates the following design: The specific method for inverting the airborne raindrop spectrum considering the phases of airborne hydrophobic condensates includes:

[0142] 1) Phase classification: First, read the raw reflectivity spectrum data observed in the micro-rain radar base data. Introducing speed The reflectance spectral density of the function:

[0143]

[0144] in, Distance-based grading ; Divide the Doppler velocity into gears. ; For intermediate parameters, , The frequency resolution of the Doppler spectrum; For light rain radar wavelengths, generally, Take 30.52 Hz. Taking 1.24 cm, the calculation yields... It is 0.1905 m / s;

[0145] Then analyze the original reflectance spectrum data Perform signal enhancement processing, peak signal determination, and Doppler velocity deblurring processing;

[0146] Then calculate the equivalent radar reflectivity factor. The average falling velocity of particles after deblurring Doppler spectral width after deblurring skewness after deblurring :

[0147]

[0148]

[0149]

[0150]

[0151] In the formula, Where is the dielectric constant. , Let be the complex refractive index of water, and be the dielectric constant of water at 24 GHz. The dielectric constant of ice is approximately 0.92, while the dielectric constant of ice, |K|², is 0.18. It is 0.1905 m / s; For the i-th Doppler velocity detected by the micro-rain radar, ;

[0152] Then calculate the equivalent radar reflectivity factor. The vertical gradient, combined with the skewness after deblurring. If a bright band exists, calculate its top height. Bright and high bottom ;

[0153] Then calculate the average velocity when the phase is rain. and the average velocity when the phase is snow , , ;

[0154] The phase classification of atmospheric condensates at different altitudes is performed using the following method; the specific logical relationships can be found [link to relevant documentation]. Figure 4 :

[0155] Will & As the first condition (Condition1), and & As the second condition (Condition 2), and will & As the third condition (Condition 3);

[0156] Bright band bottom height Existence is the fourth condition (Condition 4);

[0157] Will As the fifth condition, and & As the sixth condition, and will As the seventh condition (Condition 7);

[0158] Will As the eighth condition, and & As the ninth condition (Condition 9); The difference in equivalent radar reflectivity factor between different altitude layers is generally... The difference in equivalent radar reflectivity factor between the target altitude level and the previous altitude level;

[0159] For atmospheric condensates at altitudes that simultaneously meet the first, fourth, and fifth criteria, the phase is drizzle, rain, or hail.

[0160] For atmospheric condensates at altitudes that simultaneously meet the second, fourth, and fifth judgment conditions but do not meet the first judgment condition, the phase of the condensate is drizzle, rain, or hail.

[0161] For airborne hydrophobic substances at altitudes that simultaneously meet the third and fourth criteria but do not simultaneously meet the first, second, and seventh criteria, the phase of the hydrophobic substance is drizzle, rain, or hail.

[0162] For airborne hydrogel phases at altitudes that meet the third judgment condition but do not meet the first, second, and fourth judgment conditions, the phase is drizzle, rain, or hail.

[0163] Further classification methods for phases such as drizzle, rain, or hail include:

[0164] The atmospheric condensate phase at the altitude level that meets the eighth judgment condition is hail;

[0165] For atmospheric water condensate at altitudes that do not meet the eighth judgment condition but meet the ninth judgment condition, the phase is drizzle.

[0166] For atmospheric water condensate at altitudes that do not simultaneously meet the eighth and ninth criteria, the phase state is rain.

[0167] The phase state of airborne hydrogel at altitudes that do not simultaneously meet the first, second, and third criteria is uncertain.

[0168] For airborne hydrogel phases at altitudes that simultaneously meet the first and sixth judgment conditions but do not meet the fourth judgment condition, the phase state is a mixed phase.

[0169] The airborne hydrogel phase at a height level that simultaneously meets the second and sixth judgment conditions but does not simultaneously meet the first and fourth judgment conditions is a mixed phase.

[0170] The airborne hydrophobic phase of a height layer that simultaneously meets the third, fourth, seventh, and sixth judgment conditions but does not simultaneously meet the first and second judgment conditions is a mixed phase.

[0171] For airborne hydrogel phases at altitudes that satisfy the first judgment condition but do not simultaneously satisfy the fourth and sixth judgment conditions, the phase of the hydrogel is snow.

[0172] For airborne hydrogel phases at altitudes that satisfy the second judgment condition but do not simultaneously satisfy the first, fourth, and sixth judgment conditions, the phase is snow.

[0173] The atmospheric water condensate phase at a height level that simultaneously meets the third, fourth, and seventh judgment conditions but does not simultaneously meet the first, second, and sixth judgment conditions is snow;

[0174] 2) Based on neural network inversion, historical ground layer data for one near-surface altitude layer and historical vertical layer data for multiple airborne altitude layers are obtained; both ground layer and vertical layer data include several feature variables and PSD (Particle Spectrum Distribution of Different Phases); feature variables include radar reflectivity factor. The final velocity of the falling particles Doppler spectral width Precipitation rate R, precipitation particle phase Ph; the classification of atmospheric condensate phases is the same as that of ground condensate phases, including drizzle, rain, snow, hail and mixed phases;

[0175] Historical ground layer data and historical vertical layer data are stitched together according to height layer to form a feature variable matrix and a raindrop spectrum matrix;

[0176] Furthermore, the PSD of different phase particle spectrum distributions in the ground layer data can be obtained by the ground raindrop spectrum inversion method of the present invention, which considers the phase of ground water condensate; the liquid phase particle spectrum distribution in the vertical layer data is obtained by the following steps; the ice phase particle spectrum distribution in the vertical layer data can be obtained by aircraft observation, and the raindrop spectrum observed by the aircraft and the micro-rain radar are spatiotemporally matched at 1km in space and 10min in time, for example, the vertical particle spectrum distribution observed by the aircraft in the Meiyu front precipitation observation experiment organized by the Wuhan Institute of Rainstorm Research of China Meteorological Administration.

[0177] Furthermore, methods for inverting the liquid phase particle spectral distribution in vertical layer data include:

[0178] When water in the air is in the liquid phase of condensation, i.e., rain or drizzle, it affects the falling velocity of particles. Perform density correction with high dependence Using the terminal velocity of falling particles and diameter Using empirical relationships between them to invert the raindrop spectrum The specific calculation steps are as follows:

[0179]

[0180]

[0181] In the formula, The radar backscattering cross section is used when the diameter of the condensate (raindrop diameter) is relatively small compared to the radar wavelength; otherwise, it is Rayleigh scattering. If the raindrop diameter is relatively large compared to the radar wavelength, it is Mie scattering. The relationship between the Mie scattering cross section and the Rayleigh scattering cross section is as follows (e.g., Figure 5 (As shown)

[0182] The feature variable matrix is ​​represented as follows:

[0183]

[0184] In the formula, For the first Radar reflectivity factor at higher altitudes; For the first The terminal velocity of particles falling at higher altitudes; For the first Doppler spectral width at higher altitudes; For the first Precipitation rate at higher altitudes; For the first Phase of precipitation particles at higher altitudes;

[0185] The raindrop spectral matrix is ​​represented as: ;

[0186] In the formula, For the first Spectral distribution of different phase particles in the altitude layer;

[0187] A neural network model for aerial raindrop spectrum inversion is constructed, using the feature variable matrix as the input data of the neural network model and the raindrop spectrum matrix as the output data of the neural network model for training.

[0188] The feature variable matrix obtained from the target ground layer data and vertical layer data is input into the trained neural network model to obtain the corresponding raindrop spectrum matrix, thereby obtaining the different phase particle spectrum distributions of multiple altitude layers in the air in the raindrop spectrum matrix.

[0189] Furthermore, this example integrates mean square error and the μ-Λ relationship of particles in different phases as constraints into an improved composite loss function, constructing a fully connected neural network optimization model with multiple parameters. This fully connected neural network model mainly includes an input layer, hidden layers, and an output layer. The input layer consists of the multi-parameter input matrices of different height layers obtained in step 1. The hidden layer uses four layers (64 neurons per layer) to model the nonlinear relationship between input and output. The ReLU activation function is applied after the input and hidden layers, and the Adam optimizer is used for parameter optimization. The data is divided into training and validation sets according to a specific ratio. Finally, the output is the particle spectrum distribution (PSD) of particles in different phases in the vertical layer, fitted based on the Gamma distribution. This loss function integrates the traditional mean square error (MSE) and the physical constraint terms of different μ-Λ relationships of hydrogels, and its specific form is as follows:

[0190]

[0191]

[0192]

[0193]

[0194] After substitution, it becomes:

[0195]

[0196]

[0197]

[0198] in, This is the first term in the loss function, corresponding to the constraint term representing MSE; This is the second term in the loss function, corresponding to the liquid phase physical constraint term; This is the third term in the loss function, corresponding to the ice phase physical constraint term; The weighting coefficients for MSE, i.e., the first term. Weighting coefficients; The weighting coefficient for the reflectivity of liquid-phase particles, including rain and drizzle, is term 2. Weighting coefficients; This is the weighting coefficient for the reflectivity of ice phase particles, i.e., the third term. Weighting coefficients; The reflectivity of the liquid phase particles; The reflectivity of ice phase particles;

[0199] By adjusting these weighting coefficients, the influence of various indicators during the training process can be balanced, allowing the model to better adapt to the particle spectral inversion task.

[0200] Furthermore, step 2) related to neural network-based inversion can also be performed using the following method:

[0201] Historical surface layer data for one near-surface altitude and historical vertical layer data for multiple altitudes in the air were acquired. Both surface and vertical layer data included several characteristic variables, particle spectral distributions (PSDs) of different phases, and slope parameters obtained from the PSDs of different phases. and shape parameters The characteristic variables include the radar reflectivity factor. The final velocity of the falling particles Doppler spectral width Precipitation rate R, precipitation particle phase Ph; precipitation particle phases include drizzle, rain, snow, hail and mixed phases;

[0202] Historical ground layer data and historical vertical layer data are stitched together according to height layer to form a feature variable matrix, raindrop spectrum and parameter matrix;

[0203] The feature variable matrix is ​​represented as follows:

[0204]

[0205] In the formula, For the first Radar reflectivity factor at higher altitudes; For the first The terminal velocity of particles falling at higher altitudes; For the first Doppler spectral width at higher altitudes; For the first Precipitation rate at higher altitudes; For the first Phase of precipitation particles at higher altitudes;

[0206] The raindrop spectrum and parameter matrix are represented as follows: ;

[0207] In the formula, For the first Spectral distribution of different phase particles in the altitude layer; For the first Shape parameters of the height layer; For the first Slope parameters of the height layer;

[0208] A neural network model for aerial raindrop spectrum inversion is constructed, using the feature variable matrix as the input data of the neural network model and the raindrop spectrum and parameter matrix as the output data of the neural network model for training.

[0209] The feature variable matrix obtained from the target ground layer data and vertical layer data is input into the trained neural network model to obtain the corresponding raindrop spectrum and parameter matrix. From the raindrop spectrum and parameter matrix, the shape parameters and slope parameters of multiple altitude layers in the air, as well as the distribution of particle spectra in different phases, can be obtained. The structure of the neural network model constructed using this method is as follows: Figure 3 As shown.

[0210] Technicians can choose to construct a neural network model using either the feature variable matrix as the model input data and the raindrop spectrum and parameter matrix as the model output data, or the feature variable matrix as the model input data and the raindrop spectrum matrix as the model output data, depending on the specific inversion prediction target, hardware environment, and other factors.

[0211] Specifically, to prevent overfitting, regularization is used to improve the model's generalization performance. A dynamic weight adjustment algorithm is employed to optimize the loss function coefficients. First, initial data fitting term weights are set. Then, an adaptive algorithm dynamically adjusts λ2 and λ3 based on the rate of change of loss to balance physical constraints and data errors. Differential weight correction coefficients are set for different precipitation phases (liquid / ice phase). The specific dynamic weight adjustment rules are as follows:

[0212] When setting initial weights at the start of training, the following weight initialization scheme can be used:

[0213]

[0214] Then calculate the current loss of the second and third terms in the loss function. Assess the relative changes in various losses before taking action:

[0215]

[0216] In the formula, For the first The relative change of the loss; For the first The current loss of the item (the first item) Step loss); For the first Item Losses before the step (the first step) Step loss); This refers to the index of the loss term in the loss function. These correspond to liquid phase physical constraints and ice phase physical constraints, respectively. A decrease in losses indicates that constraints need to be strengthened; conversely, a decrease in losses indicates that constraints need to be weakened.

[0217] Use the hyperbolic tangent function (tanh) to smoothly adjust the amplitude and avoid abrupt changes:

[0218]

[0219] In the formula, and The first The first item Step and the first The weighting coefficients corresponding to the step loss; This is the learning rate for weight adjustment, used to determine the adjustment magnitude; the default value is 0.05. It also constrains the weight coefficients. and Within a reasonable range, and while maintaining overall stability, and The reasonable range can be .

[0220] Furthermore, different weighted correction coefficients are applied to different phases (liquid / ice phase).

[0221]

[0222] In the formula, This is the correction coefficient for phase correlation. These are phase indicator functions, representing the liquid and ice phases, respectively.

[0223] By monitoring the rate of change of loss as described above, the effect of physical constraints can be automatically strengthened or weakened. If the physical constraint term loses... If the value increases (the model deviates from the theoretical relationship), then the value increases. Conversely, it decreases.

[0224] Furthermore, the weights can be adjusted differently for different precipitation types (such as hail) to improve the rationality of the inversion.

[0225] Example 7:

[0226] This embodiment, based on Embodiment 1, further incorporates the following method for inverting raindrop spectral parameters at different altitudes to obtain raindrop spectral parameter profiles from the ground and airborne raindrop spectra:

[0227] Based on the ground raindrop spectrum and the airborne raindrop spectrum, through The step-moment method is used to calculate the raindrop spectral parameter profiles at different altitudes, including the slope parameter Λ of the raindrop spectrum (unit: mm). -1 and the shape parameters of the raindrop spectrum ,in, Step moment is defined as

[0228]

[0229]

[0230] In the formula, and These represent the lower and upper limits of raindrop diameter, respectively; the range of raindrop diameter detected by micro-rain radar is 0.246 mm to 5.03 mm, while the range of raindrop diameter detected by laser raindrop spectrum is 0.0625 mm to 8 mm. For concentration parameters, For the slope parameter, For shape parameters;

[0231] Then, using the second, third, and fourth moments, calculate the slope parameter Λ and shape parameter of the raindrop spectrum as follows. :

[0232]

[0233]

[0234] ;

[0235] In the formula, This is an intermediate parameter.

[0236] Furthermore, radar reflectivity can be calculated as follows: (unit: dBZ) and precipitation intensity (Unit: mm·h) -1 ):

[0237]

[0238]

[0239] in, It is the Doppler velocity observed by laser raindrop spectrum or micro-rain radar.

[0240] Example 8:

[0241] This embodiment is further designed based on Embodiment 1, in that the method of the present invention in this example also includes: classifying precipitation into stratiform cloud precipitation, convective precipitation and weak precipitation based on laser raindrop spectral base data;

[0242] Raindrop spectral parameters were inverted at different altitudes based on the precipitation type corresponding to the precipitation period in both the surface and airborne raindrop spectra, yielding raindrop spectral parameter profiles.

[0243] Example 9:

[0244] This embodiment, based on Embodiment 8, further designs the precipitation category classification by: classifying precipitation categories based on the surface precipitation rate R and its time series retrieved from the laser raindrop spectrum.

[0245] For precipitation rate observation data with a 1-minute time resolution, if the precipitation rate is greater than 0.5 mm / h for a period of more than 10 minutes and the standard deviation of the precipitation rate is less than 1.5 mm / h during that period, it is identified as stratiform cloud precipitation.

[0246] If the precipitation rate is greater than 5 mm / h and the standard deviation of the precipitation rate is greater than 1.5 mm / h, it is identified as convective precipitation.

[0247] Precipitation samples with a precipitation rate below 0.5 mm / h are identified as weak precipitation.

[0248] Example 10:

[0249] The present invention provides a raindrop spectral parameter profile inversion system, wherein the raindrop spectral parameters include shape parameters and slope parameters of a distribution model used to characterize the raindrop spectral distribution, and includes a spatiotemporal stitching module, a raindrop spectral inversion module, and a raindrop spectral parameter profile inversion module;

[0250] The spatiotemporal stitching module is used to spatiotemporally stitch together the quality-controlled micro-rain radar base data and the laser raindrop spectral base data to form a raindrop spectral fusion profile with several altitude layers. In the raindrop spectral fusion profile, one altitude layer near the ground is composed of laser raindrop spectral base data, and multiple altitude layers in the air are composed of micro-rain radar base data.

[0251] The raindrop spectrum inversion module is used to perform ground raindrop spectrum inversion considering the phase of ground hydrophobic substances and airborne raindrop spectrum inversion considering the phase of airborne hydrophobic substances, respectively, on the raindrop spectrum fusion profile.

[0252] The raindrop spectrum parameter profile inversion module is used to invert the raindrop spectrum parameters at different altitudes of the ground raindrop spectrum and the air raindrop spectrum to obtain the raindrop spectrum parameter profile.

[0253] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for inverting raindrop spectral parameter profiles, wherein the raindrop spectral parameters include shape parameters and slope parameters of a distribution model used to characterize the raindrop spectral distribution, characterized in that, include: The micro-rain radar base data and laser raindrop spectral base data after quality control are spatiotemporally stitched together to form a raindrop spectral fusion profile with several height layers. In the raindrop spectral fusion profile, one height layer near the ground is composed of laser raindrop spectral base data, and multiple height layers in the air are composed of micro-rain radar base data. The raindrop spectrum fusion profile was subjected to ground raindrop spectrum inversion considering the phase of ground hydrophobic material and airborne raindrop spectrum inversion considering the phase of airborne hydrophobic material, respectively. The raindrop spectrum parameters at different altitudes are inverted from the ground raindrop spectrum and the air raindrop spectrum to obtain raindrop spectrum parameter profiles. The specific methods for inverting the aerial raindrop spectrum considering the phases of aerial hydrophobic condensates include: 1) First, read the raw reflectivity spectrum data observed in the micro-rain radar base data. Introducing speed The reflectance spectral density of the function: ; in, Distance-based grading ; Divide the Doppler velocity into gears. ; For intermediate parameters, , The frequency resolution of the Doppler spectrum; This is the wavelength for light rain radar; Then analyze the original reflectance spectrum data Perform signal enhancement processing, peak signal determination, and Doppler velocity deblurring processing; Then calculate the equivalent radar reflectivity factor. The average falling velocity of particles after deblurring Doppler spectral width after deblurring skewness after deblurring : ; ; ; In the formula, Where is the dielectric constant. , The complex refractive index of water, For the i-th Doppler velocity detected by the micro-rain radar, ; Then calculate the equivalent radar reflectivity factor. The vertical gradient, combined with the skewness after deblurring. If a bright band exists, calculate its top height. Bright and high bottom ; Then calculate the average velocity when the phase is rain. and the average velocity when the phase is snow , , ; The phases of atmospheric condensates at different altitudes are classified according to the following method: & As the first judgment condition, and & As the second judgment condition, and will & As a third condition for judgment; Bright band bottom height Existence as the fourth condition; As the fifth judgment condition, and will & As the sixth condition for judgment, and will As the seventh condition; As the eighth judgment condition, and will & As the ninth judgment condition; This represents the difference in equivalent radar reflectivity factor between different altitude levels; For atmospheric condensates at altitudes that simultaneously meet the first, fourth, and fifth criteria, the phase is drizzle, rain, or hail. For atmospheric condensates at altitudes that simultaneously meet the second, fourth, and fifth judgment conditions but do not meet the first judgment condition, the phase of the condensate is drizzle, rain, or hail. For airborne hydrophobic substances at altitudes that simultaneously meet the third and fourth criteria but do not simultaneously meet the first, second, and seventh criteria, the phase of the hydrophobic substance is drizzle, rain, or hail. For airborne hydrogel phases at altitudes that meet the third judgment condition but do not meet the first, second, and fourth judgment conditions, the phase is drizzle, rain, or hail. Further classification methods for phases such as drizzle, rain, or hail include: The atmospheric condensate phase at the altitude level that meets the eighth judgment condition is hail; For atmospheric water condensate at altitudes that do not meet the eighth judgment condition but meet the ninth judgment condition, the phase is drizzle. For atmospheric water condensate at altitudes that do not simultaneously meet the eighth and ninth criteria, the phase state is rain. The phase state of airborne hydrogel at altitudes that do not simultaneously meet the first, second, and third criteria is uncertain. For airborne hydrogel phases at altitudes that simultaneously meet the first and sixth judgment conditions but do not meet the fourth judgment condition, the phase state is a mixed phase. The airborne hydrogel phase at a height level that simultaneously meets the second and sixth judgment conditions but does not simultaneously meet the first and fourth judgment conditions is a mixed phase. The airborne hydrophobic phase of a height layer that simultaneously meets the third, fourth, seventh, and sixth judgment conditions but does not simultaneously meet the first and second judgment conditions is a mixed phase. For airborne hydrogel phases at altitudes that satisfy the first judgment condition but do not simultaneously satisfy the fourth and sixth judgment conditions, the phase of the hydrogel is snow. For airborne hydrogel phases at altitudes that satisfy the second judgment condition but do not simultaneously satisfy the first, fourth, and sixth judgment conditions, the phase is snow. The atmospheric condensate phase at a height that simultaneously satisfies the third, fourth, and seventh judgment conditions but does not simultaneously satisfy the first, second, and sixth judgment conditions is snow.

2. The raindrop spectral parameter profile inversion method according to claim 1, characterized in that, The quality control of the light rain radar base data includes: preprocessing, first light rain radar data removal, and second light rain radar data removal. The preprocessing specifically includes noise removal, microwave attenuation correction, Mie scattering correction, air density correction, hydrophobic deformation correction, and hydrophobic terminal velocity correction based on the micro-rain radar standard inversion algorithm. The first light rain radar data removal includes removing light rain radar base data corresponding to the first altitude layer near the ground and the highest altitude layer in the air; The second light rain radar data removal includes removing Doppler velocity ambiguity data and data with a precipitation rate lower than a set precipitation rate threshold per unit time.

3. The raindrop spectral parameter profile inversion method according to claim 2, characterized in that, The quality control of the laser raindrop spectral base data includes first laser raindrop spectral base data removal and second laser raindrop spectral base data removal. The first laser raindrop spectral base data removal includes removing data where the total number of raindrops per unit time is less than a set total raindrop number threshold or the precipitation rate is less than a set precipitation rate threshold; The second laser raindrop spectral base data elimination includes eliminating the number of particles in the first two diameter levels of the laser raindrop spectral base data and eliminating data that does not satisfy the ±40% theoretical velocity-diameter relationship.

4. The raindrop spectral parameter profile inversion method according to claim 1, characterized in that, The specific method for spatiotemporally stitching together the quality-controlled micro-rain radar base data and laser raindrop spectral base data includes: The temporal stitching includes: aligning the timestamps of the micro-rain radar base data and the laser raindrop spectral base data. If the timestamps of the micro-rain radar base data and the laser raindrop spectral base data are not synchronized, a linear interpolation method is used to match the micro-rain radar base data to the sampling time point of the laser raindrop spectral base data. Spatial stitching includes: a near-ground altitude layer using laser raindrop spectral data, multiple altitude layers in the air using micro-rain radar-based data, and missing data at different altitude layers supplemented using cubic spline interpolation.

5. The raindrop spectral parameter profile inversion method according to claim 1, characterized in that, The specific methods for ground raindrop spectrum inversion considering the phase state of ground hydrophobic substances include: 1) Ground water condensate is classified into five phases: drizzle, rain, snow, hail, and mixed phases. Based on the observed diameter and terminal velocity data of water condensate from the laser raindrop spectral base data, the measured discrete curve of diameter-terminal velocity was obtained. The measured discrete curves are divided into several diameter ranges according to a pre-set droplet diameter-velocity classification standard. The measured velocities of the five types of ground water condensate phases corresponding to each diameter range are further calculated. Based on the empirical relationship between diameter and final velocity corresponding to the five types of ground hydrate phases, the theoretical velocity corresponding to each diameter range for the five types of ground hydrate phases is calculated. Calculate the discrete Friesian distance between the measured velocity and the theoretical velocity corresponding to the same ground hydrogel phase for each diameter range, and take the hydrogel phase corresponding to the minimum discrete Friesian distance as the hydrogel phase for that diameter range. 2) Ground raindrop spectrum inversion based on ground hydrophobic phase: Based on the empirical diameter-terminal velocity curves for particles of different phases, the deviation between the observed diameter-terminal velocity and the corresponding empirical curves for particles of different phases is calculated. Precipitation particles with a deviation exceeding 60% are removed. The formula for calculating the particle number concentration per unit volume is as follows: In the formula, For the diameter at the th Number concentration of hydrogel in the range; This represents the total number of diameter ranges. This refers to the serial number of the diameter range; For the diameter at the th Within the range and the final velocity of the fall is at the 1st The number of particles within the range; The sampling area of ​​the instrument; Sampling time; Indicates the first The velocity of the particle at the end of its fall corresponding to the gear. This refers to the width of the diameter gauge.

6. The raindrop spectral parameter profile inversion method according to claim 1, characterized in that, The specific method for inverting the airborne raindrop spectrum considering the phase state of airborne condensates further includes: 2) acquiring historical ground layer data at one altitude level near the ground and historical vertical layer data at multiple altitude levels in the air; both the ground layer data and the vertical layer data include several characteristic variables and the PSD of particle spectrum distributions in different phase states; the characteristic variables include radar reflectivity factor. The final velocity of the falling particles Doppler spectral width Precipitation rate R, precipitation particle phase Ph; the precipitation particle phase includes drizzle, rain, snow, hail and mixed phase; The historical ground layer data and historical vertical layer data are stitched together according to height layer to form a feature variable matrix and a raindrop spectrum matrix; The feature variable matrix is ​​represented as follows: In the formula, For the first Radar reflectivity factor at higher altitudes; For the first The terminal velocity of particles falling at higher altitudes; For the first Doppler spectral width at higher altitudes; For the first Precipitation rate at higher altitudes; For the first Phase of precipitation particles at higher altitudes; The raindrop spectral matrix is ​​represented as follows: In the formula, For the first Spectral distribution of different phase particles in the altitude layer; A neural network model for aerial raindrop spectrum inversion is constructed, using the feature variable matrix as input data and the raindrop spectrum matrix as output data to train the neural network model. The feature variable matrix obtained from the target ground layer data and vertical layer data is input into the trained neural network model to obtain the corresponding raindrop spectrum matrix. From this, the shape parameters and slope parameters of multiple altitude layers in the air, as well as the distribution of particle spectrum in different phases, are obtained from the raindrop spectrum matrix.

7. The raindrop spectral parameter profile inversion method according to claim 1, characterized in that, The specific method for inverting the raindrop spectral parameters at different altitudes from the ground raindrop spectrum and the airborne raindrop spectrum to obtain the raindrop spectral parameter profile is as follows: Based on the ground raindrop spectrum and the airborne raindrop spectrum, through The step-moment method is used to calculate the raindrop spectral parameter profiles at different altitudes, including the slope parameter Λ and the shape parameter of the raindrop spectrum. ,in, Step moment is defined as ; In the formula, and These are the lower and upper limits of the raindrop diameter, respectively; For concentration parameters, For the slope parameter, For shape parameters; Then, using the second, third, and fourth moments, calculate the slope parameter Λ and shape parameter of the raindrop spectrum as follows. : ; ; In the formula, This is an intermediate parameter.

8. The raindrop spectral parameter profile inversion method according to claim 1, characterized in that, The method further includes: classifying precipitation into stratiform cloud precipitation, convective precipitation, and weak precipitation based on the laser raindrop spectral base data; Raindrop spectrum parameters are inverted at different altitudes for the ground raindrop spectrum and the air raindrop spectrum according to the precipitation category corresponding to the precipitation period, to obtain raindrop spectrum parameter profiles; raindrop spectrum parameters are inverted at different altitudes for the ground raindrop spectrum and the air raindrop spectrum according to the precipitation category corresponding to the precipitation period, to obtain raindrop spectrum parameter profiles.

9. The raindrop spectral parameter profile inversion method according to claim 8, characterized in that, The precipitation category classification includes: classifying precipitation categories based on the surface precipitation rate R and its time series retrieved from laser raindrop spectra. For precipitation rate observation data with a 1-minute time resolution, if the precipitation rate is greater than 0.5 mm / h for a period of more than 10 minutes and the standard deviation of the precipitation rate is less than 1.5 mm / h during that period, it is identified as stratiform cloud precipitation. If the precipitation rate is greater than 5 mm / h and the standard deviation of the precipitation rate is greater than 1.5 mm / h, it is identified as convective precipitation. Precipitation samples with a precipitation rate below 0.5 mm / h are identified as weak precipitation.

10. A raindrop spectral parameter profile inversion system, wherein the raindrop spectral parameters include shape parameters and slope parameters of a distribution model characterizing the raindrop spectral distribution, characterized in that, It includes a spatiotemporal stitching module, a raindrop spectrum inversion module, and a raindrop spectrum parameter profile inversion module; The spatiotemporal stitching module is used to spatiotemporally stitch together the quality-controlled micro-rain radar base data and the laser raindrop spectrum base data to form a raindrop spectrum fusion profile with several height layers. In the raindrop spectrum fusion profile, one height layer near the ground is composed of laser raindrop spectrum base data, and multiple height layers in the air are composed of micro-rain radar base data. The raindrop spectrum inversion module is used to perform ground raindrop spectrum inversion considering the phase of ground hydrophobic substances and airborne raindrop spectrum inversion considering the phase of airborne hydrophobic substances on the raindrop spectrum fusion profile, respectively. The raindrop spectrum parameter profile inversion module is used to invert the raindrop spectrum parameters at different altitudes of the ground raindrop spectrum and the air raindrop spectrum to obtain the raindrop spectrum parameter profile. The specific methods for inverting the aerial raindrop spectrum considering the phases of aerial hydrophobic condensates include: 1) First, read the raw reflectivity spectrum data observed in the micro-rain radar base data. Introducing speed The reflectance spectral density of the function: ;in, Distance-based grading ; Divide the Doppler velocity into gears. ; For intermediate parameters, , The frequency resolution of the Doppler spectrum; The wavelength for light rain radar; then the original reflectivity spectrum data... Signal enhancement processing, peak signal determination, and Doppler velocity deblurring are performed; then the equivalent radar reflectivity factor is calculated. The average falling velocity of particles after deblurring Doppler spectral width after deblurring skewness after deblurring : ; ; ; In the formula, Where is the dielectric constant. , The complex refractive index of water, For the i-th Doppler velocity detected by the micro-rain radar, ; Then calculate the equivalent radar reflectivity factor. The vertical gradient, combined with the skewness after deblurring. If a bright band exists, calculate its top height. Bright and high bottom ; Then calculate the average velocity when the phase is rain. and the average velocity when the phase is snow , , ; The phases of atmospheric condensates at different altitudes are classified according to the following method: & As the first judgment condition, and & As the second judgment condition, and will & As a third condition for judgment; Bright band bottom height Existence as the fourth condition; As the fifth judgment condition, and will & As the sixth condition for judgment, and will As the seventh condition; As the eighth judgment condition, and will & As the ninth judgment condition; This represents the difference in equivalent radar reflectivity factor between different altitude levels; For atmospheric condensates at altitudes that simultaneously meet the first, fourth, and fifth criteria, the phase is drizzle, rain, or hail. For atmospheric condensates at altitudes that simultaneously meet the second, fourth, and fifth judgment conditions but do not meet the first judgment condition, the phase of the condensate is drizzle, rain, or hail. For airborne hydrophobic substances at altitudes that simultaneously meet the third and fourth criteria but do not simultaneously meet the first, second, and seventh criteria, the phase of the hydrophobic substance is drizzle, rain, or hail. For airborne hydrogel phases at altitudes that meet the third judgment condition but do not meet the first, second, and fourth judgment conditions, the phase is drizzle, rain, or hail. Further classification methods for phases such as drizzle, rain, or hail include: The atmospheric condensate phase at the altitude level that meets the eighth judgment condition is hail; For atmospheric water condensate at altitudes that do not meet the eighth judgment condition but meet the ninth judgment condition, the phase is drizzle. For atmospheric water condensate at altitudes that do not simultaneously meet the eighth and ninth criteria, the phase state is rain. The phase state of airborne hydrogel at altitudes that do not simultaneously meet the first, second, and third criteria is uncertain. For airborne hydrogel phases at altitudes that simultaneously meet the first and sixth judgment conditions but do not meet the fourth judgment condition, the phase state is a mixed phase. The airborne hydrogel phase at a height level that simultaneously meets the second and sixth judgment conditions but does not simultaneously meet the first and fourth judgment conditions is a mixed phase. The airborne hydrophobic phase of a height layer that simultaneously meets the third, fourth, seventh, and sixth judgment conditions but does not simultaneously meet the first and second judgment conditions is a mixed phase. For airborne hydrogel phases at altitudes that satisfy the first judgment condition but do not simultaneously satisfy the fourth and sixth judgment conditions, the phase of the hydrogel is snow. For airborne hydrogel phases at altitudes that satisfy the second judgment condition but do not simultaneously satisfy the first, fourth, and sixth judgment conditions, the phase is snow. The atmospheric condensate phase at a height that simultaneously satisfies the third, fourth, and seventh judgment conditions but does not simultaneously satisfy the first, second, and sixth judgment conditions is snow.

Images