A ground snowfall identification method based on FY3G satellite precipitation measurement radar

Abstract

The application provides a ground snow identification method based on FY3G satellite precipitation measurement radar, comprising: performing equiangular cylindrical projection, bilinear interpolation and Kriging filling operation on L1 level pretreatment data and L2 level product data of the radar to obtain equi- latitude and longitude projection data; filtering the vertical profile of the radar reflectivity factor of the equi- latitude and longitude projection data based on three-factor threshold value, then performing spatial registration and three spline interpolation double-frequency profile data calculation to obtain a double-frequency ratio profile; then calculating the snowfall index according to the slope of the double-frequency ratio profile calculated by a regression algorithm and according to the maximum reflectivity on the profile and the storm top height, so as to identify the snowfall of the screened calibration area. The method realizes effective identification of large-scale ground snowfall by combining the L1 level pretreatment data and the L2 level product data of the radar with data threshold filtering and factor discrimination technology, and improves the spatial coverage and accuracy of snowfall identification.

Description

A ground snowfall identification method based on FY3G satellite precipitation measurement radar Technical Field

[0001] This invention relates to the field of radar data identification technology, and in particular to a ground snowfall identification method based on FY3G satellite precipitation measurement radar. Background Technology

[0002] Precipitation can be divided into liquid precipitation and solid precipitation according to its different physical characteristics. Snowfall is a type of solid precipitation and an important component of precipitation in mid-to-high latitude regions. It has a significant impact on urban transportation and agricultural production. Heavy snowfall is one of the major meteorological disasters affecting my country, and obtaining information on large-scale snowfall is becoming increasingly important.

[0003] Current snowfall observation methods mainly include conventional ground observation, spaceborne passive microwave remote sensing, and active spaceborne radar. Conventional ground observation is the traditional main method, but it suffers from limited spatial coverage and uneven distribution due to the layout of observation stations, making it difficult to meet the needs of large-scale dynamic snowfall monitoring. While spaceborne passive microwave remote sensing can reflect the distribution information of cloud and rain particles, it cannot effectively detect the three-dimensional structure of precipitation and clouds, making it difficult to support vertical feature analysis of snowfall for accurate identification. Active spaceborne radar can compensate for the shortcomings of the first two methods, acquiring large-scale three-dimensional structure information of clouds and precipitation. TRMM / PR has achieved three-dimensional detection of global precipitation, but its detection capability for snowfall in mid-to-high latitude regions is limited. Although GPM / DPR is equipped with a dual-frequency precipitation detection radar, improving its detection capability for weak and solid precipitation, it still relies on near-surface echo features to identify snowfall in some scenarios and is prone to missed detections when the near-surface echo intensity is lower than the radar's minimum detection intensity. Therefore, it is essential to design a ground snowfall identification method based on the FY3G satellite precipitation measurement radar. Summary of the Invention

[0004] The purpose of this invention is to provide a ground snowfall identification method based on FY3G satellite precipitation measurement radar. By combining L1-level preprocessed data and L2-level product data from the satellite precipitation measurement radar with data threshold filtering and factor discrimination techniques, the method can effectively identify ground snowfall over a large area and improve the spatial coverage and accuracy of snowfall identification.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A ground snowfall identification method based on FY3G satellite precipitation measurement radar includes the following steps:

[0007] The L1-level preprocessed data and L2-level product data of the satellite precipitation measurement radar are subjected to isotropic cylindrical projection, bilinear interpolation and kriging filling operations to obtain isotropic latitude and longitude projection data.

[0008] Based on the three-factor threshold equilateral latitude and longitude projection data, K u Bands and K a The vertical profile of radar reflectivity factor without attenuation correction is filtered; the three-factor thresholds include: minimum height without clutter, storm top height, and minimum detection threshold of the band.

[0009] Spatial registration was performed on the vertical profile of the filtered radar reflectivity factor, and K was unified using cubic spline interpolation. u Bands and K a The vertical height layer spacing of the bands yields the dual-frequency ratio profile;

[0010] The slope of the dual-frequency ratio profile is calculated using a regression algorithm selected by the variance inflation factor; the regression algorithms include: least squares method and ridge regression algorithm;

[0011] The snowfall index is calculated based on the slope of the dual-frequency profile, the maximum reflectance, and the storm top height.

[0012] The areas to be calibrated were selected based on the comparison between the lowest altitude without clutter and the altitude of the 0℃ isotherm, and snowfall was identified in the areas to be calibrated using the snowfall index.

[0013] Optionally, the L1-level preprocessed data and L2-level product data of the satellite precipitation measurement radar are subjected to isometric cylindrical projection, bilinear interpolation, and kriging fill operations to obtain isotropic projection data, including:

[0014] Using the geodetic coordinate system as a reference, the L1 level preprocessed data and L2 level product data are converted into plane rectangular coordinates according to the preset central meridian and standard parallel, respectively, to obtain preliminary projection data containing plane coordinates and reflectivity factor;

[0015] The effective pixels within a 2×2 neighborhood of the pixel to be interpolated in the preliminary projection data are selected as the neighborhood pixels. The distance weighting calculation between the neighborhood pixels and the pixel to be interpolated is performed using the preset target resolution to obtain the projection interpolation data.

[0016] Spatial fitting of missing pixels in the projection interpolation data is performed using a spherical variogram, and the weights of the fitted missing pixels are calculated using a semi-variogram and then interpolated to fill them, thus obtaining equal latitude and longitude projection data.

[0017] Optionally, spatial registration is performed on the vertical profile of the filtered radar reflectivity factor, and K is unified using cubic spline interpolation. u Bands and K a The vertical height layer spacing of the bands yields the dual-frequency ratio profile, including:

[0018] Based on the latitude and longitude grid, the orbital deviation correction parameters in the isotropic projection data are spatially matched with the filtered radar reflectivity factor vertical profile, so that the original latitude and longitude coordinates of the radar reflectivity factor vertical profile obtain unique orbital deviation correction parameters; the orbital deviation correction parameters include: longitude offset and latitude offset.

[0019] Through formula and Correction calculations are performed on the original latitude and longitude coordinates; among which, and These are the original longitude and original latitude coordinates. and These are the longitude and latitude offsets;

[0020] When K u Bands and K a When the difference in latitude and longitude coordinates after pixel correction within the same grid of the band is ≤0.001 degrees, the vertical profile of the radar reflectivity factor after offset correction is obtained.

[0021] A piecewise cubic spline interpolation function is constructed using the original vertical height layer in the offset-corrected radar reflectivity factor vertical profile as the independent variable and the radar reflectivity factor as the dependent variable.

[0022] The target height sequence in the offset-corrected radar reflectivity factor vertical profile is input into a piecewise cubic spline interpolation function to calculate the radar reflectivity factor, thereby unifying K. u Bands and K a Vertical height layer spacing of the band;

[0023] Through formula The dual-frequency ratio profile was obtained; where and K is the unified vertical height layer. u Bands and K a Radar reflectivity factor of the band This refers to the dual-frequency ratio.

[0024] Optionally, the specific formula for calculating the slope of the dual-frequency ratio profile using the least squares method includes:

[0025] ;

[0026] ;

[0027] ;

[0028] in, The number of floors is the height. For the first The height of the layer, The average of all height values. The average of the squares of the height values. For the first The dual-frequency ratio of the layer, for The average value of the layer dual-frequency ratio, It is the average of the product of the two-frequency ratios corresponding to all heights.

[0029] Optionally, the specific formula for calculating the slope of the dual-frequency ratio profile using the ridge regression algorithm includes:

[0030] ;

[0031] ;

[0032] in, It's a dual-frequency ratio. To show the height value, and These are model parameters. It is an error term. It is the loss function of the ridge regression algorithm.

[0033] Optionally, the formula for calculating the snowfall index is:

[0034] ;

[0035] in, for The maximum reflectivity of the band, For the storm crest height, This represents the slope of the dual-frequency profile.

[0036] Optionally, the area to be calibrated is selected based on a comparison between the lowest altitude without clutter and the altitude of the 0°C isotherm, and snowfall is identified in the area to be calibrated using the snowfall index, including:

[0037] Calculate the height difference between the lowest altitude free from clutter and the altitude of the 0°C isotherm;

[0038] If the elevation difference is between 1.25km and 1.5km, it is identified as a critical area; if the elevation difference is greater than 1.5km, it is identified as a normal area. The critical area and the normal area are combined into the area to be calibrated.

[0039] If the snowfall index × 100 of the area to be calibrated is greater than 17, then it is determined that there is snowfall in the area to be calibrated.

[0040] According to specific embodiments provided by the present invention, the following technical effects are disclosed: The ground snowfall identification method based on FY3G satellite precipitation measurement radar provided by the present invention includes: performing isometric cylindrical projection, bilinear interpolation, and kriging fill operations on L1-level preprocessed data and L2-level product data of satellite precipitation measurement radar to obtain isotropic projection data; filtering the vertical profiles of radar reflectivity factors in the Ku-band and Ka-band without attenuation correction in the isotropic projection data based on three-factor thresholds; the three-factor thresholds include: the lowest altitude without clutter influence, the storm top... The method employs several techniques: first, minimum detection thresholds for high-frequency bands; second, spatial registration of the filtered radar reflectivity factor vertical profile; and third, unification of the vertical height layer intervals between the Ku and Ka bands using cubic spline interpolation to obtain a dual-frequency ratio profile. The slope of the dual-frequency ratio profile is calculated using a regression algorithm selected by variance expansion factor selection. Regression algorithms include least squares and ridge regression. A snowfall index is calculated based on the dual-frequency ratio profile slope and storm top height. The area to be calibrated is selected based on the lowest clutter-free altitude and the 0°C isotherm altitude, and snowfall is identified within this area using the snowfall index. This method, combining L1-level preprocessed data and L2-level product data from satellite precipitation measurement radar with data threshold filtering and factor discrimination techniques, achieves effective identification of large-scale ground snowfall and improves the spatial coverage and accuracy of snowfall identification. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 is a flowchart of the ground snowfall identification method of the present invention;

[0043] Figure 2 is a schematic diagram of the ground snowfall recognition strategy according to an embodiment of the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] As shown in Figures 1 and 2, this embodiment of the invention provides a ground snowfall identification method based on FY3G satellite precipitation measurement radar, comprising the following steps:

[0047] Step 100: Perform isometric cylindrical projection, bilinear interpolation, and kriging fill operations on the L1-level preprocessed data and L2-level product data of the satellite precipitation measurement radar to obtain isotropic projection data.

[0048] Step 200: Based on the three-factor threshold, equilateral latitude and longitude projection data in K u Bands and K a The vertical profile of radar reflectivity factor without attenuation correction is filtered; the three-factor thresholds include: minimum height without clutter, storm top height, and minimum detection threshold of the band.

[0049] Step 300: Spatial registration is performed on the vertical profile of the filtered radar reflectivity factor, and K is unified using cubic spline interpolation. u Bands and K a The vertical height layer spacing of the bands yields the dual-frequency ratio profile;

[0050] Step 400: Calculate the slope of the two-frequency ratio profile using a regression algorithm selected by the variance inflation factor; the regression algorithm includes: least squares method and ridge regression algorithm;

[0051] Step 500: Calculate the snowfall index based on the slope of the dual-frequency profile, the maximum reflectance, and the storm top height;

[0052] Step 600: Select the area to be calibrated based on the comparison results of the lowest height without clutter and the height of the 0℃ isotherm, and identify the snowfall in the area to be calibrated by the snowfall index.

[0053] The specific steps of step 100 in this embodiment include:

[0054] Step 101: Using the WGS84 geodetic coordinate system as the reference, determine the preset central meridian as 0° and the standard latitude as ±30°, and then input the L1 level preprocessed data (including K) from the FY3G satellite precipitation measurement radar (PMR). u Bands and K a The original radar echo information of the band and L2-level product data (including preprocessed reflectivity factor, clutter suppression related auxiliary information, and three-factor thresholds) are used to convert the longitude and latitude of the original geodetic coordinates of each pixel in both types of data into Cartesian coordinates using an isometric cylindrical projection algorithm. During the conversion process, the K coordinates of each pixel are retained. u Bands and K aThe radar reflectivity factor attribute of the band is used to obtain preliminary projection data containing planar rectangular coordinates and corresponding band reflectivity factors, so as to ensure that the projected data can initially eliminate the influence of latitude and longitude spherical coordinates on subsequent spatial processing.

[0055] Step 102: Select the effective pixels within a 2×2 neighborhood of the pixel to be interpolated in the preliminary projection data as the neighborhood pixels. Perform distance-weighted calculations between the neighborhood pixels and the pixel to be interpolated using a preset target resolution to obtain the projection interpolation data. Set the target spatial resolution to 0.05 degrees × 0.05 degrees. Based on the preliminary projection data obtained in Step 101, locate the pixel to be interpolated within this target resolution grid one by one. Then, for each pixel to be interpolated, search for effective pixels within its surrounding 2×2 neighborhood. Pixels with non-filled radar reflectivity factors are considered effective and are designated as neighborhood pixels. Calculate the weight of each neighborhood pixel based on the Euclidean distance between the neighborhood pixels and the pixel to be interpolated, and obtain the reflectivity factor value of the pixel to be interpolated through distance-weighted calculations. Finally, obtain projection interpolation data with a unified resolution of 0.05 degrees × 0.05 degrees. The expression is: ;in, This represents the radar reflectivity factor value of the pixel to be interpolated, i.e., the target pixel value calculated through bilinear interpolation. Let be the radar reflectivity factor value of the top-left effective pixel within a 2×2 neighborhood surrounding the pixel to be interpolated. This represents the radar reflectivity factor value of the top-right effective pixel within the neighborhood. This represents the radar reflectivity factor value of the effective pixel in the lower left corner of the neighborhood. Let be the radar reflectivity factor value of the effective pixel in the lower right corner of the neighborhood, and 'a' be the weighting coefficient of the pixel to be interpolated in the x-direction. The calculation formula is: , b is the weight coefficient of the pixel to be interpolated in the y direction, calculated as follows: .

[0056] Step 103: Spatial fitting of missing pixels in the projection interpolation data is performed using a spherical variogram, and the weights of the fitted missing pixels are calculated using a semi-variogram and then interpolated to fill the gaps, resulting in equal latitude and longitude projection data. First, the missing pixels in the projection interpolation data from Step 102 are marked as -9999. The spatial correlation of their surrounding valid pixels is fitted using a spherical variogram to determine the spatial variation patterns and structure among valid pixels. Then, based on the fitted spatial variation structure, the interpolation weight of each neighboring valid pixel for the missing pixel is calculated using a semi-variogram, and the missing pixels are interpolated using a weighted summation method, thus ensuring that each 0.05° × 0.05° grid contains valid K pixels. u Bands and K a Equivalent latitude and longitude projection data of band reflectivity factor.

[0057] In this embodiment, step 200 classifies profile pixels below the lowest height unaffected by clutter as invalid pixels affected by clutter, profile pixels above the storm crest height as invalid pixels in non-snowfall areas, and profile pixels below the minimum detection threshold of the band as invalid pixels with weak signals. For all three types of invalid pixels, preset invalid values ​​are used for data filling. Then, a moving average method is used to smooth the vertical profiles of the radar reflectivity factor after filling in the two bands, thereby eliminating the abrupt changes caused by the filling values, and finally obtaining the filtered vertical profiles of the radar reflectivity factor. In this embodiment, K u The minimum detection threshold for the band is -18 dBZ, K a The minimum detection threshold for the band is -12 dBZ.

[0058] In this embodiment, step 300 uses spatial registration and cubic spline interpolation operations to adjust K... u Bands and K a The vertical height layer spacing of the bands is unified to obtain the dual-frequency ratio profile. The specific process of spatial registration is as follows: First, based on the filtered K... u Bands and K a Vertical profiles of radar reflectivity factors for each band; read the original latitude and longitude coordinates corresponding to each profile. Simultaneously, data including longitude offsets were extracted from L2-level data of the FY3G satellite precipitation measurement radar. and latitude offset The orbital deviation correction parameters are then spatially matched with the filtered radar reflectivity factor vertical profile using a 0.05° × 0.05° equal latitude and longitude grid, ensuring that the original latitude and longitude coordinates of the radar reflectivity factor vertical profile obtain unique orbital deviation correction parameters. Next, the original latitude and longitude coordinates of each profile are corrected using the following formula: ; When K u Bands and K a If the difference in latitude and longitude coordinates after pixel correction within the same grid of the band is ≤0.001 degrees, the correction is deemed effective, and the vertical profile of the radar reflectivity factor after offset correction is obtained. If the difference in coordinates is >0.001 degrees, the orbital deviation correction parameters are extracted again, and the above correction operation is repeated until the correction is effective.

[0059] The specific process of cubic spline interpolation is as follows: after obtaining the profile corrected for latitude and longitude, determine K. u The original vertical height layer spacing of the band is 250m, K aThe interval between the original vertical height layers of the Ku-band is 100m, and the interval between the target uniform height layers is 200m. The target height sequence ranges from 0.2km to 15km, with a step size of 200m. Then, using the original vertical height layers of the Ku-band as the independent variable and the radar reflectivity factor as the dependent variable, a piecewise cubic spline interpolation function is constructed, expressed as:

[0060] ;

[0061] in, For any height value in the target height sequence within the offset-corrected profile, i.e., the height to be calculated. For x in the subinterval The corresponding radar reflectivity factor value, where i is the index of the sub-interval, used to locate the original height layer interval to which x belongs. This refers to the i-th original vertical height layer of the vertical profile of reflectivity factor for a certain band after spatial registration. , and Let K be the coefficient of the interpolation polynomial of the i-th subinterval. This function satisfies the following conditions: (1) the interpolation function of adjacent height segments is continuous at the junction point; (2) the first derivative is continuous at the junction point; and (3) the second derivative is continuous at the junction point. a Bands and K u The function construction process is the same for each band. Next, the target height sequence is input into a piecewise cubic spline interpolation function to calculate the radar reflectivity factor, in order to unify K. u Bands and K a The vertical altitude layer spacing of the band. Finally, the dual-frequency ratio profile is calculated, and the expression is:

[0062] ;

[0063] in and K is the unified vertical height layer. u Bands and K a Radar reflectivity factor of the band This refers to the dual-frequency ratio.

[0064] Specifically, the formula for calculating the slope of the dual-frequency ratio profile using the least squares method includes:

[0065] ;

[0066] ;

[0067] ;

[0068] in, The number of floors is the height. For the first The height of the layer, The average of all height values. The average of the squares of the height values. For the first The dual-frequency ratio of the layer, for The average value of the layer dual-frequency ratio, It is the average of the product of the two-frequency ratios corresponding to all heights.

[0069] In this embodiment, step 400 involves selecting different regression algorithms as follows: First, an independent variable matrix is ​​constructed, containing two columns: one column of constant terms (all 1s) and the other column of standardized height values. Then, for the height value column, linear regression is performed using the other columns as explanatory variables to obtain the regression determination coefficient R. 2 Then, according to the formula VIF=1 / (1-R) 2 Calculate the variance inflation factor (VIF) of the height variable. If VIF ≤ 10, it is determined that there is no significant multicollinearity of the independent variable, and the least squares method is selected to calculate the slope of the bifrequency ratio profile; if VIF > 10, it is determined that there is significant multicollinearity, and the ridge regression algorithm is selected.

[0070] Specifically, the formula for calculating the slope of the dual-frequency ratio profile using the ridge regression algorithm includes:

[0071] ;

[0072] ;

[0073] in, It's a dual-frequency ratio. To show the height value, and These are model parameters. It is an error term. It is the loss function of the ridge regression algorithm.

[0074] Specifically, the formula for calculating the snowfall index is:

[0075] ;

[0076] in, The slope is unaffected by calibration changes, therefore the slope is used instead. value, for The maximum reflectivity of the band, For the storm crest height, This represents the slope of the dual-frequency profile.

[0077] In this embodiment, step 600 specifically involves the following steps: First, the lowest clutter-free altitude corresponding to each isotropic latitude and longitude grid is read. Simultaneously, L2-level temperature profile data generated by the FY3G satellite infrared detector is extracted from the isotropic latitude and longitude projection data, and the 0℃ isotherm altitude of each grid is read from it. Then, the height difference between the lowest clutter-free altitude and the 0℃ isotherm altitude in each grid is calculated. If 1.25km ≤ height difference ≤ 1.5km, the corresponding grid is determined as a critical region; if the height difference > 1.5km, the corresponding grid is determined as a normal region, and the critical region and the normal region are integrated into the calibration region. Finally, if the snowfall index × 100 value in the calibration region is > 17, it is determined that snowfall exists in the calibration region.

[0078] The beneficial effects of this invention are as follows:

[0079] 1) By using conformal cylindrical projection, bilinear interpolation, and Kriging filling to fill data gaps, the influence of latitude and longitude spherical coordinates was eliminated, solving the problems of limited spatial coverage and uneven distribution of conventional ground observations. At the same time, it made up for the deficiency of spaceborne passive microwave remote sensing in being unable to detect the three-dimensional structure of precipitation, broke through the spatial limitations of ground observations, and realized large-scale dynamic monitoring of snowfall.

[0080] 2) By unifying the resolution of the initial projection data through bilinear interpolation, the problem of the difference in the original resolution of L1 / L2 level data is solved; by using Kriging filling to spatially fit the missing pixels in the projection interpolation data, the recognition interruption caused by data gaps is avoided; by filling and smoothing the invalid pixels of the filtered profile, the interference of abrupt data on subsequent calculations is eliminated, ensuring the integrity and stability of the entire data processing chain.

[0081] 3) The multi-condition verification mechanism of three-factor threshold avoids noise interference from a single algorithm, enabling the present invention to adapt to high-latitude weak snowfall and thick cloud attenuation scenarios, improving the accuracy of snowfall recognition while reducing signal omission and misjudgment.

[0082] 4) The regression algorithm uses VIF to determine multicollinearity for adaptive selection, enabling the method to adapt to the statistical characteristics of data at different height levels. Furthermore, an L2-level cloud top height correction mechanism is introduced during the acquisition of storm top height, avoiding parameter loss in extreme scenarios and enhancing the algorithm's environmental adaptability.

[0083] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0084] Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. Furthermore, those skilled in the art will recognize that, based on the ideas of this invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for identifying ground snowfall based on FY3G satellite precipitation measurement radar, characterized in that, The process includes the following steps: performing isometric cylindrical projection, bilinear interpolation, and kriging fill operations on the L1-level preprocessed data and L2-level product data of the satellite precipitation measurement radar to obtain isotropic projection data; and applying a three-factor threshold to the K value in the isotropic projection data. u Bands and K a The vertical profile of radar reflectivity factor without attenuation correction in the band is filtered. The three-factor thresholds include: the minimum altitude free from clutter, the storm top height, and the minimum detection threshold for the band; spatial registration is performed on the vertical profile of the filtered radar reflectivity factor, and the K value is unified using cubic spline interpolation. u Bands and the K a The vertical height layer interval of the band is used to obtain the dual-frequency ratio profile; the slope of the dual-frequency ratio profile is calculated by a regression algorithm selected by the variance expansion factor; the regression algorithm includes: least squares method and ridge regression algorithm; the snowfall index is calculated based on the slope of the dual-frequency ratio profile, the maximum reflectivity, and the storm top height; the area to be calibrated is screened based on the comparison results of the lowest height without clutter influence and the height of the 0℃ isotherm, and snowfall is identified in the area to be calibrated by the snowfall index; the vertical profile of the filtered radar reflectivity factor is spatially registered, and the K is unified by cubic spline interpolation. u Bands and the K a The vertical height layer spacing of the bands yields a dual-frequency ratio profile, including: spatially matching the orbital deviation correction parameters in the isotropic projection data with the filtered radar reflectivity factor vertical profile according to the latitude and longitude grid, so that the original latitude and longitude coordinates of the radar reflectivity factor vertical profile obtain a unique orbital deviation correction parameter; the orbital deviation correction parameter includes: longitude offset and latitude offset; using the formula... and The original latitude and longitude coordinates are corrected and calculated; wherein, and These are the original longitude and original latitude coordinates. and For longitude and latitude offsets; when the K u Bands and the K a When the difference in latitude and longitude coordinates after pixel correction within the same band is ≤0.001 degrees, the offset-corrected vertical profile of the radar reflectivity factor is obtained. A piecewise cubic spline interpolation function is constructed using the original vertical height layer in the offset-corrected vertical profile of the radar reflectivity factor as the independent variable and the radar reflectivity factor as the dependent variable. The target height sequence in the offset-corrected vertical profile of the radar reflectivity factor is input into the piecewise cubic spline interpolation function to calculate the radar reflectivity factor, thereby unifying the K... u Bands and the K a Vertical height layer spacing of the band; through formula The dual-frequency ratio profile was obtained; where and K is the unified vertical height layer. u Bands and K a Radar reflectivity factor of the band This refers to the dual-frequency ratio.

2. The ground snowfall identification method based on FY3G satellite precipitation measurement radar according to claim 1, characterized in that, The L1-level preprocessed data and L2-level product data of satellite precipitation measurement radar are subjected to isometric cylindrical projection, bilinear interpolation, and kriging fill operations to obtain isotropic projection data. This includes: using a geodetic coordinate system as a reference, converting the L1-level preprocessed data and the L2-level product data into Cartesian coordinates according to preset central meridians and standard parallels, obtaining preliminary projection data containing plane coordinates and reflectivity factors; selecting effective pixels within a 2×2 neighborhood of the pixel to be interpolated in the preliminary projection data as neighboring pixels, and performing distance-weighted calculations between the neighboring pixels and the pixel to be interpolated using a preset target resolution to obtain projection interpolation data; spatially fitting the missing pixels in the projection interpolation data using a spherical variogram, and calculating the weights of the fitted missing pixels using a semi-variogram and then interpolating to fill in the missing pixels, thus obtaining the isotropic projection data.

3. The ground snowfall identification method based on FY3G satellite precipitation measurement radar according to claim 1, characterized in that, The specific formula for calculating the slope of the dual-frequency ratio profile using the least squares method includes: ； ； ;in, The number of floors is the height. For the first The height of the layer, The average of all height values. The average of the squares of the height values. For the first The dual-frequency ratio of the layer, for The average value of the layer dual-frequency ratio, It is the average of the product of the two-frequency ratios corresponding to all heights.

4. The ground snowfall identification method based on FY3G satellite precipitation measurement radar according to claim 1, characterized in that, The specific formula for calculating the slope of the dual-frequency ratio profile using the ridge regression algorithm includes: ； ;in, It's a dual-frequency ratio. To show the height value, and These are model parameters. It is an error term. It is the loss function of the ridge regression algorithm.

5. The ground snowfall identification method based on FY3G satellite precipitation measurement radar according to claim 1, characterized in that, The formula for calculating the snowfall index is: ;in, for The maximum reflectivity of the band, For the storm crest height, This represents the slope of the dual-frequency profile.

6. The ground snowfall identification method based on FY3G satellite precipitation measurement radar according to claim 1, characterized in that, The calibration area is selected based on the comparison between the lowest clutter-free altitude and the 0℃ isotherm altitude. Snowfall is then identified in the calibration area using the snowfall index, including: calculating the height difference between the lowest clutter-free altitude and the 0℃ isotherm altitude; if 1.25km ≤ the height difference ≤ 1.5km, it is determined to be a critical area; if the height difference > 1.5km, it is determined to be a normal area. The critical area and the normal area are then combined into the calibration area; if the snowfall index × 100 of the calibration area is greater than 17, it is determined that snowfall exists in the calibration area.

Images