Networking cooperative rain detection radar data quality control method

Abstract

The invention discloses a networking cooperative rain detection radar data quality control method, which is a networking cooperative rain detection radar data quality control method based on sparse reconstruction and a local oversampling single station. Comprising a single-station quality control implementation step and a networking cooperative quality control implementation step. The method can realize full-chain automatic quality control, has the advantages of being outstanding in complex terrain and interference suppression capability, improving product quality through networking collaborative optimization and forming a closed-loop lifting mechanism, is high in adaptability and wide in application scene, can be compatible with multiple types of rain detection radar systems such as dual-polarization, Doppler and phased array systems, and is suitable for popularization and application. The method is suitable for a plurality of business scenes such as meteorology, hydrology, emergency monitoring and the like, and has wide engineering application value and popularization prospect.

Description

A method for controlling the data quality of networked collaborative rain-measuring radar Technical Field

[0001] This invention relates to a data quality control method for networked collaborative rainfall measurement radar, which is a data quality control method for networked collaborative rainfall measurement radar based on sparse reconstruction and local oversampling of a single station. Background Technology

[0002] Data quality control for rain-measuring radar is the core foundation for ensuring the accuracy of radar quantitative precipitation inversion and the reliability of hydrological applications. Its technological development directly affects the timeliness and accuracy of severe weather warnings. International research started earlier and has formed a multi-dimensional quality control technology system covering single stations and networks. At the single-station level, early methods relied on physical characteristics and empirical rules, using techniques such as distance attenuation correction, velocity domain filtering, and ground clutter removal to reduce the influence of non-precipitation echoes. Later, methods such as pulse repetition frequency folding processing, zenith echo analysis, and distance ambiguity correction were combined to further improve the physical rationality of the data. With the application of electromagnetic polarization technology, researchers have constructed abnormal echo identification models using polarization variables such as coherence coefficient and differential reflectivity to distinguish and attenuate hail, sleet, and ground clutter. However, a mature solution has not yet been developed to address the data quality degradation caused by environmental interference and radio frequency interference. At the network level, foreign countries have developed network-level quality control and deviation correction methods such as clutter map construction, ground echo filtering, radar grid mosaicking, and rain gauge assimilation, providing technical support for large-scale precipitation monitoring.

[0003] Domestically, with the rapid deployment of Doppler phased array radar and dual-polarization radar, research on quality control of rain gauge data has entered a stage of rapid development. Research directions focus on static clutter map construction, real-time removal algorithms based on empirical thresholds and polarization filtering, beam jamming identification in complex terrain, and bias correction and assimilation schemes for radar-rain gauge fusion. In engineering applications, emphasis is placed on the real-time performance of algorithms, platform adaptability, and robustness in complex terrain. Meanwhile, research on specific quality control technologies for flash flood monitoring, such as small-scale strong convection identification, local terrain echo differentiation, and rapid X-band attenuation compensation, has become a research hotspot in recent years.

[0004] Currently, the quality control of rainfall radar data is transforming from traditional single-station physical processing to multi-station gridding, polarization-enabled, and intelligent approaches. However, flash flood monitoring and high spatiotemporal resolution precipitation estimation place more stringent demands on data quality, and existing technologies still have significant shortcomings: beam blocking and multipath effects in complex terrain and densely populated urban areas lead to observation blind spots and frequent false echoes, while environmental interference and radio frequency interference further exacerbate data distortion. More importantly, existing quality control methods lack convenient, automated, and standardized operational implementation paths, relying on manual analysis and experience-based judgment, resulting in low efficiency, poor repeatability, and insufficient correction accuracy. Therefore, it is urgent to construct a standardized and intelligent single-station-network collaborative quality control method to achieve automatic identification, accurate location, and effective correction of data anomalies, terrain obstruction, radio frequency interference, and other problems. This will provide a basis for subsequent calibration, parameter optimization, and equipment upgrades, and promote the engineering application and innovative development of rainfall radar technology in complex environments. Summary of the Invention

[0005] This invention addresses the shortcomings of existing rain measurement radar data quality control technologies by proposing a network-based collaborative rain measurement radar data quality control method, which is based on sparse reconstruction and local oversampling of single stations.

[0006] To achieve the above objectives, the solution of the present invention is as follows:

[0007] A method for controlling the data quality of a networked collaborative rain-measuring radar includes data acquisition from rain-measuring radars in a networked observation system, wherein the networked observation system consists of dual-polarization rain-measuring radars at multiple observation stations, and the control method includes single-station quality control implementation steps and networked collaborative quality control implementation steps.

[0008] The single-station quality control implementation steps include:

[0009] Step 1: Raw data preprocessing: The signal-to-noise ratio (SNR) estimation algorithm is used to screen the effective data collected and remove low-quality data with an SNR of <3dB.

[0010] Step 2, External Calibration: Based on the UAV calibration data, correct the radar system deviation to ensure that the reflectivity factor measurement error is ≤1dBZ;

[0011] Step 3, Clutter Suppression: Apply clutter identification algorithms and Gaussian model adaptive filtering to filter clutter from the acquired data, achieving a clutter identification rate of ≥95%.

[0012] Step 4: Attenuation correction: For X-band radar data, an attenuation correction model is used to compensate for rainfall attenuation. The correlation between the corrected data and rain gauge observations is ≥0.9.

[0013] Step 5: Phase denoising and velocity ambiguity correction: Phase data is processed using a moving average filter and combined with a deambiguity algorithm to recover the radial velocity, with a velocity error ≤1m / s;

[0014] The network collaborative quality control implementation steps include:

[0015] Step 1: Data spatiotemporal registration: Convert the data after quality control of each station to the UTM coordinate system, with a time synchronization error of ≤1 minute;

[0016] Step 2: Multi-station data consistency check: Calculate the data consistency index CI for overlapping areas of multiple stations, and perform secondary correction on data with CI < 0.8 based on valid neighboring data;

[0017] Step 3: Weight Allocation: Based on the single-station clutter identification rate, signal-to-noise ratio, and correction accuracy, the single-station quality level is divided into three levels: A, B, and C, with corresponding weights of 0.6, 0.3, and 0.1, respectively. A comprehensive weight is generated by combining the distance weight, where the closer the distance, the greater the weight.

[0018] Step 4, Blind Spot Complementation and Integration: For single-station blind spots, distance-weighted interpolation is performed using effective data from three adjacent stations; for overlapping areas, a comprehensive weighted average fusion is used to generate a network mosaic product.

[0019] Step 5, Quality Diagnosis: Output a network product quality report, mark abnormal areas, and provide a basis for site optimization. The marked abnormal areas include areas of continuous obstruction and high-frequency interference. The basis for site optimization includes suggesting increasing the antenna height in areas of continuous obstruction.

[0020] The solution further includes: In the first step of the single-station quality control implementation process, the signal-to-noise ratio estimation algorithm is calculated using the following formula:

[0021] ;

[0022] in: For signal power, For noise power, when Data marked as low quality is removed or enhanced.

[0023] The solution further includes: in the second step of the single-station quality control implementation process, the correction formula for correcting radar system deviation is:

[0024] ;

[0025] in: To observe differential reflectance; This is a systematic bias, determined through external calibration tests.

[0026] The solution further includes: in the third step of the single-station quality control implementation process, the calculation formula for clutter identification is:

[0027] ;

[0028] in: The standard deviation of the reflectivity factor. Mean reflectivity factor The horizontal-vertical polarization correlation coefficient is given when... When the signal is detected as clutter, T is the threshold value.

[0029] The calculation formula for the Gaussian model adaptive filtering is as follows:

[0030] ;

[0031] in: The original reflectivity factor, Radial velocity, Center of clutter velocity This represents the standard deviation of clutter velocity.

[0032] The solution further includes: in the fourth step of the single-station quality control implementation process, the attenuation correction model is:

[0033] ;

[0034] in: To observe the reflectivity factor, The specific attenuation coefficient, The target distance.

[0035] The solution further includes: In the single-station quality control implementation steps, the formula for the moving average filtering process in step five is:

[0036] ;

[0037] in: This is the raw phase data; This represents the half-width of the sliding window, with a value ranging from 3 to 5.

[0038] The solution further includes the following: In the second step of the network collaborative quality control implementation process, the formula for calculating the consistency index (CI) is:

[0039] ;

[0040] in: , The reflectance factor for adjacent stations on the same grid is defined as follows: when CI < 0.8, it is considered inconsistent data and a secondary correction is triggered.

[0041] The solution further includes the following: In the fourth step of the network collaborative quality control implementation process, the spatial resolution of the network mosaic is 1km×1km, and the time resolution is 5 minutes.

[0042] The solution further includes the following steps for acquiring rain-measuring radar data:

[0043] Step 1: Establish a network observation system: Select a terrain area that includes mountains and hills, with a coverage area of ​​5,000-10,000 square kilometers, and deploy 3-5 dual-polarization rain measurement radars in the area to form a network observation system;

[0044] Step 2, Setting up data acquisition auxiliary equipment: The equipment includes a drone, a ground rain gauge array, a raindrop spectrometer, and a digital elevation model. The drone is equipped with a standardized metal sphere reflector. The ground rain gauge array is set at a spacing of 5km. The raindrop spectrometer is deployed in the center and edge areas of the network. The digital elevation model has a resolution of 30m.

[0045] Step 3, Data Acquisition: Synchronously acquire raw radar data, including: reflectivity factor Z, radial velocity V, spectral width W, and differential reflectivity Z. DR Differential phase Φ DP Correlation coefficient ρ HV External calibration data and ground observation data; acquisition time resolution of 5 minutes.

[0046] The beneficial effects of this invention are:

[0047] (1) Achieve full-chain automated quality control: Construct a full-chain quality control system covering raw data, basic data, and networking products, integrate multiple core algorithms, automatically identify and suppress various problems such as ground clutter, radio frequency interference, and system deviation, completely eliminate the dependence on manual operation, improve processing efficiency, and achieve 100% repeatability;

[0048] (2) Excellent ability to suppress complex terrain and interference: By combining digital elevation model, sparse reconstruction and local oversampling technology, the intensity of terrain occlusion is accurately quantified, and the data in blind area is efficiently recovered. The clutter identification rate is ≥95%, and the interference suppression effect is 15%-20% higher than that of traditional methods, which significantly improves the data quality in complex environments.

[0049] (3) Network collaboration optimization to improve product quality: Establish a two-way feedback mechanism between single stations and networks, and improve the spatial integrity and consistency (CI≥0.9) of network mosaic products through multi-station complementarity and weight fusion, meet the needs of high spatiotemporal resolution precipitation monitoring, and provide reliable data support for early warning of disasters such as rainstorms and flash floods;

[0050] (4) Form a closed-loop improvement mechanism: output structured quality diagnosis reports, accurately locate the source of anomalies, provide quantitative basis for radar calibration, equipment maintenance and site optimization, realize continuous optimization of data quality and equipment performance, and promote the engineering application and innovation of rain measurement radar technology;

[0051] (5) Strong adaptability and wide application scenarios: It is compatible with multiple types of rain measurement radar systems such as dual polarization, Doppler, and phased array, and is suitable for multiple business scenarios such as meteorology, hydrology, and emergency monitoring. It has broad engineering application value and promotion prospects.

[0052] The invention will be further explained in detail below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0053] Figure 1 is a schematic diagram showing the comparison results of reflectivity factor (dBZ) quality control before and after;

[0054] Figure 2 is a schematic diagram showing the comparison results before and after radial velocity (V) quality control;

[0055] Figure 3 shows the correlation coefficient (ρ). HV A diagram showing the comparison of quality control results before and after. Detailed Implementation

[0056] This embodiment constructs a full-chain quality control system covering raw data, basic data, and networking products, achieving the organic integration of independent quality control at a single station and collaborative quality control across multiple stations, thus enabling automated, intelligent, and standardized processing of rain measurement radar data.

[0057] A method for controlling the data quality of a networked collaborative rain-measuring radar includes data acquisition from rain-measuring radars in a networked observation system, wherein the networked observation system consists of dual-polarization rain-measuring radars at multiple observation stations, and the control method includes single-station quality control implementation steps and networked collaborative quality control implementation steps.

[0058] The single-station quality control implementation steps include:

[0059] Step 1: Raw data preprocessing: The signal-to-noise ratio (SNR) estimation algorithm is used to screen the effective data collected and remove low-quality data with an SNR of <3dB.

[0060] Step 2, External Calibration: Based on the UAV calibration data, correct the radar system deviation to ensure that the reflectivity factor measurement error is ≤1dBZ;

[0061] Step 3, Clutter Suppression: Apply clutter identification algorithms and Gaussian model adaptive filtering to filter clutter from the acquired data, achieving a clutter identification rate of ≥95%.

[0062] Step 4: Attenuation correction: For X-band radar data, an attenuation correction model is used to compensate for rainfall attenuation. The correlation between the corrected data and rain gauge observations is ≥0.9.

[0063] Step 5: Phase denoising and velocity ambiguity correction: Phase data is processed using a moving average filter and combined with a deambiguity algorithm to recover the radial velocity, with a velocity error ≤1m / s;

[0064] The above steps constitute a sub-method for single-station radar data quality control, the purpose of which is:

[0065] 1. Raw data quality assessment and preprocessing

[0066] 1.1 External calibration test: Relying on the field test platform, using a drone equipped with a standardized reflector, external calibration tests were carried out in various scenarios such as clear sky, light rain, heavy rain, and mixed weather. At the same time, observation data from ground rain gauges and raindrop spectrometers were collected to establish a high-quality benchmark dataset.

[0067] 1.2 Quality Assessment System: Combining sparse reconstruction technology with modern signal processing methods, key indicators such as signal-to-noise ratio, power spectrum characteristics, and polarization parameters of the original echo data are extracted to construct a multi-dimensional quality assessment model to identify problem types such as ground clutter, radio frequency interference, and system errors;

[0068] 1.3 Preprocessing: Based on the evaluation results, a noise suppression algorithm (signal-to-noise ratio threshold screening) is used to initially filter electromagnetic noise, laying the foundation for subsequent refined processing.

[0069] 2. Refined quality control of basic data

[0070] 2.1 Ground clutter suppression: A clutter identification algorithm based on statistical characteristics is combined with Gaussian model adaptive filtering to accurately identify and filter out ground clutter;

[0071] 2.2 Attenuation Correction: To address the attenuation characteristics of radars in different bands (such as the X-band), an attenuation correction model based on differential phase is adopted to achieve accurate compensation of echo intensity.

[0072] 2.3 Phase Denoising: A moving average filtering algorithm is used to reduce phase noise and improve the stability of speed and spectral width data;

[0073] 2.4 Velocity fuzziness correction: By combining the pulse repetition frequency and Doppler velocity characteristics, a velocity fuzziness recognition and defuzzification model is established to recover the true radial velocity.

[0074] 3. Handling obstructions in complex terrain

[0075] 3.1. Obstruction Quantification: Combining the digital elevation model (DEM) and the beam propagation geometry model, the beam obstruction angle and obstruction intensity are calculated, and the observation blind zone is automatically identified;

[0076] 3.2 Observation recovery: A local oversampling strategy is adopted to perform super-resolution reconstruction of the effective observation data around the blind zone. Combined with the spatial consistency constraint of the neighborhood, the blind zone data is accurately compensated.

[0077] The network collaborative quality control implementation steps include:

[0078] Step 1: Data spatiotemporal registration: Convert the data after quality control of each station to the UTM coordinate system, with a time synchronization error of ≤1 minute;

[0079] Step 2: Multi-station data consistency check: Calculate the data consistency index CI for overlapping areas of multiple stations, and perform secondary correction on data with CI < 0.8 based on valid neighboring data;

[0080] Step 3: Weight Allocation: Based on the single-station clutter identification rate, signal-to-noise ratio, and correction accuracy, the single-station quality level is divided into three levels: A, B, and C, with corresponding weights of 0.6, 0.3, and 0.1, respectively. A comprehensive weight is generated by combining the distance weight, where the closer the distance, the greater the weight.

[0081] Step 4, Blind Spot Complementation and Integration: For single-station blind spots, distance-weighted interpolation is performed using effective data from three adjacent stations; for overlapping areas, a comprehensive weighted average fusion is used to generate a network mosaic product.

[0082] Step 5, Quality Diagnosis: Output a network product quality report, mark abnormal areas, and provide a basis for site optimization. The marked abnormal areas include areas of continuous obstruction and high-frequency interference. The basis for site optimization includes suggesting increasing the antenna height in areas of continuous obstruction.

[0083] The above steps constitute a sub-method for network collaborative quality control, the purpose of which is:

[0084] 1. Multi-site data consistency verification:

[0085] 1.1 Data Spatiotemporal Registration: Convert the data after quality control of each station to a unified coordinate system and achieve spatiotemporal synchronization based on timestamps;

[0086] 1.2 Consistency Test: The data consistency index (CI) is used to assess the differences in data in overlapping areas of multiple stations;

[0087] 2. Multi-site collaborative integration:

[0088] 2.1 Site weight allocation: Based on the data quality level of a single site (comprehensively evaluated by indicators such as clutter identification rate, signal-to-noise ratio, and correction accuracy) and the spatial location of the site, a dynamic weight model is established. The higher the quality level and the closer to the grid center, the greater the weight.

[0089] 2.2 Blind Zone Complementarity: By utilizing the differences in observation angles from multiple stations, overlapping area data are fused through weighted averaging, and distance-weighted interpolation is used to supplement the blind zones of individual stations, thereby improving the spatial integrity of the network data.

[0090] 2.3 Network mosaic generation: Based on the collaborative fusion results, generate a high spatiotemporal resolution network mosaic product to ensure the continuity and consistency of the product.

[0091] 3. Quality diagnosis and closed-loop optimization:

[0092] 3.1 Structured quality output: Perform quality diagnosis, including anomaly types (clutter, interference, blockage, etc.), spatial distribution, temporal evolution patterns, and potential sources of system errors;

[0093] 3.2 Closed-loop optimization mechanism: Based on the diagnostic results, it provides quantitative basis for radar calibration (such as reflectivity factor calibration), equipment maintenance (such as antenna angle adjustment), and site optimization (such as raising the antenna or adjusting the site in complex terrain), so as to achieve continuous improvement in data quality and equipment performance.

[0094] In the above method: in the first step of the single-station quality control implementation process, the signal-to-noise ratio estimation algorithm calculation formula is:

[0095] ;

[0096] in: For signal power, The noise power is used to mark data as low quality when SNR < 3dB, and then it is either discarded or enhanced.

[0097] In the second step of the single-station quality control implementation process, the correction formula for the radar system deviation is:

[0098] ;

[0099] in: To observe differential reflectance; This is a systematic bias, determined through external calibration tests.

[0100] In the third step of the single-station quality control implementation process, the calculation formula for clutter identification is as follows:

[0101] ;

[0102] in: The standard deviation of the reflectivity factor. Mean reflectivity factor The horizontal-vertical polarization correlation coefficient is given when... When the signal is detected as clutter, T is the threshold value.

[0103] The calculation formula for the Gaussian model adaptive filtering is as follows:

[0104] ;

[0105] in: The original reflectivity factor, Radial velocity, Center of clutter velocity This represents the standard deviation of clutter velocity.

[0106] In the fourth step of the single-station quality control implementation process, the attenuation correction model is:

[0107] ;

[0108] in: To observe the reflectivity factor, The specific attenuation coefficient, The target distance.

[0109] In the single-station quality control implementation steps, the formula for the moving average filtering process in step five is:

[0110] ;

[0111] in: This is the raw phase data; This represents the half-width of the sliding window, with a value ranging from 3 to 5.

[0112] In the second step of the network collaborative quality control implementation process, the formula for calculating the consistency index (CI) is as follows:

[0113] ;

[0114] in: , The reflectance factor for adjacent stations on the same grid is defined as follows: when CI < 0.8, it is considered inconsistent data and a secondary correction is triggered.

[0115] In the fourth step of the network collaborative quality control implementation process, the spatial resolution of the network mosaic is 1km×1km, and the time resolution is 5 minutes.

[0116] The steps for acquiring rainfall radar data include:

[0117] Step 1: Establish a network observation system: Select a terrain area that includes mountains and hills, with a coverage area of ​​5,000-10,000 square kilometers, and deploy 3-5 dual-polarization rain measurement radars in the area to form a network observation system;

[0118] Step 2, Setting up data acquisition auxiliary equipment: The equipment includes a drone, a ground rain gauge array, a raindrop spectrometer, and a digital elevation model. The drone is equipped with a standardized metal sphere reflector. The ground rain gauge array is set at a spacing of 5km. The raindrop spectrometer is deployed in the center and edge areas of the network. The digital elevation model has a resolution of 30m.

[0119] Step 3, Data Acquisition: Synchronously acquire raw radar data, including: reflectivity factor Z, radial velocity V, spectral width W, and differential reflectivity Z. DR Differential phase Φ DP Correlation coefficient ρ HV External calibration data and ground observation data; acquisition time resolution of 5 minutes.

[0120] The above embodiments are based on "precise quality control at a single station, network-wide collaborative optimization as the core, and quality closed-loop improvement as the goal," constructing a technical framework of "three-layer quality control + collaborative integration + diagnostic feedback."

[0121] The first layer is the raw data quality control layer, which performs quality assessment and preprocessing of raw radar echo data and identifies system errors and initial interference.

[0122] The second layer is the basic data quality control layer, which performs fine processing on radar basic data, such as clutter suppression, attenuation correction, and phase denoising, to improve the reliability of single-station data.

[0123] The third layer: the network collaboration quality control layer, which generates high-quality network mosaic products through multi-station data consistency verification, blind spot complementarity, and weight fusion;

[0124] The method is constructed with a collaborative fusion module and a diagnostic feedback module:

[0125] Collaborative Integration Module: Establishes a two-way feedback mechanism between individual stations and the network, using individual station quality control results to support the accuracy of network integration, and optimizing individual station quality control parameters based on network consistency requirements;

[0126] Diagnostic feedback module: Outputs structured quality reports, identifies the types and sources of anomalies, and provides a basis for equipment maintenance and site optimization.

[0127] The above embodiments address the following issues: (1) In complex terrain (mountains, hills) and urban environments, there is a lack of automated and precise identification and correction methods for blind spots and false echoes caused by beam blocking, ground clutter, and multipath effects; (2) Existing algorithms struggle to effectively suppress and compensate for data distortion caused by non-meteorological factors such as radio frequency interference and electromagnetic noise; (3) There is a disconnect between single-station quality control and network quality control, resulting in poor consistency of multi-station data and spatial discontinuity and cumulative deviations in network mosaic products, which cannot meet the needs of high spatiotemporal resolution precipitation monitoring; (4) The quality control process lacks standardized specifications and relies on manual operation, resulting in low efficiency, poor repeatability, and unclear quality diagnosis, making it difficult to form a closed-loop optimization mechanism for data quality and equipment performance.

Claims

1. A method for controlling the data quality of a networked collaborative rainfall radar system, comprising data acquisition from rainfall radars in a networked observation system, wherein the networked observation system consists of dual-polarization rainfall radars at multiple observation stations, characterized in that... The control method includes single-station quality control implementation steps and network-based collaborative quality control implementation steps; The single-station quality control implementation steps include: Step 1, raw data preprocessing: using a signal-to-noise ratio estimation algorithm to screen valid data and remove low-quality data with SNR < 3dB; Step 2, external calibration: based on UAV calibration data, correcting radar system deviations to ensure reflectivity factor measurement error ≤ 1dBZ; Step 3, clutter suppression: applying a clutter identification algorithm and Gaussian model adaptive filtering to filter clutter in the acquired data, with a clutter identification rate ≥ 95%; Step 4, attenuation correction: for X-band radar data, using an attenuation correction model to compensate for rainfall attenuation, with the correlation between the corrected data and rain gauge observations ≥ 0.9; Step 5, phase denoising and velocity ambiguity correction: using a moving average filter to process phase data, combined with a deambiguity algorithm to recover radial velocity, with a velocity error ≤ 1m / s; The network collaborative quality control implementation steps include: Step 1, data spatiotemporal registration: converting the quality-controlled data from each single station to the UTM coordinate system. The time synchronization error is ≤1 minute; the second step is multi-station data consistency verification: calculate the data consistency index (CI) for overlapping areas of multiple stations, and perform secondary correction on data with CI < 0.8 based on effective neighboring data; the third step is weight allocation: divide the quality level of a single station into three levels, A, B, and C, based on the clutter identification rate, signal-to-noise ratio, and correction accuracy, with corresponding weights of 0.6, 0.3, and 0.1, respectively, and generate a comprehensive weight based on distance weight, where the closer the distance, the greater the weight; the fourth step is blind zone complementarity and fusion: for single-station blind zones, use effective data from three adjacent stations for distance-weighted interpolation; for overlapping areas, use comprehensive weighted average fusion to generate a network mosaic product; the fifth step is quality diagnosis: output a network product quality report, mark abnormal areas, and provide a basis for site optimization. The marked abnormal areas include persistent obstruction areas and high-frequency interference areas; the basis for site optimization includes suggesting increasing the antenna height in persistent obstruction areas.

2. The control method according to claim 1, characterized in that, In the first step of the single-station quality control implementation process, the signal-to-noise ratio estimation algorithm is calculated using the following formula: ;in: For signal power, For noise power, when Data marked as low quality is removed or enhanced.

3. The control method according to claim 1, characterized in that, In the second step of the single-station quality control implementation process, the correction formula for the radar system deviation is: ;in: To observe differential reflectance; This is a systematic bias, determined through external calibration tests.

4. The control method according to claim 1, characterized in that, In the third step of the single-station quality control implementation process, the calculation formula for clutter identification is as follows: ;in: The standard deviation of the reflectivity factor. Mean reflectivity factor The horizontal-vertical polarization correlation coefficient is used. When Clutter Index > T, it is considered clutter, where T is the threshold. The calculation formula for the Gaussian model adaptive filtering is as follows: ;in: The original reflectivity factor, Radial velocity, Center of clutter velocity This represents the standard deviation of clutter velocity.

5. The control method according to claim 1, characterized in that, In the fourth step of the single-station quality control implementation process, the attenuation correction model is: ;in: To observe the reflectivity factor, The specific attenuation coefficient, The target distance.

6. The control method according to claim 1, characterized in that, In the single-station quality control implementation steps, the formula for the moving average filtering process in step five is: ;in: This is the raw phase data; This represents the half-width of the sliding window, with a value ranging from 3 to 5.

7. The control method according to claim 1, characterized in that, In the second step of the network collaborative quality control implementation process, the formula for calculating the consistency index (CI) is as follows: ;in: 、 The reflectance factor for adjacent stations on the same grid is defined as follows: when CI < 0.8, it is considered inconsistent data and a secondary correction is triggered.

8. The control method according to claim 1, characterized in that, In the fourth step of the network collaborative quality control implementation process, the spatial resolution of the network mosaic is 1km×1km, and the time resolution is 5 minutes.

9. The control method according to claim 1, characterized in that, The steps for acquiring rain-measuring radar data include: Step 1: Establishing a network observation system: Selecting a terrain area including mountains and hills, with a coverage area of ​​5000-10000 square kilometers, and deploying 3-5 dual-polarization rain-measuring radars within the area to form a network observation system; Step 2: Setting up data acquisition auxiliary equipment: The equipment includes a drone, a ground rain gauge array, a raindrop spectrometer, and a digital elevation model. Specifically: the drone carries a standardized metal spherical reflector; the ground rain gauge array is spaced 5 km apart; the raindrop spectrometer is deployed in the center and edge areas of the network; and the digital elevation model has a resolution of 30 m; Step 3: Data acquisition: Synchronously acquiring raw radar data, including: reflectivity factor Z, radial velocity V, spectral width W, and differential reflectivity Z. DR Differential phase Φ DP Correlation coefficient ρ HV External calibration data and ground observation data; acquisition time resolution of 5 minutes.

Images