A GNSS and rain radar cooperative networking system for mountain flood warning

Abstract

The application discloses a GNSS and rain measurement radar cooperative networking system for mountain flood early warning, and is used in the technical field of mountain flood early warning.The system comprises a monitoring station network layout and data acquisition module, a radar blind area is identified based on terrain and radar coverage analysis, and a GNSS monitoring station network is laid out, and GNSS atmospheric water vapor, rain measurement radar reflectivity and hydrological station data are acquired; a surface rainfall forecast module cooperatively processes GNSS atmospheric water vapor data, rain measurement radar reflectivity and hydrological station data by using a water vapor and precipitation cooperative fusion algorithm, and generates a surface rainfall forecast product within a preset future time; a flood risk simulation and graded early warning module inputs the surface rainfall forecast product into a watershed hydrological model, deduces a flood risk in combination with real-time hydrological data, and generates graded early warning information.The application generates high-precision surface rainfall forecast, deduces a flood risk in real time in combination with a watershed hydrological model, and realizes accurate graded early warning.

Description

Technical Field

[0001] This application relates to the field of flash flood early warning technology, and more specifically, to a GNSS and rain-measuring radar collaborative networking system for flash flood early warning. Background Technology

[0002] Flash floods refer to rapid floods in mountainous and hilly areas caused by torrential rain or continuous rainfall within a short period of time. These floods are usually sudden, intense, and localized, and can quickly inundate low-lying areas, causing serious disasters. Especially in mountainous or sloping areas, the water flow is concentrated and violent, posing a great threat to people's lives and property.

[0003] The importance of flash flood warnings lies in their ability to provide early risk alerts to disaster-stricken areas, helping governments and relevant departments to take flood prevention measures, such as evacuating people and strengthening dikes, thereby reducing losses caused by flash floods. Timely and effective warnings can significantly improve the public's emergency response capabilities, avoid casualties and reduce economic losses, especially when flash floods occur, speed of response is crucial.

[0004] Traditional radar monitoring often has blind spots in complex terrain, and these blind spots are mostly located in valleys. Valleys are both water vapor transport channels and the core areas for flash floods, making it impossible to directly monitor the most critical precipitation data. Secondly, traditional early warning methods are difficult to capture the precursory information before precipitation forms, and may not fully consider hydrological data and real-time changes, resulting in inaccurate or delayed early warning information.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] To overcome the above problems, this application aims to propose a GNSS and rainfall radar collaborative networking system for flash flood early warning. The purpose is to solve the problems that traditional radar monitoring often has blind spots in complex terrain and cannot fully cover the watershed area, resulting in the inability to obtain key data in a timely manner in the blind spots of flash flood monitoring. Furthermore, existing flood warnings may not fully consider hydrological data and real-time changes when assessing flood risk, resulting in inaccurate or delayed warning information.

[0007] Therefore, the specific technical solution adopted in this application is as follows: A GNSS and rainfall radar collaborative networking system for flash flood early warning, the system includes: The monitoring station network deployment and data acquisition module identifies radar blind spots based on terrain and radar coverage analysis and deploys a GNSS monitoring station network to collect GNSS atmospheric water vapor, rainfall radar reflectivity and hydrological station data. The areal rainfall forecast module uses a water vapor and precipitation fusion algorithm to collaboratively process GNSS atmospheric water vapor data, rain measurement radar reflectivity, and hydrological station data to generate areal rainfall forecast products for a preset time period in the future. The flood risk simulation and graded early warning module inputs areal rainfall forecast products into the watershed hydrological model, combines real-time hydrological data to deduce flood risks and generate graded early warning information.

[0008] Alternatively, the method for identifying radar blind spots and deploying a GNSS monitoring network based on terrain and radar coverage analysis is as follows: Collect topographic data, radar parameters, and information on potential flash flood sites in the target watershed; Based on terrain data and radar parameters, a beam propagation algorithm is used to simulate the effective radar coverage area and generate a radar coverage analysis map. Based on the radar coverage analysis map, spatial overlay analysis is used to identify radar detection blind spots, and a list of key blind spots is obtained by combining the distribution of information on flash flood hazard points; Based on the list of key blind spots, objectives were set to improve water vapor observation and satellite signal quality, and the specific locations for GNSS monitoring stations were determined.

[0009] Optionally, the method for generating the radar coverage analysis map is as follows: Acquire digital elevation data of the target area, as well as the coordinates, frequency, and transmission power of each radar unit; Using the coordinates and frequencies of each radar as input, and based on the geometric propagation principle of radar beams, the line-of-sight path between each radar is calculated grid by grid in the digital elevation data, and the presence of terrain obstruction is determined based on the line-of-sight path. Based on the determination of terrain obstruction, each grid point is comprehensively identified. If a grid point is directly visible to the radar, it is identified as having effective coverage. If a grid point is not directly visible to the radar, it is marked as terrain occlusion, and a coverage status dataset is generated. The coverage status dataset is overlaid with the geographic base map and rendered to generate a spatial distribution map showing the effective radar coverage area and terrain-obstructed blind spots.

[0010] Optionally, a list of key blind spots can be obtained by combining information on the distribution of flash flood hazard points, including: Obtain the distribution of potential flash flood sites from the information on potential flash flood sites; The spatial distribution map was then overlaid with the distribution of flash flood hazard points for spatial analysis, and overlapping areas were identified. The coverage priority of overlapping areas is assessed based on the density and level of potential hazards in the overlapping areas. Based on the coverage priority, areas with higher than the preset priority are selected as a list of key blind spots for station deployment.

[0011] Optionally, the method for collaboratively processing GNSS atmospheric water vapor data with rainfall radar reflectivity and hydrological station data using a water vapor and precipitation fusion algorithm is as follows: Spatial interpolation was performed on GNSS atmospheric water vapor data, grid resampling was performed on rain measurement radar reflectivity data, and quality control was implemented in conjunction with hydrological station data to generate a multi-source grid dataset. Using a multi-source grid dataset as input, a co-kriging algorithm is employed for spatial fusion to generate a joint analysis field of water vapor and radar. The time series data of the water vapor and radar joint analysis field are input into a pre-trained spatiotemporal deep learning model, which outputs gridded products of areal rainfall forecast within a preset time period. The forecast grid products are corrected based on real-time hydrological station observation data.

[0012] Alternatively, the method for generating a multi-source grid dataset is as follows: Time synchronization preprocessing was performed on the raw data of GNSS atmospheric water vapor, rain measurement radar reflectivity and hydrological stations to obtain a time-consistent multi-source data sequence. Define a unified regular grid coordinate system based on the geographical scope and spatial resolution requirements of the target area; Interpolate GNSS water vapor data and hydrological station precipitation data from multi-source data sequences to a unified grid, and resample rain-measuring radar reflectivity data to the same grid to obtain water vapor, precipitation, and reflectivity grid data layers; By integrating water vapor, precipitation, and reflectivity grid data layers and performing joint quality control, a multi-source grid dataset is generated.

[0013] Optionally, the calculation formula for the joint analysis field of water vapor and radar is as follows:

[0014] In the formula, Represents target grid points At any moment t The fused atmospheric water vapor estimate; Indicates the first s One observation sample point; This represents the total number of observation sample points for three types of observation data: GNSS water vapor, rainfall radar, and hydrological stations. Indicates the first s The overall weight of each observation sample point; Indicates the first s Multi-source observation variables for each observation sample point; Indicates the current time t Reference time The temporal correlation between them; Indicates at time t Below, grid points to be estimated With the s Observation sample points Spatial covariance of water vapor between them; Indicates at time t Next, the s Observation sample points With the grid points to be estimated The spatial covariance of water vapor between them.

[0015] Optionally, the areal rainfall forecast grid products for a preset future time period are output, including: The time series data of the water vapor and radar joint analysis field are preprocessed to obtain the characteristic sequence; The feature sequence is input into a pre-trained spatiotemporal deep learning model for forward propagation calculation, and standardized prediction features are output. The standardized forecast features are denormalized and reconstructed using spatial grids to generate gridded products for areal rainfall forecasts within a preset time period.

[0016] Optionally, the method for generating tiered early warning information is as follows: Preprocess the areal rainfall forecast products and real-time hydrological data to generate a dataset of precipitation and runoff conditions. The precipitation and runoff generation data set is input into a pre-trained watershed hydrological model to simulate runoff generation, confluence, and flood evolution, and output the hydrological process lines of each control section. Based on the simulated hydrological process line, combined with the watershed digital elevation model and the distribution of flood protection objects, an assessment of the flood inundation range and the population affected was carried out. Based on preset warning indicators and risk level thresholds, tiered warning information covering different regions and different degrees of severity is generated.

[0017] Compared with the prior art, this application has the following beneficial effects: 1. This application comprehensively utilizes topographic, radar, GNSS water vapor, and hydrological station data, and employs precise beam propagation and spatial analysis algorithms to effectively identify radar blind spots and optimize the deployment of monitoring stations, thereby improving the accuracy of water vapor and precipitation observations. Through collaborative fusion algorithms and deep learning models, it generates high-precision areal rainfall forecasts and combines them with watershed hydrological models to simulate flood risks in real time, achieving accurate graded early warning.

[0018] 2. This application accurately identifies radar blind spots and optimizes the deployment of GNSS monitoring stations by combining terrain data, radar parameters, and information on flash flood hazard points; it ensures the effectiveness of radar coverage by using beam propagation algorithms and spatial overlay analysis, and prioritizes the deployment of monitoring stations according to the distribution of flash flood hazard points, thereby improving the monitoring accuracy of blind spots.

[0019] 3. This application combines multi-source data fusion with a spatiotemporal deep learning model to efficiently integrate GNSS atmospheric water vapor, rain-measuring radar reflectivity, and hydrological station data to generate accurate areal rainfall forecasts. This collaborative processing method significantly improves the spatiotemporal accuracy of precipitation forecasts, especially in flash flood warnings, where forecast results can be updated and corrected in real time.

[0020] 4. This application uses a watershed hydrological model and flood evolution simulation. The system not only provides high-precision flood risk assessment, but also generates graded early warning information according to different risk levels, thereby improving the accuracy and response speed of early warning and ensuring timely and effective flood prevention and mitigation. Attached Figure Description

[0021] The above-mentioned features, characteristics, and advantages of this application, as well as their implementation methods, will become clearer and more understandable in conjunction with the following description of the embodiments, which are illustrated in detail with reference to the accompanying drawings. Schematic diagrams are shown here: Figure 1 This is a schematic diagram of the GNSS and rainfall radar collaborative networking system for flash flood early warning in this application. Detailed Implementation

[0022] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0023] According to embodiments of this application, a GNSS and precipitation radar collaborative networking system for flash flood early warning is provided. By comprehensively utilizing terrain, radar, GNSS water vapor, and hydrological station data, and employing precise beam propagation and spatial analysis algorithms, it can effectively identify radar blind spots and optimize monitoring station deployment, thereby improving the accuracy of water vapor and precipitation observations. Figure 1 As shown, the system includes: The monitoring station network deployment and data acquisition module identifies radar blind spots based on terrain and radar coverage analysis and deploys a GNSS monitoring station network to collect GNSS atmospheric water vapor, rainfall radar reflectivity, and hydrological station data.

[0024] Preferably, the method for identifying radar blind spots and deploying a GNSS monitoring network based on terrain and radar coverage analysis is as follows: Collect topographic data, radar parameters, and information on potential flash flood sites in the target watershed; Based on terrain data and radar parameters, a beam propagation algorithm is used to simulate the effective radar coverage area and generate a radar coverage analysis map. Based on the radar coverage analysis map, spatial overlay analysis is used to identify radar detection blind spots, and a list of key blind spots is obtained by combining the distribution of information on flash flood hazard points; Based on the list of key blind spots, objectives were set to improve water vapor observation and satellite signal quality, and the specific locations for GNSS monitoring stations were determined.

[0025] Preferably, the method for generating the radar coverage analysis map is as follows: Acquire digital elevation data of the target area, as well as the coordinates, frequency, and transmission power of each radar unit; Using the coordinates and frequencies of each radar as input, and based on the geometric propagation principle of radar beams, the line-of-sight path between each radar is calculated grid by grid in the digital elevation data, and the presence of terrain obstruction is determined based on the line-of-sight path. Based on the determination of terrain obstruction, each grid point is comprehensively identified. If a grid point is directly visible to the radar, it is identified as having effective coverage. If a grid point is not directly visible to the radar, it is marked as terrain occlusion, and a coverage status dataset is generated. The coverage status dataset is overlaid with the geographic base map and rendered to generate a spatial distribution map showing the effective radar coverage area and terrain-obstructed blind spots.

[0026] Preferably, a list of key blind spots is obtained by combining information on the distribution of flash flood hazard points, including: Obtain the distribution of potential flash flood sites from the information on potential flash flood sites; The spatial distribution map was then overlaid with the distribution of flash flood hazard points for spatial analysis, and overlapping areas were identified. The coverage priority of overlapping areas is assessed based on the density and level of potential hazards in the overlapping areas. Based on the coverage priority, areas with higher than the preset priority are selected as a list of key blind spots for station deployment.

[0027] It should be explained that, at the data acquisition level, terrain data usually comes from high-precision digital elevation models, which can be obtained through UAV lidar mapping, airborne SAR mapping data, or 1–5m resolution DEM products provided by existing surveying and mapping departments; radar parameters are provided by existing rain-measuring radars, including antenna height, transmit power, beamwidth, elevation angle settings, etc. of C-band or S-band weather radars; information on flash flood hazard points comes from the historical disaster database of the water resources department, remote sensing interpretation results, or field survey results; Typically, a GIS spatial analysis platform is deployed in conjunction with a radar beam propagation model for calculation. The core implementation method is to construct a three-dimensional terrain field based on a DEM, and run line-of-sight analysis or terrain occlusion analysis on a computing server or high-performance workstation. Starting from the radar station, rays are constructed for each target grid point, and it is gradually determined whether there are terrain units higher than the radar beam height on the ray path to determine whether it is occluded. The calculation results are output as coverage status raster data, and then a coverage analysis map is generated through GIS rendering. Spatial overlay analysis is performed using a spatial database. The radar blind zone grid is overlaid with vector data of flash flood hazard points to calculate the number, density and risk level of hazard points in the blind zone and form a risk index evaluation result. It adopts dual-frequency or multi-frequency GNSS receivers, equipped with high-stability choke coil antennas and anti-multipath observation environment design. During site selection, GNSS site selection evaluation equipment is used to conduct sky visibility tests and electromagnetic environment detection to ensure unobstructed views, high signal-to-noise ratio and good data continuity. The site data is transmitted back to the data center in real time via 4G / 5G or Beidou short message, and is aggregated with radar data and hydrological station data on a unified server platform.

[0028] The areal rainfall forecast module uses a water vapor and precipitation fusion algorithm to collaboratively process GNSS atmospheric water vapor data, rainfall radar reflectivity, and hydrological station data to generate areal rainfall forecast products for a preset time period in the future.

[0029] Preferably, the method for collaboratively processing GNSS atmospheric water vapor data, rainfall radar reflectivity, and hydrological station data using a water vapor and precipitation fusion algorithm is as follows: Spatial interpolation was performed on GNSS atmospheric water vapor data, grid resampling was performed on rain measurement radar reflectivity data, and quality control was implemented in conjunction with hydrological station data to generate a multi-source grid dataset. Using a multi-source grid dataset as input, a co-kriging algorithm is employed for spatial fusion to generate a joint analysis field of water vapor and radar. The time series data of the water vapor and radar joint analysis field are input into a pre-trained spatiotemporal deep learning model, which outputs gridded products of areal rainfall forecast within a preset time period. The forecast grid products are corrected based on real-time hydrological station observation data.

[0030] Preferably, the method for generating a multi-source grid dataset is as follows: Time synchronization preprocessing was performed on the raw data of GNSS atmospheric water vapor, rain measurement radar reflectivity and hydrological stations to obtain a time-consistent multi-source data sequence. Define a unified regular grid coordinate system based on the geographical scope and spatial resolution requirements of the target area; Interpolate GNSS water vapor data and hydrological station precipitation data from multi-source data sequences to a unified grid, and resample rain-measuring radar reflectivity data to the same grid to obtain water vapor, precipitation, and reflectivity grid data layers; By integrating water vapor, precipitation, and reflectivity grid data layers and performing joint quality control, a multi-source grid dataset is generated.

[0031] Preferably, the calculation formula for the joint analysis field of water vapor and radar is as follows:

[0032] In the formula, Represents target grid points At any moment t The fused atmospheric water vapor estimate; Indicates the first s One observation sample point; This represents the total number of observation sample points for three types of observation data: GNSS water vapor, rainfall radar, and hydrological stations. Indicates the first s The overall weight of each observation sample point; Indicates the first s Multi-source observation variables for each observation sample point; Indicates the current time t Reference time The temporal correlation between them; Indicates at time t Below, grid points to be estimated With the s Observation sample points Spatial covariance of water vapor between them; Indicates at time t Next, the s Observation sample points With the grid points to be estimated The spatial covariance of water vapor between them.

[0033] Preferably, the output of areal rainfall forecast grid products within a preset future time period includes: The time series data of the water vapor and radar joint analysis field are preprocessed to obtain the characteristic sequence; The feature sequence is input into a pre-trained spatiotemporal deep learning model for forward propagation calculation, and standardized prediction features are output. The standardized forecast features are denormalized and reconstructed using spatial grids to generate gridded products for areal rainfall forecasts within a preset time period.

[0034] It needs to be explained that, taking a certain flash flood-prone small watershed (area 100 km²) as an example... 2 Central grid point Taking (e.g., 119.0°E, 30.0°N) as an example, the specific parameters are as follows: Observation sites: 5 GNSS water vapor stations (G1-G5), 1 C-band rainfall radar (R1), and 3 hydrological stations (H1-H3). Time setting: Target time t =14:00, reference time t 0 = 13:00 (1 hour interval); Topographic data: 5m resolution DEM The elevation at point G5 is 300m, at point R1 it is 350m, and at point R1 it is 200m. The radar parameters include an antenna height of 200m, a beam elevation angle of 0.5°, and a transmit power of 500kW. In line of sight analysis The distance to R1 is 10km. The radar beam height = 200 + 10 × tan(0.5°) ≈ 287.2m. There is no terrain higher than this height in the path. For effective coverage; The distance from G5 to R1 is 5km. The radar beam height is approximately 243.6m (200 + 5 × tan(0.5°)). There is a mountain peak with an altitude of 320m in the path, which is higher than the beam height. Therefore, G5 is a radar blind spot. There are two high-risk flash flood hazard points at G5, with a hazard density of 0.5 points / km². 2 The level is high, so G5 is set as a critical blind zone and one GNSS station (G5) is deployed to supplement water vapor observations; The input data is shown in Table 1. Table 1 Multi-source data fusion

[0035] Based on the parameters in Table 1, substitute the target grid points. At any moment t Combined atmospheric water vapor estimates The estimated value of fused atmospheric water vapor was calculated. It is 3.77mm; Will The fused values ​​from the previous hour (13:00) form a time-series feature sequence; Input a pre-trained spatiotemporal deep learning model and output standardized prediction features; After inverse normalization, the isal rainfall forecast value at 15:00 was 12 mm. Then, based on real-time hydrological observation (10 mm of precipitation at 14:00), the final forecast value was 11 mm. By supplementing radar blind spots with GNSS measurements, the observation gaps caused by terrain obstruction were resolved. Furthermore, the accuracy of water vapor estimation was improved through the spatiotemporal fusion of three sources, ultimately providing reliable areal rainfall forecasts for flash flood warnings.

[0036] The flood risk simulation and graded early warning module inputs areal rainfall forecast products into the watershed hydrological model, combines real-time hydrological data to deduce flood risks and generate graded early warning information.

[0037] Preferably, the method for generating tiered early warning information is as follows: Preprocess the areal rainfall forecast products and real-time hydrological data to generate a dataset of precipitation and runoff conditions. The precipitation and runoff generation data set is input into a pre-trained watershed hydrological model to simulate runoff generation, confluence, and flood evolution, and output the hydrological process lines of each control section. Based on the simulated hydrological process line, combined with the watershed digital elevation model and the distribution of flood protection objects, an assessment of the flood inundation range and the population affected was carried out. Based on preset warning indicators and risk level thresholds, tiered warning information covering different regions and varying degrees of severity is generated. It should be noted that the areal rainfall forecast is 11 mm from 14:00 to 15:00 (after correction based on real-time hydrological data). Real-time hydrological data: 14:00 measured rainfall 10mm, current water level 5.0m, flow rate 20m³ / h 3 / s, soil moisture content is 0.3 (slightly dry); Warning thresholds: Blue (<6.5m), Yellow (6.5–8.0m), Orange (8.0–10.0m), Red (≥10.0m); The generated precipitation and runoff condition dataset is input into a pre-trained watershed hydrological model to conduct runoff generation and flood evolution simulations. Based on conditions such as total precipitation and soil moisture content, the process of precipitation being converted into runoff is calculated, and the propagation and evolution of floods within the watershed are simulated. Finally, the hydrological process lines of each control section are output: the water level is expected to rise to 7.2m at 15:00, reach a peak of 9.1m at 16:00, and fall back to 6.8m at 17:00, clearly showing the trend of water level changes at different times in the future. Based on the simulated hydrological process line, combined with the watershed digital elevation model and the spatial distribution of flood protection objects, an assessment of the flood inundation range and the affected population was conducted. By analyzing the relationship between the simulated peak water level of 9.1m and the topographic elevation, it was determined that downstream village B (elevation 8.5–9.0m) would be completely flooded, and the low-lying areas of village C (elevation 9.0–9.5m) would also be flooded. At the same time, it was assessed that approximately 500 people would be affected, and one rural road would be blocked by the flood. Based on preset warning indicators and risk level thresholds (blue <6.5m, yellow 6.5–8.0m, orange 8.0–10.0m, red ≥10.0m), and referring to the simulated hydrological process line, graded warning information covering different areas and different degrees of severity is generated. At 15:00, the water level reached 7.2m, triggering a yellow alert, indicating that low-lying areas in Village C would be affected. At 16:00, the water level reached 9.1m, triggering an orange alert. It was determined that the entire village of B and low-lying areas of village C would be threatened by flooding, requiring the immediate evacuation of residents in the relevant areas and the suspension of traffic on rural roads.

[0038] It should be noted that the calculation formulas and all parameters involved in the calculations in this application have been dimensionless beforehand. The process of dimensionless processing is well known in the industry and will not be described here.

[0039] Although the present application has disclosed the preferred embodiments above, the embodiments are merely examples for the purpose of illustration and are not intended to limit the present application. Those skilled in the art can make some modifications and refinements without departing from the spirit and scope of the present application. The scope of protection claimed by the present application should be determined by the claims.

Claims

1. A GNSS and rainfall radar collaborative networking system for flash flood early warning, characterized in that, The system includes: The monitoring station network deployment and data acquisition module identifies radar blind spots based on terrain and radar coverage analysis and deploys a GNSS monitoring station network to collect GNSS atmospheric water vapor, rainfall radar reflectivity and hydrological station data. The areal rainfall forecast module uses a water vapor and precipitation fusion algorithm to collaboratively process GNSS atmospheric water vapor data, rain measurement radar reflectivity, and hydrological station data to generate areal rainfall forecast products for a preset time period in the future. The flood risk simulation and graded early warning module inputs areal rainfall forecast products into the watershed hydrological model, combines real-time hydrological data to deduce flood risks and generate graded early warning information.

2. The GNSS and rainfall radar collaborative networking system for flash flood early warning as described in claim 1, characterized in that, The method for identifying radar blind spots and deploying a GNSS monitoring network based on terrain and radar coverage analysis is as follows: Collect topographic data, radar parameters, and information on potential flash flood sites in the target watershed; Based on terrain data and radar parameters, a beam propagation algorithm is used to simulate the effective radar coverage area and generate a radar coverage analysis map. Based on the radar coverage analysis map, spatial overlay analysis is used to identify radar detection blind spots, and a list of key blind spots is obtained by combining the distribution of information on flash flood hazard points; Based on the list of key blind spots, objectives were set to improve water vapor observation and satellite signal quality, and the specific locations for GNSS monitoring stations were determined.

3. The GNSS and rainfall radar collaborative networking system for flash flood early warning according to claim 2, characterized in that, The method for generating the radar coverage analysis map is as follows: Acquire digital elevation data of the target area, as well as the coordinates, frequency, and transmission power of each radar unit; Using the coordinates and frequencies of each radar as input, and based on the geometric propagation principle of radar beams, the line-of-sight path between each radar is calculated grid by grid in the digital elevation data, and the presence of terrain obstruction is determined based on the line-of-sight path. Based on the determination of terrain obstruction, each grid point is comprehensively identified. If a grid point is directly visible to the radar, it is identified as having effective coverage. If a grid point is not directly visible to the radar, it is marked as terrain occlusion, and a coverage status dataset is generated. The coverage status dataset is overlaid with the geographic base map and rendered to generate a spatial distribution map showing the effective radar coverage area and terrain-obstructed blind spots.

4. The GNSS and rainfall radar collaborative networking system for flash flood early warning according to claim 3, characterized in that, The key blind spot list obtained by combining the distribution of information on potential flash flood hazards includes: Obtain the distribution of potential flash flood sites from the information on potential flash flood sites; The spatial distribution map was then overlaid with the distribution of flash flood hazard points for spatial analysis, and overlapping areas were identified. The coverage priority of overlapping areas is assessed based on the density and level of potential hazards in the overlapping areas. Based on the coverage priority, areas with higher than the preset priority are selected as a list of key blind spots for station deployment.

5. The GNSS and rainfall radar collaborative networking system for flash flood early warning according to claim 1, characterized in that, The method for collaboratively processing GNSS atmospheric water vapor data, rainfall radar reflectivity, and hydrological station data using a water vapor and precipitation fusion algorithm is as follows: Spatial interpolation was performed on GNSS atmospheric water vapor data, grid resampling was performed on rain measurement radar reflectivity data, and quality control was implemented in conjunction with hydrological station data to generate a multi-source grid dataset. Using a multi-source grid dataset as input, a co-kriging algorithm is employed for spatial fusion to generate a joint analysis field of water vapor and radar. The time series data of the water vapor and radar joint analysis field are input into a pre-trained spatiotemporal deep learning model, which outputs gridded products of areal rainfall forecast within a preset time period. The forecast grid products are corrected based on real-time hydrological station observation data.

6. The GNSS and rainfall radar collaborative networking system for flash flood early warning according to claim 5, characterized in that, The method for generating multi-source grid datasets is as follows: Time synchronization preprocessing was performed on the raw data of GNSS atmospheric water vapor, rain measurement radar reflectivity and hydrological stations to obtain a time-consistent multi-source data sequence. Define a unified regular grid coordinate system based on the geographical scope and spatial resolution requirements of the target area; Interpolate GNSS water vapor data and hydrological station precipitation data from multi-source data sequences to a unified grid, and resample rain-measuring radar reflectivity data to the same grid to obtain water vapor, precipitation, and reflectivity grid data layers; By integrating water vapor, precipitation, and reflectivity grid data layers and performing joint quality control, a multi-source grid dataset is generated.

7. The GNSS and rainfall radar collaborative networking system for flash flood early warning according to claim 6, characterized in that, The calculation formula for the combined water vapor and radar analysis field is as follows: In the formula, Represents target grid points At any moment t The fused atmospheric water vapor estimate; Indicates the first s One observation sample point; This represents the total number of observation sample points for three types of observation data: GNSS water vapor, rainfall radar, and hydrological stations. Indicates the first s The overall weight of each observation sample point; Indicates the first s Multi-source observation variables for each observation sample point; Indicates the current time t Reference time The temporal correlation between them; Indicates at time t Below, grid points to be estimated With the s Observation sample points Spatial covariance of water vapor between them; Indicates at time t Next, the s Observation sample points With the grid points to be estimated The spatial covariance of water vapor between them.

8. The GNSS and rainfall radar collaborative networking system for flash flood early warning according to claim 7, characterized in that, The output of the areal rainfall forecast grid product for a preset future time period includes: The time series data of the water vapor and radar joint analysis field are preprocessed to obtain the characteristic sequence; The feature sequence is input into a pre-trained spatiotemporal deep learning model for forward propagation calculation, and standardized prediction features are output. The standardized forecast features are denormalized and reconstructed using spatial grids to generate gridded products for areal rainfall forecasts within a preset time period.

9. The GNSS and rainfall radar collaborative networking system for flash flood early warning according to claim 1, characterized in that, The method for generating graded early warning information is as follows: Preprocess the areal rainfall forecast products and real-time hydrological data to generate a dataset of precipitation and runoff conditions. The precipitation and runoff generation data set is input into a pre-trained watershed hydrological model to simulate runoff generation, confluence, and flood evolution, and output the hydrological process lines of each control section. Based on the simulated hydrological process line, combined with the watershed digital elevation model and the distribution of flood protection objects, an assessment of the flood inundation range and the population affected was carried out. Based on preset warning indicators and risk level thresholds, tiered warning information covering different regions and different degrees of severity is generated.

Images