A ground-based radar calibration bias estimation method based on star-ground precipitation radar verification

Abstract

The application discloses a ground-based radar calibration deviation estimation method based on a star-ground precipitation radar calibration, and mainly comprises the following steps: performing Ku-band to S-band frequency correction on a spaceborne radar (SR) reflectivity factor, and performing polar coordinate system projection conversion with the ground-based radar (GR) as the center after the correction; performing time and space overlap matching on the converted SR and GR reflectivity factors with the accuracy of seconds; then improving mean accuracy through inverse distance weighting and reducing precipitation spatial heterogeneity of the SR and the GR in the matching space through a non-uniform beam filling quality control technology; and a star-ground precipitation radar quality control factor threshold standard system is established, quality control is performed in the whole process during the calibration deviation estimation process, and finally, the spaceborne radar (SR) and the ground-based radar (GR) reflectivity factors after the quality control processing are compared and calibrated, so that the ground-based radar (GR) calibration deviation estimation is realized. The application can effectively reduce the error of the star-ground radar calibration and improve the accuracy of the precipitation estimation.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of satellite ground equipment calibration systems, and particularly relates to a ground-based radar calibration deviation estimation method based on satellite-ground precipitation radar verification. BACKGROUND

[0002] With the continuous development of meteorological observation technology, ground-based radar plays an increasingly important role in precipitation monitoring. Ground-based radar realizes quantitative estimation of precipitation by detecting the scattering and attenuation of electromagnetic waves by precipitation particles. However, in the actual observation process, the calibration accuracy of ground-based radar is affected by various factors, such as atmospheric environment and equipment performance, which directly affects the accuracy of precipitation estimation. SUMMARY

[0003] In order to solve the problems existing in the prior art, the application provides a ground-based radar calibration deviation estimation method based on satellite-ground precipitation radar verification, which adopts a new frequency correction method of full phase state and full precipitation type, a pixel time difference matching method accurate to seconds, and a variety of factor sensitivity experiments, analyzes in detail the influence of factors such as precipitation type, bright band, and attenuation correction deficiency on test accuracy, designs threshold standards for quality control factor system, reduces the error caused by different frequencies and relative movement of the instrument, effectively improves the matching accuracy of satellite-ground radar and product test accuracy, and realizes accurate estimation of the calibration deviation of ground-based radar, providing strong support for meteorological observation and weather forecasting.

[0004] The application provides a ground-based radar calibration deviation estimation method based on satellite-ground precipitation radar verification, which includes:

[0005] S1, obtaining reflectivity factor raw data of a spaceborne radar SR and a ground-based radar GR, and performing quality control on the raw data;

[0006] S2, performing frequency correction from Ku band to S band on the SR reflectivity factor after quality control;

[0007] S3, performing polar coordinate system projection conversion with the GR as the center on the corrected SR reflectivity factor;

[0008] S4, correcting the horizontal latitude and longitude information on each height layer of the SR based on the satellite zenith angle;

[0009] S5, converting the reflectivity factor of the SR and the GR from logarithmic scale to linear scale respectively by using an improved radar reflectivity factor average method;

[0010] S6, performing time and space overlap matching of the converted SR and GR reflectivity factors accurate to seconds;

[0011] S7, respectively improve the mean accuracy of the SR and GR reflectivity factors in each matched overlapping spatial range by inverse distance weighting;

[0012] S8, reduce the spatial heterogeneity of the SR and GR precipitation in the matched space by the in-pixel non-uniform beam filling quality control technology;

[0013] S9, respectively control the precipitation type, precipitation phase and bright band factor of the SR in the matched space to reduce the influence on the evaluation accuracy;

[0014] S10, respectively convert the reflectivity factors of the SR and GR from linear scale to logarithmic scale by using the improved radar reflectivity factor averaging method;

[0015] S11, based on the comparison and verification of the converted SR reflectivity factor and GR reflectivity factor, perform ground-based radar GR calibration bias estimation.

[0016] Further, the S1, respectively acquire the reflectivity factor raw data of the spaceborne radar SR and the ground-based radar GR, and perform quality control on the raw data, including:

[0017] S11, acquire the reflectivity factor raw data of the spaceborne radar SR;

[0018] S12, perform quality control on the acquired reflectivity factor raw data of the spaceborne radar SR by using dual-frequency observation quality, precipitation classification quality and dual-frequency precipitation identification parameters;

[0019] S13, acquire the reflectivity factor raw data of the ground-based radar GR;

[0020] S14, perform quality control on the GR reflectivity factor by removing the calibration distance data.

[0021] Further, the S2, perform frequency correction of the Ku band to the S band on the quality-controlled SR reflectivity factor, including:

[0022] S21, acquire the raindrop number concentration Nw, precipitation particle phase and temperature T data from the SR product data; S22, calculate the backscattering cross section σ b and the corresponding scattering function fz(Dm) by integral calculation according to the acquired data values;

[0023] The scattering function fz(Dm) is:

[0024]

[0025] Wherein, λ is the wavelength of S-band radar, Kw is the complex refractive index of water, T is temperature, σb is the backscattering cross section of raindrop particle, D is the diameter of raindrop particle, and Dm is the median diameter of raindrop particle.

[0026] S23, combining the raindrop number concentration Nw provided in the product and fz(Dm) calculated in S22, the radar reflectivity factor Zs of the S-band is calculated to realize the correction of frequency.

[0027] The corrected radar reflectivity factor Zs of the S-band is:

[0028] Zs=Nwfz(Dm);

[0029] Wherein, Nw is the raindrop number concentration, Dm is the median diameter of raindrop particle, and fz(Dm) is the backscattering function of the S-band.

[0030] Further, the S4 corrects the horizontal latitude and longitude information on each height layer of SR based on the satellite zenith angle, including:

[0031] S41, the satellite observation field deviation distance is calculated by using the satellite zenith angle;

[0032] S42, the field distance deviation of the horizontal latitude and longitude on each height layer is respectively calculated based on the satellite scanning angle and the observation field deviation distance;

[0033] S43, the deviation amount is subtracted from the SR ground horizontal latitude and longitude information, and the horizontal latitude and longitude information on each height layer is obtained by correction.

[0034] Further, the S5 converts the reflectivity factor of SR and GR from logarithmic scale to linear scale by using the improved radar reflectivity factor average method, including:

[0035] The reflectivity factor of SR and GR is converted from logarithmic scale dBZ to linear scale dB by using the expression dBZ=10logdB.

[0036] Further, the S6 performs the overlap matching of the converted SR and GR reflectivity factors in time and space to seconds, including:

[0037] S61, the SR transit GR matching is screened;

[0038] S62, the observation time of each row of the screened SR is extracted, and the value is expanded to a three-dimensional observation distance library time, which is recorded as TSR;

[0039] S63, the time of each beam of each elevation angle of GR is extracted, and the value is expanded to a three-dimensional observation distance library time, which is recorded as TGR;

[0040] S64, calculate the SR to GR station center distance closest distance library, the SR beam time of the distance is marked as TSR0;

[0041] S65, calculate the average value of the first three elevation angle scanning time of SR, marked as TGR0;

[0042] S66, calculate the time difference between GR and SR Screening from GR base data The smallest GR base data is screened and screened;

[0043] S67. Screen each overlapping spatial range that matches, respectively calculate the average time of GR and SR in the overlapping space, and calculate the time difference between the two The time accuracy is accurate to seconds, and the selected spatial overlap range is determined.

[0044] Further, the S7, the SR and GR reflectivity factor in each overlapping spatial range that matches is respectively improved by inverse distance weighting to improve the mean value accuracy, comprising:

[0045] S71, according to the unified polar coordinate projection of GR and SR, first from the SR beam distance library cycle, and then from the GR elevation cycle, traverse each distance library corresponding fixed GR elevation angle relationship;

[0046] S72. Through the spatiotemporal matching, the star-ground radar beam cross space is obtained, the matched GR distance library is used for inverse distance weighted average within the 5 kilometer range of SR footprint area, and the matched SR distance library is used for inverse distance weighted average within the GR elevation beam widening range.

[0047] Further, the S8, the non-uniform beam filling quality control technology is used to reduce the precipitation spatial heterogeneity of SR and GR in the matching space, comprising:

[0048] S81, non-uniform beam filling effect improvement technology of GR

[0049] The minimum detection reflectivity factor threshold of GR is set as Zg* = 15dBZ, and for each volume matched sample, the proportion of GR distance library in the volume that satisfies Zg ≥ Zg* is recorded as fg;

[0050] By excluding samples with fg less than the beam filling percentage factor, the influence of GR non-uniform beam filling in the matching volume is reduced, and here the non-uniform beam filling factor fg = 0.9.

[0051] S82, non-uniform precipitation filling effect improvement technology of SR

[0052] The minimum detection reflectivity factor threshold of the SR is set as Zs* = 15 dBZ, and for each volume-matched sample, the proportion of the SR distance library in the volume satisfying Zs >= Zs* is recorded as fs;

[0053] The influence of the non-uniform beam filling of the SR in the matching volume is reduced by excluding the samples with fs less than the beam filling percentage factor, where the non-uniform beam filling factor fs = 0.9.

[0054] S83, non-uniform beam filling quality control technology in a pixel of the SR

[0055] The proportion of the distance library of the SR with a non-uniform beam correction parameter paramNUBF greater than 0.1 in the pixel to the SR distance library in the volume is recorded as fs_nubf.

[0056] The influence of the non-uniform beam filling in the pixel area of the SR is reduced by excluding the samples with fs_nubf less than the beam filling percentage factor, where fs_nubf = 1.0.

[0057] Further, the S9, the precipitation type, the precipitation phase state and the bright band factor of the SR in the matching space are respectively subjected to quality control, and the influence on the evaluation accuracy is reduced, including:

[0058] S91. For each sample of the SR in the matching volume, the proportion of stratiform precipitation in the volume is recorded as f_stratiform, and the influence of the SR precipitation type on the accuracy evaluation is reduced by excluding the samples with f_stratiform less than the precipitation classification factor.

[0059] S92. For each sample of the SR in the matching volume, the proportion of convective precipitation in the volume is recorded as f_liquid, and the influence of the SR precipitation phase state in the matching volume is reduced by excluding the samples with f_liquid less than the phase state classification factor.

[0060] S93. For each sample of the SR in the matching volume, the bright band identification factor f_bb in the volume is recorded, and the influence of the observed precipitation bright band of the SR in the matching volume is reduced by excluding the samples with f_bb equal to 0.

[0061] The beneficial effects of the present application are:

[0062] 1. The present application adopts a new frequency correction method of full phase state and full precipitation type and a pixel time difference matching method with an accuracy of 1s, which effectively reduces the errors caused by different frequencies and precipitation intensities and movement changes in the evaluation process.

[0063] 2、The present application analyzes in detail how the precipitation type, bright band, and insufficient attenuation correction affect the test accuracy through a variety of factor sensitivity experiments, designs a threshold standard for the quality control factor system, and improves the matching accuracy of the space-ground radar and the product test accuracy;

[0064] 3、The present application adopts an improved radar reflectivity factor average method, converts dBZ into dB, averages, and then converts into dBZ, can correct the BIAS by about 1.6 dB, proves that the direct average of the previous dBZ leads to low BIAS, and effectively improves the accuracy of the radar calibration bias estimation. BRIEF DESCRIPTION OF DRAWINGS

[0065] Figure 1 is a flowchart of a ground-based radar calibration bias estimation method based on space-ground precipitation radar verification of the present application;

[0066] Figure 2 is a sensitivity relationship diagram of the difference ΔZ between the reflectivity factors of the spaceborne radar SR and the ground-based radar GR and the time bias factor of an embodiment of the present application;

[0067] Figure 3 is a statistical probability density distribution diagram of the difference ΔZ between the reflectivity factors of the three ground-based radars GR and the spaceborne radar SR in the Guangdong-Hong Kong-Macao Greater Bay Area of an embodiment of the present application. DETAILED DESCRIPTION

[0068] To make the purpose, technical scheme and advantages of the present application clearer, further description will be made in combination with the drawings and embodiments.

[0069] It should be noted that, in this paper, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between the entities or operations. Moreover, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0070] As shown in the accompanying Figure 1 , the present application provides a ground-based radar calibration bias estimation method based on space-ground precipitation radar verification, which comprises the following steps:

[0071] S1, obtaining radar reflectivity factor raw data of a spaceborne radar SR and a ground-based radar GR, and performing quality control on the raw data, including:

[0072] S11, obtaining the raw data of the radar reflectivity factor of the spaceborne radar SR;

[0073] GPM Core Observatory (GPM-CO) was successfully launched on February 27, 2014, to measure global precipitation in space. GPM-CO runs at an altitude of 407 km, with an expanded range of 65°S-65°N in the polar orbit, covering the subtropical and tropical regions, with an orbital period of about 93 min, 15 or 16 orbits a day. Dual frequency Precipitation Radar (DPR is the precipitation radar carried on GPM-CO (hereinafter referred to as "spaceborne radar SR")), scanning width 245km, beam width 0.71°, detection accuracy 1dB, vertical observation height 19km, space resolution 5.2km at the subsatellite point. The stable operation and high observation accuracy of DPR (hereinafter referred to as "spaceborne radar SR") can provide reference for global ground-based weather radar (hereinafter referred to as "ground-based radar GR") to carry out ground-based radar calibration bias estimation based on space-ground precipitation radar calibration.

[0074] S12, quality control of the obtained spaceborne radar SR radar reflectivity factor raw data, including:

[0075] S121, quality control of Ku band radar reflectivity factor of spaceborne radar SR through dual frequency observation quality;

[0076] Spaceborne radar SR has two detection bands of Ku at 13.6 GHz and Ka at 35.5 GHz. KuPR vertical resolution is 250 meters, and minimum detection accuracy is 18dBZ (0.5mm / h). KaPR vertical resolution is 250m / 500m, and minimum detection accuracy is 12dBZ (0.2mm / h). Quality control of Ku band radar reflectivity factor of spaceborne radar SR raw data can improve the accuracy and reliability of the data through the quality control method of dual frequency observation. In dual frequency observation, Ka band has shorter wavelength, which is suitable for higher resolution observation and can more accurately capture the details of precipitation particles, especially light precipitation; Ku band has moderate wavelength, which is commonly used in weather radar, especially in the observation of moderate intensity precipitation and snowfall.

[0077] The different attenuation characteristics of the two frequency bands make them complementary to correct radar data. Through the comparison of dual frequency data, the rainfall can be more accurately estimated, and the performance difference of different bands under different precipitation intensities can also be corrected, especially in the relationship between radar reflectivity factor and precipitation intensity. Through the collaborative processing of data of two frequency bands, it is helpful to improve the accuracy of quality control.

[0078] Further, in dual-frequency observations, Ku-band radar reflectivity factor data quality control for spaceborne radar SR can be conducted in conjunction with auxiliary dual-frequency observation quality codes (auxiliary products of spaceborne radar SR). By filtering out data with optimal quality levels, effective optimization can be achieved in the Ku-band radar reflectivity factor quality control process of spaceborne radar SR raw data, thereby improving the reliability of observation data.

[0079] S122, using the precipitation classification quality parameter and the dual-frequency precipitation identification parameter to classify and control the quality of the Ku-band radar reflectivity factor;

[0080] To improve the accuracy and reliability of precipitation data, when performing precipitation classification quality control, the precipitation classification quality parameter and the dual-frequency precipitation identification parameter are used to help identify and eliminate potential measurement errors, noise and inconsistencies of the Ku-band radar reflectivity factor, wherein: the precipitation classification quality parameter is used for quality control: in the auxiliary products of spaceborne radar SR, the precipitation classification quality parameter is identified as 1-layer, 2-convective, 3-other, -1111-no rain, and -9999.9-missing. By filtering out data with parameters of 1-layer and 2-convective, the rest is excluded.

[0081] Quality control using dual-frequency precipitation identification parameters: dual-frequency radar provides the ability to reflect echo data at two frequencies, which helps improve the accuracy of precipitation identification. The optimal data in the dual-frequency precipitation identification parameter is used for quality control, and the rest is excluded.

[0082] By using the precipitation classification quality and dual-frequency precipitation identification parameters to classify and control the quality of the Ku-band radar reflectivity factor, the accuracy of radar observation data can be effectively improved, and errors caused by meteorological phenomena or technical reasons can be reduced. This method is of great significance for the improvement of precipitation type classification and intensity estimation, especially in complex meteorological conditions, which can help improve the accuracy of precipitation monitoring and early warning.

[0083] S13, obtaining raw data of radar reflectivity factor of ground-based radar GR;

[0084] The original data of ground-based radar GR is derived from the meteorological information system. The newly-built weather radar network (CINRAD) has hundreds of weather radars (S-band radar and C-band weather radar). The S-band dual-polarization weather radar covers four types of SAD, SB, SC and WSR88D0, is equipped with a horizontally polarized parabolic antenna, and changes its azimuth and elevation by mechanical means. The volume scanning mode of the S-band dual-polarization radar (ground-based radar GR) includes 9 layers of elevation (0.5°, 1.5°, 2.4°, 3.3°, 4.3°, 6.0°, 9.9°, 14.6°, 19.5°), the beam width is 0.98°, the distance library interval is 250 meters, the volume scanning time resolution is 6 minutes, and the effective business detection range is 230 km. The S-band dual-polarization weather radar can obtain quantitative information such as differential radar reflectivity factor and ground surface rainfall intensity, and is one of the most important detection means for weather monitoring and early warning business. Therefore, the calibration and performance maintenance of the S-band dual-polarization weather radar are also very important, which directly determines the accuracy of precipitation measurement.

[0085] S14, quality control on the original data of ground-based radar GR;

[0086] Considering that there is a certain quality difference between the near-distance and far-distance observations of ground-based radar GR, the original data of ground-based radar GR is quality controlled by eliminating the radar reflectivity factor data within 15 km and outside 230 km.

[0087] S2, frequency correction from Ku band to S band on the quality-controlled reflectivity factor of satellite-borne radar SR;

[0088] For the same observation target, there is a difference in the radar reflectivity factor detected by radars of different wavelengths. Under Rayleigh scattering conditions, the S-band ground-based radar reflectivity factor is the sum of the particle diameters to the sixth power, which is independent of the radar wavelength. The satellite Ku-band radar detects large-scale precipitation particles, which is suitable for Mie scattering approximation, so it is necessary to correct the Ku band of satellite-borne radar SR to S band, which is conducive to subsequent estimation of the calibration bias of ground-based radar GR. The corrected data can improve the observation accuracy of the multi-band radar system and ensure the accuracy and consistency in data fusion, precipitation estimation and meteorological analysis. It includes the following steps:

[0089] S21, obtaining data including raindrop number concentration Nw, precipitation particle phase and temperature T from the secondary product of dual-frequency precipitation measurement radar DPR carried on GPM core observation satellite GPM-CO;

[0090] S22, integrating and calculating the backscattering cross section σ of raindrop particles according to the obtained data values b and the corresponding scattering function fz(Dm) value;

[0091] The scattering function fz(Dm) is:

[0092]

[0093] where λ is the wavelength of the S-band radar, Kw is the complex refractive index of water, T is the temperature, σb is the backscattering cross section of the raindrop, D is the diameter of the raindrop, and Dm is the median diameter of the raindrop.

[0094] S23, combining the raindrop number concentration Nw provided in the product and fz(Dm) calculated in S22, the radar reflectivity factor Z of the S-band is calculated s , and the frequency is corrected.

[0095] The corrected radar reflectivity factor Z of the S-band is s .

[0096] Z s =Nwfz(Dm);

[0097] where Nw is the raindrop number concentration, Dm is the median diameter of the raindrop, and fz(Dm) is the backscattering function of the S-band.

[0098] Combining Nw and fz(Dm), Z is calculated using the expression s , and the radar reflectivity factor of the S-band is obtained, and the frequency is corrected.

[0099] S3, the corrected SR reflectivity factor is projected and converted in a polar coordinate system centered on the ground-based radar GR, including:

[0100] S31. Determine the coordinate system and parameters of the ground-based radar GR.

[0101] Determine the geographical position (latitude and longitude) of the ground-based radar GR, the antenna feedback source horizontal altitude, and define the polar coordinate system centered on GR, including radial distance r, azimuth angle θ, and elevation angle β.

[0102] S32, convert the geographical coordinates of the spaceborne radar SR data into polar coordinates centered on GR, including:

[0103] Use the azimuth equidistant projection coordinate system to project the latitude and longitude of each SR beam distance library with the Earth Reference Coordinate System WGS84 into relative geographical information centered on the GR station.

[0104] To project the latitude and longitude of each SR beam distance library with the Earth Reference Coordinate System (WGS84) into relative geographical information centered on the ground-based radar GR station, an azimuth and distance coordinate system (such as a polar coordinate system) can be used for projection. The following are the steps to achieve this process:

[0105] S321. Define coordinate system and variables

[0106] GR station: Located at a known latitude and longitude position, with longitude λGRand latitude φGR.

[0107] SR beam: The observation point position of each beam is given by the corresponding latitude and longitude, with longitude λSRand latitude φSR.

[0108] All position data is based on latitude and longitude in the WGS84 coordinate system.

[0109] S322. Calculate distance and azimuth angle between GR and SR beams

[0110] In order to project the data of SR beams into the relative coordinate system centered on GR, it is first necessary to calculate the distance and azimuth angle between the GR station and each SR beam.

[0111] Calculate the distance between ground radar GR and observation point: For latitude and longitude in the WGS84 coordinate system, use the great circle distance formula (Haversine formula) to calculate the spherical distance (i.e. radial distance r) :

[0112]

[0113] where: R is the radius of the Earth, approximately 6371 km; Δλ = λSR- λGRis the longitude difference, in radians; Δφ = φSR- φGRis the latitude difference, in radians.

[0114] Calculate the azimuth angle: In order to determine the azimuth angle, i.e. the direction from the GR station to the SR beam, the following formula can be used to calculate the azimuth angle θ:

[0115] θ = arctan2(sin(Δλ) · cos(φ SR ), cos(φ GR ) · sin(φ SR - sin(φ GR ) · cos(φ SR ) · cos(Δλ))

[0116] where: arctan2(y, x) is a function that calculates the four-quadrant inverse tangent, and the output angle θ is the azimuth angle (positive direction from north to east).

[0117] The azimuth angle θ gives the direction from the GR station to the SR beam, in radians, which is usually converted to degrees.

[0118] S323. Convert to relative coordinate system

[0119] After calculating the distance d and azimuth θ for each SR beam, these data can be converted to relative positions in a polar coordinate system. Assume that the GR station is the origin of the polar coordinate system (i.e., distance is 0 and azimuth is 0). Convert the distance and azimuth to (x, y) coordinates in a relative coordinate system: x = d*sin(θ); y = d*cos(θ). Where x is the easting distance relative to the GR station, and y is the northing distance relative to the GR station.

[0120] In this way, the original geographic coordinates (latitude and longitude) are converted to a relative coordinate system centered on the GR station. Furthermore, the observation position of each SR beam can also be converted from the geographic coordinate system (WGS84) to a relative coordinate system centered on the GR station.

[0121] S33. Using the GR antenna height, the Earth's radius at the center of the Earth, and the SR's relative geographic information, calculate the coordinates and data of the SR on the elevation cone relative to the GR. The data here include all data including related auxiliary parameters such as radar reflectivity factor and time.

[0122] Based on the GR antenna height and the Earth's radius, the elevation angle β of the SR beam relative to the GR is calculated to further refine the position of the SR beam in the polar coordinate system:

[0123] Calculate the elevation angle β: Calculate the elevation angle of the SR beam based on the GR station's antenna height (which affects the signal propagation angle) and the relative position of the SR (converted to polar coordinates in S32). The elevation angle is generally the angle between the horizontal plane and the line connecting the GR station to the SR target.

[0124] Geocentric Earth Radius: Taking into account the curvature of the Earth, the Geocentric Earth Radius is used to more accurately calculate the distance and elevation angle to the target, ensuring the accuracy of the calculation.

[0125] Calculating Conical Coordinates: Elevation angle calculation is based on spatial geometry, particularly on a spherical surface. Elevation angle and distance together determine the spatial position of the SR beam. This data is converted to polar coordinates and further combined with azimuth and radial distance to produce the final coordinate information.

[0126] The polar coordinate data converted in S32 provides the necessary foundation for this step. The GR station's antenna height and geocentric radius are essential physical parameters for calculating elevation angles and spatial coordinates. These parameters are used to ultimately calculate the elevation angle and spatial position of the SR beam. Through these steps, or by using the haversine formula to obtain SR and GR distances, the geographic coordinate data (WGS84) is effectively converted to a polar coordinate system centered on the GR. This, combined with physical parameters such as the radar's antenna height and geocentric radius, allows for precise calculations, laying the foundation for further spatial analysis and visualization.

[0127] According to the quality control threshold standard system, the data application range is limited:

[0128] 1. Horizontal distance r factor: The spherical distance between the spaceborne radar SR beam and the ground-based radar GR site is calculated by the Haversine formula, that is, the radial distance r. r represents the radial distance in the polar coordinate system with GR as the center, that is, the horizontal distance.

[0129] 2. Vertical distance z factor: Based on the geometric relationship of the polar coordinate system, the vertical height z is calculated in combination with the elevation angle β and the horizontal distance r: z = r sin(β) + h GR ;

[0130] Where: β is the elevation angle calculated from the GR antenna height, the earth curvature and the relative position of the SR beam (refer to the above content); hGR: the antenna altitude of the ground-based radar GR.

[0131] If the earth curvature is ignored, it can be simplified as a plane approximation: z ≈ r tan(β)z.

[0132] According to the quality control system standard, the application range is limited, that is, only the data within the following range is retained for subsequent calculation:

[0133] Horizontal distance r: 15km < r < 120km; Vertical height z: 0km < z < 4km.

[0134] S4, based on the satellite zenith angle, the horizontal latitude and longitude information on each height layer of SR is corrected, including:

[0135] S41, the satellite observation field of view deviation distance is calculated by using the satellite zenith angle;

[0136] To calculate the satellite observation field of view deviation distance, the satellite zenith angle and the geometric relationship between the satellite and the ground target can be used. The zenith angle is the angle between the direction from the satellite to the ground target and the straight line directly above the satellite observation point. Generally, the larger the zenith angle, the more the observation distance (i.e. the distance between the satellite and the ground target) will be affected by the deviation.

[0137] If the height h of the satellite and the horizontal distance (also known as "ground projection distance") dz between the ground point and the projection point directly perpendicular to the satellite are known, the field of view deviation can be described by the relationship between the zenith angle and the distance: assuming that the satellite is on the orbit and the target point is on the ground, the deviation distance (deviation along the ground parallel line) between the satellite and the target can be estimated by the following formula:

[0138] dz = h tan(θz)

[0139] where dz is the deviation distance of the satellite observation field of view (horizontal distance), h is the satellite height to the ground, and θz is the satellite zenith angle.

[0140] The larger the zenith angle θz, the greater the horizontal deviation between the satellite and the ground target. This means that the satellite's field of view will deviate more from its vertical direction, resulting in more obvious deviation of the observation data. The deviation distance d of the satellite field of view is the horizontal distance between the actual position observed by the satellite and the projection point directly below the satellite.

[0141] For example: Assuming the satellite height is 800 km and the zenith angle is 30°, the field of view deviation distance d can be calculated as follows:

[0142] d = 800 * tan(30°) = 462.4 km

[0143] To calculate the deviation distance of the satellite observation field of view using the zenith angle, the key is to know the satellite height and the zenith angle, and the deviation distance of the satellite observation field of view can be obtained.

[0144] S42, based on the satellite scanning angle and the observation field of view deviation distance, respectively calculate the horizontal latitude and longitude field of view distance deviation on each height layer;

[0145] Calculate the deviation in longitude: Since the Earth's equatorial circumference is approximately 40,075 km (i.e. the total distance of 360° longitude), the field of view deviation in longitude can be calculated by the following formula:

[0146]

[0147] where ΔLongitude is the deviation in longitude (unit: degrees), d is the field of view deviation distance (unit: meters or kilometers), R is the Earth's radius (unit: meters or kilometers, usually 6371 km), and φ is the satellite observation point latitude (unit: degrees).

[0148] Calculate the deviation in latitude: The distance in the latitude direction changes with the change in latitude. Near the equator, the Earth's arc length is longer, while near the poles, the distance between latitudes is shorter. Therefore, the deviation in the latitude direction is fixed, and the calculation method is:

[0149]

[0150] where ΔLatitude is the deviation in the latitude direction (unit: degrees).

[0151] Example: Suppose there is a satellite with an orbital height of 800 km, a zenith angle of 30°, and the satellite is observing above the equator (latitude φ = 0°).

[0152] Calculate the longitude deviation:

[0153]

[0154] Calculate the latitude deviation:

[0155]

[0156] Therefore, for this satellite, the field of view deviation will shift approximately 0.0725° in each of the longitude and latitude directions.

[0157] By this method, the horizontal longitude and latitude field of view deviations for a satellite at different altitudes can be calculated. The deviations will vary for different satellite altitudes, zenith angles, and the latitude of the observation point. This method can be used to estimate the accuracy of satellite observation data and its impact on ground targets.

[0158] S43, Subtract the deviation from the SR ground horizontal longitude and latitude information to obtain the horizontal longitude and latitude information at each altitude layer by correction.

[0159] Subtracting the calculated deviation from the SR ground horizontal longitude and latitude information to correct to obtain the horizontal longitude and latitude information at each altitude layer is actually correcting the observation error (deviation) of the satellite from the original ground longitude and latitude data. The original ground longitude and latitude can be adjusted according to the previously calculated field of view deviation (including longitude and latitude direction deviation).

[0160] After calculating the deviation, the longitude and latitude information can be updated in the following ways:

[0161] Corrected latitude: φcorrected = φ0 - Δφ (Note: The latitude deviation is usually corrected in the positive and negative directions, and the specific direction depends on the sign of the deviation)

[0162] Corrected longitude: λcorrected = λ0 - Δλ (Similarly, the correction direction depends on the sign of the deviation)

[0163] Suppose the following conditions:

[0164] Original ground longitude and latitude: φ0 = 10°, λ0 = 20°

[0165] Original ground latitude correction: φcorrected = φ0 - Δφ = 0° - 0.0725° = -0.0725°

[0166] Original ground longitude correction: λcorrected = λ0 - Δλ = 20° - 0.0725° = 19.9275°

[0167] Through these steps, the SR ground latitude and longitude information of any height layer can be corrected to obtain a more accurate observation result. In actual operation, especially in places far from the satellite orbit or when the zenith angle is large, the correction deviation is particularly important.

[0168] S5. Adopting the improved radar reflectivity factor averaging method, the reflectivity factors of SR and GR are converted from logarithmic scale to linear scale respectively;

[0169] The radar reflectivity factor is an important parameter in meteorology, which is used to represent the reflection intensity of the target object to the radar signal. It is usually represented by logarithmic scale (dBZ). The improved radar reflectivity factor averaging method provided by the present application is: converted to dB by dBZ=10log(dB), then averaged, and converted to dBZ again. That is, first convert (dBZ) back to linear scale (Z), then average, and then convert back to (dBZ) representation.

[0170] For example, assuming there is (dBZ) data: [dBZ_1=30, dBZ_2=35, dBZ_3=40], the average (dBZ) value obtained by adopting the improved radar reflectivity factor averaging method is 36.74, instead of the directly averaged value (assuming the directly averaged value is ((30+35+40) / 3=35)). When processing multiple reflectivity data, if the (dBZ) is directly averaged, it may cause deviation. First convert (dBZ) to (Z) for averaging, and then convert back to (dBZ), which can provide more accurate average reflectivity factor.

[0171] Adopting the improved radar reflectivity factor averaging method, the reflectivity factors of SR and GR obtained are converted from logarithmic scale to linear scale respectively.

[0172] S6, the converted SR and GR reflectivity factors are matched in time and space to seconds;

[0173] The observation data of two different radar systems (SR and GR) are accurately matched according to the overlap in time and space, and finally the desired result is obtained. The core goal of each step is to accurately determine the time difference and spatial overlap between the two by mapping time and space, combining the observation information of different radars, so as to help further data analysis.

[0174] S61, screening the SR transit GR matching;

[0175] Determine whether the number of effective precipitation within 15-230km of the GR site is greater than 10. If it is less than 10, it is considered to be local sporadic precipitation, and the loop is exited to proceed to the next transit matching.

[0176] S62, extract the SR observation time for each row, assign the extended three-dimensional observation distance library time, time accurate to milliseconds, record as T SR .

[0177] Extracting the SR observation time for each row and assigning the extended three-dimensional observation distance library time usually involves expanding the two-dimensional scanning process to three-dimensional space to obtain observation data at different time and spatial positions.

[0178] Extract the latitude and longitude information of each orbit (scan line, each scan line corresponds to a set of latitude and longitude). Assign the time of each scan line to each spatial position. For example, for each row, according to the radar scanning mode and satellite speed, determine the spatial position (x1, y1) corresponding to each time point t. Assuming the radar performs N rows of scanning in total, then the time t can be used as the third dimension to construct a three-dimensional observation distance library. Each spatial point (x1, y1) corresponds to an observation time t and an observation distance D, which can be represented by the following three-dimensional data structure: D(t, x1, y1) = observation distance, distance changes with time.

[0179] Through the extraction of observation time for each row and the expansion of space-time, the two-dimensional scanning data of spaceborne radar can be converted into three-dimensional space-time data. This not only helps to understand the dynamic process of radar scanning, but also enables more complex observation analysis, such as data processing and analysis at different times and spatial positions.

[0180] S63, extract the GR observation time for each beam at each elevation angle, assign the extended three-dimensional observation distance library time, time accurate to milliseconds, record as T GR .

[0181] Similar to S62, but here for GR radar, extract the observation time for each beam at each elevation angle. These time data also need to be expanded into an observation library in three-dimensional space, and maintain millisecond-level accuracy, record as TGR. GR radar has multiple elevations and multiple beams, each beam corresponds to an observation position and time. Through this step, these information is also converted into a three-dimensional data structure, in order to carry out more accurate time and space matching.

[0182] Assigning SR observation time for each row to three-dimensional observation distance library time means expanding the scanning time information of each observation row to the area scanned by the row, and giving each data point in the three-dimensional data structure a corresponding timestamp. This approach helps to establish a connection between time, space and observation data, forming a complete spatio-temporal data set.

[0183] S64, calculate the distance library from the SR to the center of the GR site, and the SR beam time at that distance is the SR transit GR time T SR0 .

[0184] The closest distance point is found by calculating the distance between the SR radar and the center of the GR station, and the beam observation time of the SR radar at this point is taken as the time TSR0 when the SR passes through the GR station. This step involves matching of spatial positions. The station coordinates of the SR and the GR are compared, and the SR observation point closest to the GR station is selected to determine the transit time when the SR passes through the GR station. This time value TSR0 will serve as the basis for subsequent calculations.

[0185] S65, calculate the average of the first three elevation angle scan times of the GR, marked as T GR0 .

[0186] In order to obtain a higher precision time match between the GR and the SR, the observation times of the first few elevations of the SR when passing through the GR station need to be averaged. This step averages the scan times of the first 3 elevations to obtain TGR0 as the reference time for the time match between the GR and the SR.

[0187] Generally, the GR radar scans different elevations at certain time intervals, and after knowing the scan time or scan period of each elevation, since the GR radar has multiple elevations, the scan time of each elevation may be different. By averaging the times of the first 3 elevations, the influence of accidental factors can be reduced, making the matched time more accurate.

[0188] S66, calculate the time difference between the GR and the SR Select the smallest GR base data from the GR base data within the past 10 minutes Determine whether the selected GR base data is within 6 minutes, if greater than 6 minutes, exit the matching.

[0189] Calculate the time difference between TGR0 and TSR0 Evaluate the time difference between the GR and the SR. Then, in the base data of the GR station, select the base data closest to the SR, and check whether the time difference is within a reasonable range (within 6 minutes). If the time difference is greater than 6 minutes, exit the matching process, indicating that this matching is invalid.

[0190] The core of this step is the calculation and reasonableness judgment of the time difference. If the time difference exceeds 6 minutes, it means that the time precision of the matching is not enough, and there may be a large deviation between the data, so this matching process needs to be skipped. Through this step, it is ensured that the selected matching data is accurate.

[0191] S67. Calculate the time difference of the overlapping spatial matching between the GR and the SR observation. Select each overlapping spatial range for matching, calculate the average time of the GR and the SR within the overlapping space respectively, and calculate the time difference between the two ​The time accuracy is accurate to seconds.

[0192] By overlapping the range in space, the time difference of GR and SR observation data in each space is calculated (the time accuracy is accurate to seconds), and the time deviation factor is screened out The overlapping space range within the threshold (-90s<Δt<90s) is the selected space overlapping range. Through the sensitivity test established by the pre-value system, it is found that the time deviation within 90 seconds represents the observation of the same precipitation, which is one of the purposes of establishing the pre-standard system.

[0193] S7, for each overlapping space range matched, the SR and GR reflectivity factors are respectively improved by inverse distance weighting to improve the mean accuracy;

[0194] S71, according to the unified polar coordinate projection of GR and SR, first from the SR beam distance library cycle, and then from the GR elevation cycle. Traverse each distance library corresponding to the fixed GR elevation relationship.

[0195] The observation data of SR and GR are projected and matched through the unified polar coordinate system to ensure that they can be compared in the same coordinate system, including:

[0196] Unified polar coordinate projection: the observation data of GR and SR are located in different spatial positions and coordinate systems. In order to ensure that they can be effectively compared and matched, their coordinate systems need to be unified first, and they are converted into the same polar coordinate system. The definition of polar coordinate system is usually based on a center point, which is usually the GR station or the SR station.

[0197] SR beam distance library cycle: the observation data of SR radar is usually divided into multiple beams according to distance. Each beam has a certain coverage range, and the observation information of the beam can obtain data of different positions. SR beam distance library is to organize these beams according to different distances to form a library.

[0198] GR elevation cycle: the observation data of GR radar is divided according to elevation. Each elevation corresponds to an observation beam to obtain data of a specific area. Therefore, the elevation cycle of GR is to process the observation data of different elevations one by one.

[0199] Traversal of fixed GR elevation relationship: by traversing the relationship between each SR beam distance and GR elevation, a corresponding relationship between SR and GR can be constructed to ensure that each SR beam distance and fixed elevation GR beam can be matched in space.

[0200] S72. Obtain the cross-space of the star-ground radar beam by time-space matching, and use inverse distance weighted average for the matched GR distance library within the 5 km range of the SR footprint area, and use inverse distance weighted average for the matched SR distance library within the GR elevation beam widening range.

[0201] Weighted average is performed on the cross-space matched by SR and GR radars in time-space, ensuring that the data matched in S71 is more accurate and reliable, including:

[0202] Cross-space of star-ground radar beam: In the matching process of SR and GR, the observation range of each must be considered. Through time-space matching, the cross-space between SR and GR is obtained, i.e., the part where their observation areas overlap. This cross-space determines which data can be used for weighted average calculation.

[0203] Inverse distance weighted average for GR distance library within 5 km range of SR footprint area: After obtaining the cross-space of SR and GR, inverse distance weighted average is performed on the distance library of GR radar observation points within 5 km range of the SR footprint area (i.e., within the observation area range of SR radar). Inverse distance weighted average is based on the distance between each data point and the target point: the closer the distance, the greater the weight, and vice versa. This method ensures that the closer GR data has a greater impact on the result, ensuring spatial consistency of the data.

[0204] Inverse distance weighted average for SR distance library within the GR elevation beam widening range: Similar to GR data, inverse distance weighted average is also needed for SR radar observation data within the elevation beam range of GR radar. Here, "elevation beam widening range" refers to the area covered by GR radar due to different elevations, and SR data within this range also uses inverse distance weighting to ensure spatial consistency of the weighted result.

[0205] The method of inverse distance weighted average further refines the matching results of observation data of SR and GR radars, ensuring that the data within the cross-space is weighted more reasonably, thereby improving the accuracy and reliability of data matching.

[0206] The present application adopts high-precision time matching method improvement technology: including accurate to 1s pixel time difference matching method. The time average of GR and SR within the independent cross-pixel space obtained by effective irradiation volume method is used to perform difference, and the maximum matching time difference ΔT at 230 km is within 1s.

[0207] High-precision spatial matching method improvement technology: including SR and GR respectively using inverse distance weighted average method. The antenna gain of the SR pixel center is twice that of the edge, and the inverse distance weight is used to improve the accuracy of the mean value of the reflectivity factor of GR and SR within the irradiation volume.

[0208] Firstly, in step S6, the method of matching the GR single elevation scan time with the SR time closest to the GR in distance (transit time) is adopted. Specifically, the scan time of the GR stored in the radar data needs to be extracted and converted to datetime format for calculation. At the same time, the transit time of the SR, i.e. the time of the point closest to the GR in distance, can be directly read from the SR data set. Next, the time difference between each elevation scan time of the GR and the SR transit time is calculated, and the GR scan time with the smallest time difference is found as the matching time point.

[0209] Then, in step S7, the method of matching the pixel time according to the cross-overlapping space of the GR and SR beams is adopted. First, the overlapping area of the GR and SR in space is determined, which can be achieved by calculating their distance and azimuth angle, and the SR data is projected into the coordinate system centered on the GR. Next, for each pixel of the GR and SR in the overlapping area, its observation time is extracted. These time data are usually stored in the respective data sets. Finally, the mean values of the pixel times of the GR and SR in the overlapping area are calculated, and the time difference of these mean values is calculated to ensure that the maximum matching time difference AT is within 1 second.

[0210] Through these two matching methods, the improvement from the traditional transit time-based matching to the high-precision time matching based on the cross-overlapping space can be achieved. Step S6 provides a basic matching method, while step S7 further improves the accuracy and reliability of the matching, ensuring the consistency of the GR and SR data in time and space, and providing a more accurate basis for subsequent data analysis and application.

[0211] S8, reduce the spatial heterogeneity of SR and GR in the matching space by using the non-uniform beam filling quality control technology;

[0212] S81, GR non-uniform beam filling effect improvement technology

[0213] The minimum detection reflectivity factor threshold of the GR is set to Zg* = 15 dBZ, and for each volume-matched sample, the proportion of the GR distance library in the volume that satisfies Zg ≥ Zg* is recorded as fg.

[0214] The influence of GR non-uniform beam filling in the matching volume is reduced by excluding samples with fg less than the beam filling percentage factor, where fg = 0.9.

[0215] S82, SR non-uniform precipitation filling effect improvement technology

[0216] The minimum detectable reflectivity factor threshold of SR is set as Zs* = 15dBZ, and for each volume-matched sample, the proportion of the distance bin in the volume that satisfies Zs >= Zs* is recorded as fs.

[0217] The influence of non-uniform beam filling of SR in the matching volume is reduced by excluding samples with fs less than the beam filling percentage factor, where fs = 0.9.

[0218] In order to avoid the difference caused by the minimum detectable sensitivity as much as possible, the minimum threshold of reflectivity factor of GR and SR is uniformly set to 15dBZ.

[0219] S83, non-uniform beam filling quality control technology in the pixel of SR

[0220] The proportion of the distance bin with the pixel non-uniform beam correction parameter paramNUBF of SR greater than 0.1 to the distance bin of SR in the volume is fs_nubf.

[0221] The influence of non-uniform beam filling in the pixel area of SR is reduced by excluding samples with fs_nubf less than the beam filling percentage factor, where fs_nubf = 1.0.

[0222] The invention uses non-uniform beam filling quality control technology in the pixel, which significantly reduces the spatial heterogeneity of precipitation.

[0223] S9, the precipitation type, precipitation phase state and bright band factor of SR in the matching space are respectively controlled, and the influence on the evaluation accuracy is reduced;

[0224] S91. Precipitation type quality control factor establishment.

[0225] The precipitation type observed by SR is divided into convective precipitation and stratiform precipitation. For each sample in the effective irradiation volume, the proportion of stratiform precipitation in the volume is recorded as f_stratiform.

[0226] The influence of SR precipitation type on accuracy evaluation is reduced by excluding samples with f_stratiform less than the precipitation classification factor, where f_stratiform = 1.0.

[0227] S92. Precipitation phase state quality control factor establishment.

[0228] The precipitation phase state observed by SR is divided into liquid state, solid state and mixed state. For each sample in the effective irradiation volume, the proportion of convective precipitation in the volume is recorded as f_liquid.

[0229] The influence of SR observed precipitation bright band in the matching volume is reduced by excluding samples with f_liquid less than the phase classification factor, where f_liquid = 1.0.

[0230] S93. Bright band quality control factor is established.

[0231] SR observed precipitation is divided into yes and no according to the bright band, and the bright band is set to 1 when there is a bright band, and 0 when there is no bright band. For each sample in the effective irradiation volume, the bright band identification factor f_bb in the volume is recorded.

[0232] The influence of SR observed precipitation bright band in the matching volume is reduced by excluding samples with f_bb equal to 0, where f_bb = 1.0.

[0233] S10, the reflectivity factor of SR and GR is converted from linear scale to logarithmic scale by using improved radar reflectivity factor averaging method;

[0234] After the above S5 step, the radar reflectivity factor dBZ is converted to dB. Here, the improved radar reflectivity factor averaging method is used to convert the average value of the reflectivity factor dB after S9 from linear scale to logarithmic scale by formula to convert to dBZ, which effectively reduces the calculation deviation caused by direct averaging of dBZ.

[0235] S11, based on the comparison and verification of the converted SR reflectivity factor and the GR reflectivity factor, the ground-based radar GR calibration bias estimation is carried out;

[0236] S111. Star-ground precipitation radar consistency. By comparing the average GR radar reflectivity factor Zg and the SR radar reflectivity factor Zs, ΔZ (ΔZ = Z g - Z s ) evaluates the consistency of star-ground precipitation radar.

[0237] S112 combines the ground-based precipitation radar calibration maintenance record twice a year, usually once before the flood season and once after the flood season. By statistically analyzing the time series of ΔZ during the calibration period of the ground-based radar, the calibration error of the ground-based radar is evaluated.

[0238] The southern region of China belongs to the monsoon climate of the southern subtropical region with oceanic characteristics, and the strong convective weather events of typhoon rainstorm occur frequently during the flood season period (April-October) every year. The three dual-polarization S-band radars (Guangzhou, Shenzhen and Zhaoqing, hereinafter referred to as GZ, SZ and ZQ) in the Guangdong-Hong Kong-Macao Greater Bay Area selected by the present application are arranged in a triangular shape, close to each other, and the straight-line distance is only 56, 125 and 82 km respectively. According to the effective observation range of 230 km, the effective observation area of a single radar is 1.7 x 10 6 km 2, , and the area observed by the three radars together is 1.1 x 10 6 km2 The common observation area of the three radars is large and has high overlap, which enables them to effectively observe strong precipitation events in the same area and provide reliable ground verification data for spaceborne radars, thereby being suitable for ground-based radar network verification. The three radars account for 65% of the effective observation range of a single radar, indicating that the three radars can approximately represent strong precipitation events in the same area and are suitable for ground-based radar network verification using DPR. In the present application, the technical effects of the three-radar network observation mainly manifest in the following two points:

[0239] 1. Network observation: The three radars are arranged in a triangular shape and close to each other, with straight-line distances of 56 km (Guangzhou-Shenzhen), 125 km (Guangzhou-Zhaoqing), and 82 km (Shenzhen-Zhaoqing), respectively. This layout enables their observation areas to have a large overlap and form a dense observation network, thereby improving the temporal and spatial resolution and monitoring accuracy in the same area;

[0240] 2. Ground verification: The common observation area of the three radars is large and has high overlap, which enables them to provide reliable ground verification data for spaceborne radars. By comparing with the data of spaceborne radars, the precipitation estimation algorithm of spaceborne radars can be verified and improved.

[0241] In the above technical solution, the quality control factor is applied in multiple steps of the present application for quality control of GR and SR data to reduce estimation bias, including: application of horizontal distance r factor in S32 step, application of vertical distance z factor in S32 step, application of non-uniform beam filling factor in S7 step, application of precipitation classification factor in S81 step, application of precipitation phase factor in S82 step, application of precipitation bright band factor in S83 step, application of time bias factor in S5 step, and application of reflectivity factors Zg and Zs: In the matching of spaceborne and ground radars, data with intensity below 15 dbz is filtered out in advance due to the different sensitivities of the two. After the space-time matching of spaceborne and ground radars is completed, it is found through experiments that the data above 40 dbz has large differences between the two, which is caused by the instruments, so the data above 40 dbz is also excluded. Therefore, the final result is only the comparison of data between 15-40 dbz of reflectivity factors Zg and Zs. Based on the sensitivity error source analysis, the threshold standard system of spaceborne and ground precipitation radar quality control factors is determined through sensitivity experiments of multiple factors, and the systematic scientific quality control standard system is applied to the evaluation data set to improve the consistency test accuracy of spaceborne and ground precipitation radar reflectivity factors.

[0242] As shown in the following Table 1 of spaceborne and ground precipitation radar quality control factor and scheme optimization combination experiment:

[0243] Table 1 Spaceborne and ground precipitation radar quality control factor and scheme optimization combination experiment

[0244]

[0245]

[0246] The above table is first sequentially do 1 to 4 quality control factor experiments, on this basis, respectively, do 5 to 13 quality control factor experiments in order, and finally give is the result of quality control factor optimization combination.

[0247] Compared with dBZ direct average, the BIAS of dBZ converted to dB and then averaged and converted back to dBZ increases by 1.6 dB. The reason is that dBZ is actually compressed for values above 10 dB, and direct averaging without conversion will significantly reduce the actual BIAS value. Similarly, the BIAS of SZ radar averaged dBZ is -0.67 dB, and after conversion to dB and then averaged, the BIAS increases to -0.17 dB (table omitted). Converting dBZ to dB, averaging, and then converting back to dBZ can correct the BIAS by about 1.6 dB, proving that direct averaging of dBZ leads to a lower BIAS.

[0248] The time offset factor is the largest factor affecting the correlation coefficient. When the time is set to within 90 s, the ΔZ variability is significantly reduced, and the CC reaches 0.93. For example, Figure 2 shows the sensitivity of the reflectivity factor difference ΔZ (ΔZ = Z g - Z s ) of the star-ground radar to the time offset ΔT, where (a) is GZ, (b) is SZ, (c) is ZQ, (d) is three superimposed, (e) is stratiform, and (f) is convective. It is found that ΔZ has a large divergence with increasing time offset ΔT, especially between 200 s and 300 s, with a maximum ΔZ of more than 24 dB. Figures e-f reveal that the ΔZ of convective precipitation is significantly more variable than that of stratiform precipitation. Obviously, compared with stratiform clouds, the rapid development and movement speed of convective clouds have an impact on the matching accuracy of star-ground radar reflectivity factors.

[0249] In terms of phase classification test indicators, liquid is better than solid, and mixed state is better than solid. Therefore, setting the vertical height from 0 to 4 km (the bottom of the Guangdong bright band is 4.1 km) has the greatest impact on BIAS, and BIAS decreases to 0.87 below 4 km, which is basically liquid precipitation. The difference standard deviation STD of stratiform precipitation is the smallest at 2.23, and the STD of convective precipitation is the largest at 3.6, indicating that unstable convective precipitation has a large ΔZ divergence, while using stable precipitation can significantly reduce the divergence of ΔZ.

[0250] As shown in Table 2 below, the present invention provides a standard system for the threshold values ​​of quality control factors for satellite-to-ground precipitation radar. Table 2 shows the threshold values ​​composed of eight main quality control factors determined after the experiment. By setting these parameter thresholds, the reflectivity error of satellite-to-ground radar inspection can be effectively reduced.

[0251] Table 2: Standard system for quality control factor thresholds of satellite-to-ground precipitation radar

[0252]

[0253] The present invention provides a method for estimating the calibration deviation of a ground-based radar based on the calibration of a satellite-to-ground precipitation radar. The method actually includes two parts: the first is a pre-processing part, i.e., the establishment of a standard system for the quality control factor threshold of the satellite-to-ground precipitation radar; the second is an application part, i.e., the application of the standard system for the quality control factor threshold of the satellite-to-ground precipitation radar to estimate the calibration deviation of the ground-based radar based on the calibration of the satellite-to-ground precipitation radar.

[0254] After the quality control factor threshold experiment, the threshold standards of 8 types of quality control factors were set to effectively reduce the reflectivity deviation of satellite-to-ground radar inspection. Figure 3 As shown in the figure, the statistical probability density distribution of the difference ΔZ between the reflectivity factors of three ground-based radars (GR) and spaceborne radars (SR) in the Guangdong-Hong Kong-Macao Greater Bay Area (Guangzhou GZ, Shenzhen SZ, and Zhaoqing ZQ) using the method of the present invention is statistically analyzed. Part b of the figure shows the estimated statistical probability density distribution of the calibration deviation between the reflectivity factors of the spaceborne radar SR and the Guangzhou GZ ground-based radar GR after quality control processing using the method of the present invention through the highest quality control factor threshold standard. Part d of the figure shows the estimated statistical probability density distribution of the calibration deviation between the reflectivity factors of the spaceborne radar SR and the Shenzhen SZ ground-based radar GR after quality control processing using the method of the present invention through the highest quality control factor threshold standard. Part f of the figure shows the estimated statistical probability density distribution of the calibration deviation between the reflectivity factors of the spaceborne radar SR and the Zhaoqing ZQ ground-based radar GR after quality control processing using the method of the present invention through the highest quality control factor threshold standard. Statistics show that the correlation CC of ΔZ based on the three ground-based radars reaches between 0.9 and 0.95, proving that the combined threshold standard can improve the matching accuracy of satellite-ground radars and indirectly enhance the accuracy of authenticity verification. Statistics show that the standard deviation (STD) of the ΔZ differences between the three ground-based radars in the Greater Bay Area ranges from 1.23 to 1.62 dB. The overall sampling points are stably concentrated near the fitted line, with few scattered sampling points, indicating a high level of consistency in the matching of satellite-ground radar data. For example, the ΔZ correlation CC for the GZ radar is 0.95, with an STD of only 1.23 dB, a significant improvement over traditional statistical conclusions (the ΔZ correlation CC from 2016 to 2019 was 0.91 and the STD was 2.05 dB).

[0255] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely divergences of the principles and spirit of the application and that numerous modifications, changes, substitutions, and alterations can be made thereto without departing from the spirit and scope of the application as defined by the appended claims and their equivalents.

[0256] The above description of the application and its embodiments is not restrictive, and the embodiments shown in the drawings are only one of the embodiments of the application, and the actual structure is not limited thereto. In general, if a person skilled in the art is inspired thereby, without departing from the purpose of the application, without creative design, similar structure and embodiments to the technical solution should belong to the protection scope of the application.

Claims

1. A method for estimating ground-based radar calibration bias based on satellite-ground precipitation radar calibration, characterized in that: It includes: S1. Obtain the original data of reflectivity factors of space-borne radar SR and ground-based radar GR, and perform quality control on the original data; S2. Perform frequency correction from Ku band to S band on the SR reflectivity factor after quality control; S3, performing polar coordinate projection transformation on the corrected SR reflectivity factor with GR as the center; S4. Correct the horizontal latitude and longitude information at each SR altitude layer based on the satellite zenith angle; S5. Use the improved radar reflectivity factor averaging method to convert the reflectivity factors of SR and GR from logarithmic scale to linear scale, specifically: The reflectivity factors of SR and GR are converted from logarithmic scale dBZ to linear scale dB using the expression dBZ = 10logdB; S6. Perform temporal and spatial overlap matching of the converted SR and GR reflectivity factors with an accuracy of seconds; S7, for the SR and GR reflectivity factors within each overlapping spatial range of the matching, improve the mean accuracy by using the inverse distance weighting; S8. Reduce the spatial heterogeneity of precipitation in the matching space of SR and GR through the intra-pixel non-uniform beam filling quality control technology, including: S81, GR non-uniform beam filling effect improvement technology; The minimum detection reflectivity factor threshold of GR is set to Zg*=15dBZ. For each volume-matched sample, the proportion of GR distance libraries that satisfy Zg ≥ Zg* in the volume is recorded as fg; The effect of GR non-uniform beam filling in the matching volume is reduced by excluding samples whose fg is less than the beam filling percentage factor, where the non-uniform beam filling factor fg = 0.9; S82, SR non-uniform precipitation plugging effect improvement technology; The minimum detection reflectivity factor threshold of SR is set to Zs*=15dBZ. For each volume-matched sample, the proportion of SR distance libraries that satisfy Zs ≥ Zs* in the volume is recorded as fs; The effect of SR non-uniform beam filling in the matching volume is reduced by excluding samples with fs less than the beam filling percentage factor, where the non-uniform beam filling factor fs = 0.9; S83 and SR pixel non-uniform beam filling quality control technology; The ratio of the distance library with the pixel non-uniform beam correction parameter paramNUBF greater than 0.1 to the SR distance library in the volume is fs_nubf; The effect of non-uniform beam filling within the SR pixel area is reduced by excluding samples where fs_nubf is less than the beam filling percentage factor, where fs_nubf = 1.0; S9. Perform quality control on the precipitation type, precipitation phase and bright band factor of SR in the matching space to reduce the impact on the evaluation accuracy; S10, using the improved radar reflectivity factor averaging method, convert the reflectivity factors of SR and GR from linear scale to logarithmic scale; S11. Estimating the ground-based radar GR calibration deviation based on the comparison and verification of the converted SR reflectivity factor and the GR reflectivity factor.

2. The method for estimating ground-based radar calibration deviation based on satellite-to-ground precipitation radar calibration according to claim 1, characterized in that: The S1, respectively obtaining the original data of the reflectivity factors of the spaceborne radar SR and the ground-based radar GR, and performing quality control on the original data, includes: S11. obtaining raw data of radar reflectivity factor of spaceborne radar SR; S12. Quality control of the original data of the spaceborne radar SR reflectivity factor obtained by dual-frequency observation quality, precipitation classification quality and dual-frequency precipitation identification parameters; S13, obtaining original data of radar reflectivity factor of ground-based radar GR; S14. Perform quality control on the GR reflectivity factor by eliminating the calibration distance data.

3. The method for estimating ground-based radar calibration deviation based on satellite-to-ground precipitation radar calibration according to claim 1, characterized in that: S2, performing frequency correction from Ku band to S band on the SR reflectivity factor after quality control, includes: S21, obtain raindrop number concentration Nw, precipitation particle phase phase and temperature T data from SR product data; S22. Calculate the backscattering cross section σ of the rainfall particles based on the integral of the acquired data values b And the corresponding scattering function fz(Dm); The scattering function fz(Dm) is: ； ； Where λ is the wavelength of the S-band radar, Kw is the complex refractive index of water, T is the temperature, and σ b is the backscattering cross section of the rainfall particles, D is the diameter of the precipitation particles, and Dm is the median diameter of the precipitation particles; S23, combined with the raindrop number concentration Nw provided in the product and fz(Dm) calculated in S22, calculate the radar reflectivity factor Z of the S band s , realize frequency correction; Corrected S-band radar reflectivity factor Z s for: Z s =Nwfz(Dm); Where Nw is the raindrop number concentration, Dm is the median diameter of precipitation particles, and fz(Dm) is the backscattering function in the S band.

4. The method for estimating ground-based radar calibration deviation based on satellite-to-ground precipitation radar calibration according to claim 1, characterized in that: The S4, correcting the horizontal latitude and longitude information at each SR altitude layer based on the satellite zenith angle, includes: S41. Calculate the satellite observation field deviation distance using the satellite zenith angle; S42. Calculate the field of view distance deviation of the horizontal latitude and longitude at each altitude layer based on the satellite scanning angle and the observed field of view deviation distance; S43. Subtract the field of view distance deviation from the SR ground horizontal longitude and latitude information, and obtain the horizontal longitude and latitude information of each altitude layer through correction.

5. The method for estimating ground-based radar calibration deviation based on satellite-to-ground precipitation radar calibration according to claim 1, characterized in that: S6, performing temporal and spatial overlap matching of the converted SR and GR reflectivity factors with an accuracy of seconds, includes: S61, screening SR transit GR matching; S62, extract the filtered SR line-by-line observation time, assign it to the three-dimensional observation distance library time, and record it as T SR； S63, extract the time of each beam of GR elevation angle, assign it to the three-dimensional observation distance library time, and record it as T GR； S64. Calculate the distance library with the closest distance from the SR to the GR site center, and record the SR beam time of the distance library as the SR transit GR time T SR0； S65. Calculate the average time of the first three elevation scans of SR, marked as T GR0； S66. Calculate the time difference between GR and SR ∇T = T GR0- T SR0 , select the GR basis data with the smallest ∇T from the GR basis data and perform the screening; S67. Filter out each matching overlapping spatial range, calculate the average time of GR and SR in the overlapping space respectively, and calculate the time difference ∇T between the two with time accuracy to seconds, and determine the selected spatial overlapping range.

6. The method for estimating ground-based radar calibration deviation based on satellite-to-ground precipitation radar calibration according to claim 1, characterized in that: S7, for each SR and GR reflectivity factor within each overlapping spatial range of the matching, respectively improving the mean accuracy by using the inverse distance weight, includes: S71. Based on the unified polar coordinate projection of GR and SR, first loop through the SR beam range library, then loop through the GR elevation angle, and traverse the relationship between each range library and the fixed GR elevation angle. S72. Obtain the space of the satellite-to-ground radar beam intersection through cyclic space-time matching. Use the inverse distance weighted average for the matched GR range library within the 5 km SR footprint area. Use the inverse distance weighted average for the matched SR range library within the GR elevation beam width range.

7. The method for estimating ground-based radar calibration deviation based on satellite-to-ground precipitation radar calibration according to claim 1, characterized in that: S9, respectively, performs quality control on the precipitation type, precipitation phase, and bright band factor of the SR in the matching space to reduce the impact on the evaluation accuracy, including: S91. For each SR sample within the matching volume, record the fraction of stratiform precipitation within the volume as f_stratiform. Reduce the impact of SR precipitation type on the accuracy assessment by excluding samples where f_stratiform is less than the precipitation classification factor. S92. For each SR sample within the matching volume, record the proportion of convective precipitation within the volume as f_liquid; reduce the influence of the SR precipitation phase within the matching volume by excluding samples whose f_liquid is less than the phase classification factor; S93. For each SR sample within the matching volume, record the bright band identification factor within the volume as f_bb, and reduce the influence of the precipitation bright band observed by the SR within the matching volume by excluding samples with f_bb equal to 0.

Images