A method for calculating the beam occlusion rate of weather radar based on the improvement of spatio-temporal resolution

Abstract

The present invention relates to a method for calculating the beam blockage rate of weather radar based on the improvement of spatio-temporal resolution, belonging to the field of radio. The present invention realizes the adaptive and precise spatio-temporal matching of meteorological unmanned aerial vehicle sounding data and ECMWF forecast data under different weather echo types, realizes the calculation of the spatial distance in the extended domain grid, realizes the preliminary extraction of spatio-temporal features, constructs a self-attention mechanism model integrated with physical perception, successfully generates high-spatial-resolution datasets of atmospheric temperature, atmospheric relative humidity, and atmospheric pressure, realizes the generation of high-spatio-temporal-resolution datasets of atmospheric temperature, atmospheric relative humidity, and atmospheric pressure, realizes the accurate calculation of the propagation path of radar beam rays with high spatio-temporal resolution, realizes the calculation of the occlusion azimuth and maximum elevation angle of the occluder in each azimuth ray beam block, and finally introduces a beam blockage power integration calculation model to successfully realize the accurate calculation of the weather radar beam blockage rate.

Description

Technical Field

[0001] The present invention relates to the field of radio, and particularly to a method for calculating the weather radar beam occlusion rate based on the improvement of spatio-temporal resolution. Background Art

[0002] The terrain in mountainous areas has a significant impact on the formation of airflow, precipitation, and meteorological phenomena. In particular, extreme weather phenomena such as hail and heavy precipitation usually develop complexly and are difficult to predict. Therefore, it is of great practical significance to conduct real-time monitoring of weather radars in such areas. Among various remote sensing observation means, weather radars have become the only effective means for monitoring the weather in complex terrain areas due to their ability to provide high temporal and spatial resolutions. Through radar detection, it is possible to more accurately monitor and predict weather phenomena such as precipitation and storms, especially catastrophic weather such as local heavy precipitation and mountain torrents. The real-time data of the radar can effectively capture these weather signals, providing key support for the disaster warning system, thereby reducing the risk of casualties and property losses. In addition, due to the large differences in climate characteristics in mountainous areas and the relatively unique local meteorological phenomena, radar data provides valuable local climate information for climatologists, helping to study the microscopic changes and weather laws of mountain climates. Through the accurate monitoring and forecasting of weather radars, relevant departments can issue meteorological warnings in a timely manner, improve the traffic management's response ability to extreme weather, and ensure traffic safety. At the same time, radar data can also provide more accurate meteorological services for local governments, agriculture, tourism and other industries, helping them make scientific decisions. However, the complex terrain in mountainous areas has a significant impact on the propagation of radar signals, especially the occlusion effect of tall mountains on radar beams. This occlusion effect will cause errors in the weather target parameters measured by the radar. In order to accurately describe the occlusion effect of the radar beam, the radar beam occlusion rate is usually used to quantify this impact. The calculation of the radar beam occlusion rate is a key issue in the application of weather radars in mountainous areas, aiming to quantify the impact of complex terrain on the propagation of radar signals. To ensure effective correction of measurement errors caused by the occlusion effect, it is particularly important to design an accurate method for calculating the radar beam occlusion rate.

[0003] Currently, there are mainly the following methods for calculating the beam blockage rate of weather radar: 1. The blockage rate calculation method based on terrain undulation. This method calculates the relative height relationship between the propagation path of the radar beam and the terrain through the elevation data of the terrain, so as to determine whether the beam is blocked by the terrain. However, it requires high-quality terrain data support. When the data accuracy and resolution are low, it may lead to errors, and the calculation process ignores the atmospheric refraction effect and the propagation path of the radar beam, which may also lead to certain errors. 2. The blockage rate calculation based on the ray tracing method. This method simulates the propagation path of the radar beam, traces the whole process of the radar wave from the emission point to the target point, and checks whether it intersects with the terrain to determine whether there is blockage. Its defect is that the calculation amount is large, especially when used in a large area, the calculation efficiency is low. 3. The statistical method based on the terrain blockage coefficient. This method estimates the blockage rate of the radar beam by statistically analyzing the high and low undulation characteristics of the terrain, grouping and weighting the terrain in the area. However, its disadvantage is that the accuracy is low, it ignores the specific propagation path, and it cannot consider the blockage effect of individual points. 4. The blockage rate calculation based on numerical simulation and wind field analysis. This method combines meteorological factors such as wind field, temperature, and pressure for numerical simulation, calculates the obstruction of the radar wave propagation in the atmosphere, and can simulate the influence of complex weather and atmospheric conditions on the radar beam. However, the calculation complexity is high, it has high requirements for the real-time and accuracy of data, and this method does not introduce accurate and real-time meteorological element observation data to effectively correct the numerical model output data, which will also produce large errors.

[0004] Therefore, when calculating the radar beam blockage rate, the existing methods do not fully consider the influence of the real atmospheric environment change on the electromagnetic wave propagation path. And the solution to this problem is crucial for improving the calculation accuracy of the weather radar beam blockage rate. The influence of the real atmospheric environment change on the electromagnetic wave propagation path highly depends on the real-time and accurate acquisition of high spatio-temporal resolution meteorological element data. However, a single observation device cannot provide the required high spatio-temporal resolution meteorological element data. Therefore, how to obtain the three-dimensional data of meteorological elements with both accuracy and high spatio-temporal resolution is particularly important. In addition, the existing weather radar beam blockage rate calculation methods usually do not fully utilize the high spatial resolution contour information of the obstacles for more refined small area division, which makes the calculation of the blockage rate in the blocked area unable to reach a higher accuracy. Summary of the Invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art, and provides a method for calculating the weather radar beam blockage rate based on the improvement of spatio-temporal resolution, which solves the deficiencies existing in the prior art.

[0006] The purpose of the present invention is achieved through the following technical solutions: A method for calculating the weather radar beam blockage rate based on the improvement of spatio-temporal resolution, the calculation method includes:

[0007] S1. Collect data, input the collected data into a geographic coordinate system generation model to obtain first data, determine the spatial search and matching area range based on the collected data, optimize the time matching maximum threshold by combining the first data, the initial classification scheme, and the time matching threshold optimization scheme, and extract all products within the time matching range and the spatial search area based on the collected data and the optimized time matching maximum threshold.

[0008] S2. Construct a first data set based on the collected data and the results obtained in S1, input the first data set and the collected data into a geodetic coordinate spatial distance calculation model to obtain second data, input the first data set and the results obtained in S1 into a three-dimensional grid center difference calculation model to obtain third data, and process the third data to obtain a second data set.

[0009] S3. Input the second data set into an architecture model to obtain a third data set, input the third data set and the second data set into a self-attention mechanism model to output a fourth data set, construct a refined prediction model based on the fourth data set, the second data set, and the collected data, obtain a fifth data set based on the collected data, the second data set, and the refined model, and obtain a sixth data set based on the fifth data set and a multivariate nonlinear fitting model.

[0010] S4. Obtain fourth data through the sixth data set, an atmospheric refractive index calculation model, and the collected data, screen and extract the collected data, screen the collected data with the extracted data to obtain a seventh data set, input the extracted data and the collected data into a geographic information processing system to obtain fifth data, obtain an eighth data set and a ninth data set through the collected data and the seventh data set, input the eighth data set and the ninth data set into the fourth data to obtain sixth data, input the sixth data and the collected data into a beam occlusion power integration calculation model to obtain seventh data, and process the seventh data to obtain the calculation result of the radar beam occlusion rate.

[0011] The data collection in S1 includes:

[0012] Collect real-time forecast products and auxiliary data of the European Centre for Medium-Range Weather Forecasts;

[0013] Collect meteorological data and auxiliary data of high-frequency downward radiosondes of meteorological drones;

[0014] Collect weather echo intensity data detected by weather radars and auxiliary data; collect spatial distribution data of geomorphic types with high spatial resolution and digital elevation data.

[0015] The real-time forecast products of the European Centre for Medium-Range Weather Forecasts include: three-dimensional atmospheric temperature datasets, atmospheric relative humidity datasets, and atmospheric pressure datasets at different longitudes, latitudes, and altitudes at different times, as well as surface pressure data. Among them, the atmospheric temperature dataset is denoted as the T_EC dataset, the atmospheric relative humidity dataset is denoted as the Q_EC dataset, the atmospheric pressure dataset is denoted as the P_EC dataset, and the surface pressure dataset is denoted as the P0_EC dataset;

[0016] The auxiliary data of the European Centre for Medium-Range Weather Forecasts include: time resolution, spatial resolution, longitude, latitude, and altitude information;

[0017] The meteorological data of the high-frequency downward radiosonde of the meteorological unmanned aerial vehicle include: atmospheric temperature datasets, atmospheric relative humidity datasets, and atmospheric pressure datasets at different longitudes, latitudes, and altitudes at different times. Among them, the atmospheric temperature dataset is denoted as the T_SO dataset, the atmospheric relative humidity dataset is denoted as the Q_SO dataset, and the atmospheric pressure dataset is denoted as the P_SO dataset;

[0018] The auxiliary data of the high-frequency downward radiosonde of the meteorological unmanned aerial vehicle include: the longitude, latitude, and altitude corresponding to the meteorological data, as well as the observation time data;

[0019] The auxiliary data detected by the weather radar include: the height, longitude, latitude, azimuth angle, and elevation angle of the radar antenna, the radar radial distance resolution, the radial distance, azimuth angle, elevation angle, and observation time where the weather echo intensity is located, the radar transmit power, antenna gain, and antenna beam width.

[0020] The specific content of S1 is as follows:

[0021] S101. Input the height, longitude, and latitude data of the radar antenna in the auxiliary data detected by the weather radar, as well as the radial distance, azimuth angle, and elevation angle data where the weather echo intensity is located, into the geographical coordinate system generation model. Run the model to output the longitude, latitude, and altitude information of the weather echo intensity data, denoted as [lon_wea, lat_wea, hei_wea], that is, the first data;

[0022] S102. Determine the spatial search and matching interval range, denoted as [lon_so, lat_so, hei_so], according to the maximum and minimum values of the longitude, latitude, and altitude information of the observation data of the high-frequency downward radiosonde of the meteorological unmanned aerial vehicle. Combine with [lon_wea, lat_wea, hei_wea] obtained in S101. First, extract the weather echo intensity data within the range of [lon_so, lat_so, hei_so], and then calculate its average value, denoted as Z_ave;

[0023] S103. Set the maximum initial threshold for time matching of different datasets as Δt_ori. Classify the weather echoes in the spatial search and matching region according to the initial classification scheme based on weather echo intensity. Then, optimize Δt_ori based on the time matching threshold optimization scheme for weather echo categories. The optimized maximum time matching threshold is denoted as Δt_new.

[0024] S104. Taking the moment of the meteorological data of the high-frequency downward radiosonde collected by the meteorological drone as the reference, form a time matching range by adding and subtracting Δt_new / 2 at this moment. Taking [lon_so, lat_so, hei_so] as the spatial search and matching region, extract all the real-time forecast products of the European Centre for Medium-Range Weather Forecasts within the time matching range and the spatial search and matching region, denoted as T1_EC, Q1_EC, and P1_EC.

[0025] The specific content of S2 is as follows:

[0026] S201. Select any point in the collected T_SO dataset, denoted as the T_SO_j point. Taking the spatial position of the T_SO_j point as the origin and using T1_EC as a three-dimensional grid, select all the vertices of the cubic grid where the T_SO_j point is located and the closest neighboring three-dimensional grids, thereby forming an extended domain dataset around the T_SO_j point, denoted as T2_EC_j_i, that is, the first dataset.

[0027] S202. Starting from i = 1, set the step as 1. The maximum value of i is the total number of grid points in the T2_EC_j_i dataset. Input the longitude, latitude, and altitude information of both the T_SO_j point and T2_EC_j_i into the geodetic coordinate spatial distance calculation model to obtain the extended domain spatial distance from the T_SO_j point to each point in the T2_EC_j_i dataset, denoted as T2_R_j_i, that is, the second data.

[0028] S203. Input T1_EC and T2_EC_j_i together into the three-dimensional grid central difference calculation model to obtain the gradients of each grid point in the T2_EC_j_i dataset in the longitude, latitude, and altitude directions, denoted as [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i], that is, the third data.

[0029] S204. Sum and average [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i] at each grid point to obtain the gradient average values in multiple dimensions, denoted as T2_G_j. Calculate the second-order derivatives of [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i] in the longitude, latitude, and height directions respectively, and input the generated second-order derivatives into the Hessian matrix calculation model to construct the curvature tensor matrix at each grid point. Sum all the elements in the curvature tensor matrix at each grid point to obtain the average value of the curvature tensor matrix elements, denoted as T2_H_j;

[0030] S205. Repeat S201 - S204 until all points in the T_SO dataset are selected, to obtain the extended domain dataset T3_EC, extended domain spatial distance T3_R, multi-dimensional gradient average value T3_G, and curvature tensor matrix element average value T3_H dataset that are spatio-temporally matched with the T_SO dataset;

[0031] S206. Apply the schemes of S201 - S205 to the Q_SO and P_SO data respectively to obtain the [Q3_EC, Q3_R, Q3_G, Q3_H] and [P3_EC, P3_R, P3_G, P3_H] datasets. The [T3_EC, T3_R, T3_G, T3_H] dataset, [Q3_EC, Q3_R, Q3_G, Q3_H] dataset, and [P3_EC, P3_R, P3_G, P3_H] dataset are the second datasets.

[0032] The step of inputting the second dataset into the architecture model to obtain the third dataset, inputting the third dataset and the second dataset into the self-attention mechanism model to output the fourth dataset, and constructing a refined prediction model based on the fourth dataset, the second dataset, and the collected data includes:

[0033] A1. Denote the generated [T3_EC, T3_R, T3_G, T3_H] dataset as T3_DS, and input T3_DS into the encoder and decoder architecture model with self-attention mechanism. Run the model to obtain the preliminary spatial feature dataset of T3_DS, denoted as T3_DS_FEA1, which is the third dataset;

[0034] A2. Input T3_DS_FEA1 and T3_DS together into the self-attention mechanism model integrated with physical perception. Run the model to obtain the refined spatial feature dataset of T3_DS after physical perception, denoted as T3_DS_FEA2, which is the fourth dataset;

[0035] A3. Take T3_DS_FEA2 and T3_DS as the input datasets, and the collected T_SO dataset as the output dataset. Construct a dynamic convolutional neural model based on thermodynamic consistency constraints, design a joint loss function, and conduct model training to obtain a refined spatial prediction model of atmospheric temperature.

[0036] A4. Apply the solutions in A1 - A3 to the [Q3_EC, Q3_R, Q3_G, Q3_H] and [P3_EC, P3_R, P3_G, P3_H] datasets respectively to obtain refined spatial prediction models of atmospheric relative humidity and atmospheric pressure.

[0037] The obtaining of the fifth dataset from the collected data, the second dataset, and the refined model in S3 includes:

[0038] B1. Set a spatial resolution equal to the obtained radar radial distance resolution, denoted as ΔR_high, and construct a three - dimensional grid with high spatial resolution, denoted as Grid_high.

[0039] B2. Replace the T_SO dataset with the Grid_high grid, and replace the T1_EC dataset with the collected T_EC dataset. Repeat the processing flow of S2 and steps A1 - A4 to obtain a three - dimensional atmospheric temperature dataset with high spatial resolution, denoted as T_space_high.

[0040] B3. Replace the T1_EC dataset with the Q_EC and P_EC datasets respectively to obtain three - dimensional atmospheric relative humidity and atmospheric pressure datasets with high spatial resolution, denoted as Q_space_high and P_space_high. T_space_high, Q_space_high, and P_space_high are the fifth dataset.

[0041] The obtaining of the sixth dataset from the fifth dataset and the multi - variable non - linear fitting model in S3 includes:

[0042] C1. For the generated T_space_high dataset, take the observed data of this dataset changing with time in the three - dimensional directions of longitude, latitude, and altitude, denoted as T_time_lon, T_time_lat, T_time_hei. The corresponding observed time vector is denoted as time1, and the corresponding longitude, latitude, and altitude vectors are denoted as lon1, lat1, and height1 respectively.

[0043] C2. Input [T_time_lon, time1, lon1], [T_time_lat, time1, lat1], and [T_time_hei, time1, height1] into the multivariate non - linear fitting model respectively. Run the model to obtain non - linear trend models of high - spatial - resolution atmospheric temperature in the longitude, latitude, and height directions at the observation time, denoted as Model1, Model2, and Model3 respectively;

[0044] C3. Set the high - time resolution as Δt_high, and construct an observation - time vector with high - time resolution, denoted as time_high = [0, Δt_high, 2×Δt_high, …, n×Δt_high], where n is the length of the time vector;

[0045] C4. Input time_high and the corresponding longitude, latitude, and height data into Model1, Model2, and Model3 respectively, and combine the outputs of the three models in the three - dimensional directions of longitude, latitude, and height to obtain a three - dimensional atmospheric temperature dataset with high spatio - temporal resolution, denoted as T_space_time_high;

[0046] C5. Apply the solutions of C1 - C4 to the Q_space_high and P_space_high datasets to obtain three - dimensional atmospheric relative humidity and atmospheric pressure datasets with high spatio - temporal resolution, denoted as Q_space_time_high and P_space_time_high respectively. T_space_time_high, Q_space_time_high, and P_space_time_high are the sixth datasets.

[0047] The obtaining of the fourth data from the sixth dataset, the atmospheric refractive index calculation model, and the collected data in S4 includes:

[0048] D1. Input the generated sixth dataset into the atmospheric refractive index calculation model and run the model to obtain three - dimensional atmospheric refractive index data with high spatio - temporal resolution;

[0049] D2. Input the collected radar antenna azimuth and elevation angles, and the generated three - dimensional atmospheric refractive index data into the radar ray propagation path model and run the model to obtain the propagation paths of the radar beam center, upper edge, and lower edge rays starting from the radar antenna and ending at any point in the high - spatio - temporal resolution grid, which is the fourth data.

[0050] In step S4, the collected data is screened and extracted. The data after extraction is used to screen the collected data to obtain the seventh dataset. The data after extraction and the collected data are input into a geographic information processing system to obtain the fifth data. The eighth dataset and the ninth dataset are obtained from the collected data and the seventh dataset. The eighth dataset and the ninth dataset are input into the fourth data to obtain the sixth data. The sixth data and the collected data are input into a beam occlusion power integration calculation model to obtain the seventh data. After processing the seventh data, the calculation result of the radar beam occlusion rate includes:

[0051] E1. According to the collected spatial distribution data of high-spatial-resolution landform types, the data belonging to the mountain and tall building types are screened out, marked as occlusion object data, and the longitude and latitude data of the locations of such occlusion object data are extracted, denoted as [lon_c, lat_c];

[0052] E2. According to the collected high-spatial-resolution DEM data and [lon_c, lat_c], the DEM dataset of the occlusion object data is screened out, denoted as DEM_c, that is, the seventh dataset. DEM represents digital elevation;

[0053] E3. [lon_c, lat_c], the altitude, longitude, and latitude of the collected radar antenna are input into a geographic information processing system. After running the system, the azimuth range [AZ_c1, AZ_c2] of the distribution of the occlusion object data in the polar coordinate system with the radar antenna as the origin is obtained, that is, the fifth data;

[0054] E4. Using the beam width of the collected radar antenna as the azimuth division interval, the space within [AZ_c1, AZ_c2] is divided into several azimuth beam resolution volumes. Taking any one azimuth beam resolution volume, with the spatial resolution of the DEM_c dataset as the step size, in the azimuth direction, the azimuth beam resolution volume is further divided into several azimuth ray beam blocks, and the number is denoted as N_c2;

[0055] E5. Combining with the DEM_c dataset, the maximum value of the DEM in each azimuth ray beam block is screened out to form the dataset DEM_c_max, that is, the eighth dataset. At the same time, the longitude and latitude where each data in the DEM_c_max dataset appears are recorded to form the dataset lon_lat_c_max, that is, the ninth dataset;

[0056] E6. Input DEM_c_max and lon_lat_c_max into the propagation paths of the beam center, upper edge, and lower edge rays of the generated radar beam at any point in space to obtain the azimuth range of the obstacle relative to the radar beam centerline and the maximum elevation angle of occlusion in each azimuth ray beam block, denoted as [AZ_c3, AZ_c4] and ELE_c_max respectively, which is the sixth data.

[0057] E7. Input [AZ_c3, AZ_c4], ELE_c_max, and the collected radar transmission power and antenna gain into the beam occlusion power integration calculation model, and run the model to obtain the power occluded by the obstacle in each azimuth ray beam block, denoted as Power_ci, which is the seventh data, where ci = 1, 2, …, N_c2.

[0058] E8. Sum up these N_c2 power values to obtain the sum of the radar transmission power occluded by the obstacle in the azimuth beam resolution volume, denoted as Power_C. Divide Power_C by the collected radar transmission power to obtain the calculation result of the radar beam occlusion rate in the azimuth beam resolution volume.

[0059] The present invention has the following advantages:

[0060] 1. Based on weather radar echo data, meteorological unmanned aerial vehicle (UAV) high-frequency downward radiosonde data, and ECMWF forecast data, a time-matching threshold optimization scheme based on echo categories is designed to achieve the adaptive and accurate spatio-temporal matching of meteorological UAV sounding data and ECMWF forecast data under different weather echo types. On this basis, using the accurately matched ECMWF forecast data and meteorological UAV sounding data, an extended domain three-dimensional grid determination scheme is designed, an extended domain three-dimensional grid data set is constructed, and the calculation of spatial distance in the extended domain grid is realized.

[0061] 2. Combine the three-dimensional grid central difference calculation model and the Hessian matrix calculation model to calculate the multi-dimensional gradients and curvature tensors at each point in the extended domain three-dimensional grid. Based on these data sets (including ECMWF data set, spatial distance, multi-dimensional gradients, and curvature tensor matrix), an encoder-decoder architecture model with self-attention mechanism is constructed to realize the preliminary extraction of spatio-temporal features, a self-attention mechanism model integrating physical perception is constructed to refine the preliminarily extracted spatial features, and a dynamic convolutional neural network model based on thermodynamic consistency constraints is introduced to successfully generate high-spatial-resolution data sets of atmospheric temperature, atmospheric relative humidity, and atmospheric pressure.

[0062] 3. Based on the generated multi-moment high-spatial-resolution atmospheric dataset, a multivariate non-linear trend model was designed to further refine the fitting relationship between the data in the directions of longitude, latitude, altitude, etc. and the observation time, thereby realizing the generation of high spatio-temporal resolution datasets of atmospheric temperature, atmospheric relative humidity, and atmospheric pressure. Based on the generated high spatio-temporal resolution datasets of atmospheric temperature, atmospheric relative humidity, and atmospheric pressure, as well as the radar observation parameters, a calculation model of atmospheric refractive index and radar ray propagation path was introduced to accurately calculate the propagation path of high spatio-temporal resolution radar beams.

[0063] 4. Based on the spatial distribution data of landform types, digital elevation model (DEM), and radar detection auxiliary data, a method for extracting obstacle data and determining the azimuth range was designed. Combining the calculation results of the propagation path of high spatio-temporal resolution radar beams, a refined beam occlusion rate calculation method based on azimuth ray beam block division was constructed to calculate the occlusion azimuth and maximum elevation angle of obstacles in each azimuth ray beam block. Finally, a beam occlusion power integration calculation model was introduced to successfully achieve the accurate calculation of the weather radar beam occlusion rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a flowchart of the process of the present invention;

[0065] Figure 2 It is a flowchart for realizing the generation of the HRES dataset of ECMWF within the time matching and spatial search matching ranges of the present invention;

[0066] Figure 3 It is a flowchart for realizing the calculation of the extended domain spatial distance, multi-dimensional gradient calculation, and curvature tensor matrix generation of the present invention;

[0067] Figure 4 It is a flowchart for realizing the atmospheric temperature spatial refinement prediction model of the present invention;

[0068] Figure 5 It is a flowchart for realizing the generation of a high spatio-temporal resolution three-dimensional atmospheric temperature dataset of the present invention;

[0069] Figure 6 It is a flowchart for realizing the propagation path of high spatio-temporal resolution weather radar beams of the present invention;

[0070] Figure 7 It is a flowchart for realizing the calculation of the weather radar beam occlusion rate of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0071] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Usually, the components of the embodiments of this application described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application provided in the following with reference to the accompanying drawings is not intended to limit the protection scope of the claimed application, but only represents the selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative efforts belong to the protection scope of this application. The following further describes the present invention with reference to the accompanying drawings.

[0072] As Figure 1 shown, the present invention specifically relates to a method for calculating the beam occlusion rate of a weather radar based on the improvement of spatio-temporal resolution, which specifically includes the following contents:

[0073] Step 1: Collect real-time forecast products (HRES) of the European Centre for Medium-Range Weather Forecasts (ECMWF) and auxiliary data; collect meteorological data and auxiliary data of high-frequency down-dropped radiosondes of meteorological drones; collect weather echo intensity data detected by weather radars and auxiliary data; collect spatial distribution data of landform types with high spatial resolution and digital elevation (DEM) data;

[0074] Among them, the HRES products of ECMWF mainly include: three-dimensional atmospheric temperature (T_EC), atmospheric relative humidity (Q_EC), and atmospheric pressure (P_EC) data, etc. at different longitudes, latitudes, and altitudes at different times, as well as surface pressure data (P0_EC);

[0075] The auxiliary data of the HRES products of ECMWF mainly include: a time resolution of 1 hour, a spatial resolution of 9 km, and longitude, latitude, and altitude information;

[0076] The meteorological data of high-frequency down-dropped radiosondes of meteorological drones mainly include: atmospheric temperature data sets (T_SO data sets), atmospheric relative humidity data sets (Q_SO data sets), atmospheric pressure data sets (P_SO) data sets, etc. at different longitudes, latitudes, and altitudes at different times;

[0077] Among them, the auxiliary data of high-frequency down-dropped radiosondes of meteorological drones mainly include: longitude, latitude, altitude corresponding to the meteorological data, and observation time data;

[0078] Among them, the auxiliary data detected by the weather radar mainly include: the altitude, longitude, latitude, azimuth angle, elevation angle where the radar antenna is located, the radar radial distance resolution of 125 meters, the radial distance, azimuth angle, elevation angle where the weather echo intensity is located, and the observation time, etc., the radar transmitting power of 75 kw, the antenna gain of 44 dBi, and the antenna beam width of 1 degree;

[0079] Step 2: As Figure 2 shown, first, input the altitude, longitude, and latitude data of the radar antenna in the auxiliary data detected by the radar collected in Step 1, as well as the radial distance, azimuth angle, and elevation angle data where the weather echo intensity is located into the geographic coordinate system generation model, and run the model. The output of the model is the longitude, latitude, and altitude information of the weather echo intensity data, denoted as [lon_wea, lat_wea, hei_wea].

[0080] Secondly, determine the range of the spatial search and matching area, denoted as [lon_so, lat_so, hei_so], according to the maximum and minimum values of the longitude, latitude, and altitude information in the observation data of the high-frequency downward radiosonde of the meteorological unmanned aerial vehicle collected in Step 1. Combine [lon_wea, lat_wea, hei_wea] processed in this step. First, extract the weather echo intensity data within the range of [lon_so, lat_so, hei_so], and then calculate its average value, denoted as Z_ave.

[0081] After that, set the maximum initial threshold Δt_ori = 10 minutes for time matching of different data sets, classify the weather echoes in the spatial search and matching area according to the initial classification scheme based on the weather echo intensity, and then optimize Δt_ori according to the time matching threshold optimization scheme for the weather echo category. The optimized maximum time matching threshold is denoted as Δt_new. Finally, taking the time of the observation data (T_SO, Q_SO, P_SO) of the high-frequency downward radiosonde of the meteorological unmanned aerial vehicle collected in Step 1 as the benchmark, form a time matching range by adding and subtracting Δt_new / 2 at this benchmark time, and taking [lon_so, lat_so, hei_so] as the spatial search and matching area, extract all ECMWF's HRES products (T_EC, Q_EC, P_EC) within the time matching range and the spatial search and matching area, denoted as T1_EC, Q1_EC, P1_EC.

[0082] Further, the initial classification scheme based on weather echo intensity is as follows: when the average value of weather echo intensity is less than or equal to 10 dBz, the weather phenomenon is determined to be clear sky; when the average value of weather echo intensity is greater than 10 dBz and less than or equal to 30 dBz, the weather phenomenon is determined to be light rain; when the average value of weather echo intensity is greater than 30 dBz and less than or equal to 40 dBz, the weather phenomenon is determined to be moderate rain; when the average value of weather echo intensity is greater than 40 dBz and less than or equal to 50 dBz, the weather phenomenon is determined to be heavy rain; when the average value of weather echo intensity is greater than 50 dBz, the weather phenomenon is determined to be rainstorm.

[0083] Further, the time matching threshold optimization scheme based on weather echo category is: Δt_new = N_kind × Δt_ori, where N_kind is the time scaling factor, which is N1 = 6, N2 = 4, N3 = 3, N4 = 2, N5 = 1 under the conditions of clear sky, light rain, moderate rain, heavy rain and rainstorm respectively.

[0084] Step 3: As Figure 3 shown, first, select any point in the T_SO dataset collected in Step 1, denoted as the T_SO_j point. Taking the spatial position (longitude, latitude and altitude) of the T_SO_j point as the origin, and using the T1_EC dataset generated in Step 2 as a three-dimensional grid, select all the vertices of the cubic grid where the T_SO_j point is located and the closest neighboring three-dimensional grids, so as to form an extended domain dataset around the T_SO_j point, denoted as T2_EC_j_i, where i starts from 1, with a step size of 1, and the maximum value of i is 26, which is the total number of grid points in the T2_EC_j_i dataset. Input the longitude, latitude and altitude information of both the T_SO_j point and T2_EC_j_i into the geodetic coordinate space distance calculation model, and the output of the model is the extended domain space distance from the T_SO_j point to each point in the T2_EC_j_i dataset, denoted as T2_R_j_i.

[0085] Secondly, input both T1_EC and T2_EC_j_i into the three-dimensional grid central difference calculation model, run the model, and the output of the model is the gradient of each grid point in the T2_EC_j_i dataset in the longitude, latitude and altitude directions, denoted as [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i].

[0086] After that, sum and average [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i] at each grid point to obtain the average gradient values in multiple dimensions, denoted as T2_G_j. Then, calculate the second-order derivatives of [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i] in the three directions of longitude, latitude, and altitude respectively, and input the generated second-order derivatives into the Hessian matrix calculation model to construct the curvature tensor matrix at each grid point. After that, sum all the elements in the curvature tensor matrix at each grid point to obtain the average value of the curvature tensor matrix elements, denoted as T2_H_j.

[0087] Finally, apply the above scheme in this step to all points in the T_SO dataset, and datasets such as the extended domain dataset T3_EC, the extended domain spatial distance T3_R, the multi-dimensional gradient average value T3_G, and the average value of the curvature tensor matrix elements T3_H that are spatio-temporally matched with the T_SO dataset can be obtained. Apply the above scheme to the Q_SO and P_SO datasets respectively, and datasets such as [Q3_EC, Q3_R, Q3_G, Q3_H] and [P3_EC, P3_R, P3_G, P3_H] can be obtained respectively.

[0088] Step 4: As Figure 4 shown, first, denote the [T3_EC, T3_R, T3_G, T3_H] datasets generated in Step 3 as T3_DS. Input T3_DS into the encoder and decoder architecture model introduced with the self-attention mechanism, run the model, and the output of the model is the preliminary spatial feature dataset of T3_DS, denoted as T3_DS_FEA1.

[0089] Secondly, input T3_DS_FEA1 and T3_DS together into the self-attention mechanism model integrated with physical perception, run the model, and the output of the model is the refined spatial feature dataset of T3_DS after physical perception, denoted as T3_DS_FEA2. Then, use T3_DS_FEA2 and T3_DS as the input datasets, and T_SO collected in Step 1 as the output dataset, construct a dynamic convolutional neural model based on thermodynamic consistency constraints, design a joint loss function, and conduct the training of the model, that is, obtain the spatial refinement prediction model of atmospheric temperature. Apply the above scheme to datasets such as [Q3_EC, Q3_R, Q3_G, Q3_H] and [P3_EC, P3_R, P3_G, P3_H] respectively, and the spatial refinement prediction models of atmospheric relative humidity and atmospheric pressure can be obtained respectively.

[0090] Furthermore, design the joint loss function as the sum of the mean square error between the model output value and the observation value of the high-frequency down-dropped radiosonde of the meteorological unmanned aerial vehicle, the gradient smoothing loss, and the physical perception loss.

[0091] Step 5: First, set the spatial resolution to be the same as the radar radial distance resolution obtained in Step 1, denoted as ΔR_high = 125 m, and construct a three-dimensional grid with high spatial resolution, denoted as Grid_high.

[0092] Secondly, replace the T_SO dataset in Step 3 with the Grid_high grid, and replace the T1_EC dataset with the T_EC dataset collected in Step 1. Repeat the processing flow of Steps 3 and 4 to obtain a three-dimensional atmospheric temperature dataset with high spatial resolution, denoted as T_space_high.

[0093] Finally, replace the T1_EC dataset in this step with the Q_EC and P_EC datasets respectively to obtain three-dimensional atmospheric relative humidity and atmospheric pressure datasets with high spatial resolution, denoted as Q_space_high and P_space_high.

[0094] Step 6: As Figure 5 shown, first, collect the T_space_high datasets at multiple moments generated in Step 5, and respectively take the observational data of the dataset varying with time in the three-dimensional directions of longitude, latitude, and altitude, denoted as T_time_lon, T_time_lat, T_time_hei. The corresponding observational moment vector is denoted as time1, and the corresponding longitude, latitude, and altitude vectors are denoted as lon1, lat1, and height1 respectively.

[0095] Secondly, input [T_time_lon, time1, lon1], [T_time_lat, time1, lat1], and [T_time_hei, time1, height1] into the multivariate non-linear fitting model respectively, and run the model. The outputs of the model are non-linear trend models of the atmospheric temperature with high spatial resolution and the observational moment in the longitude, latitude, and altitude directions respectively, denoted as Model1, Model2, and Model3.

[0096] After that, set the high time resolution to be Δt_high, and construct an observational moment vector with high time resolution, denoted as time_high = [0, Δt_high, 2×Δt_high, …, n×Δt_high], where n is the length of the time vector.

[0097] Finally, input time_high and the corresponding longitude, latitude, and altitude data into Model1, Model2, and Model3 respectively, and combine the outputs of the three models in the three-dimensional directions of longitude, latitude, and altitude to obtain a three-dimensional atmospheric temperature dataset with high spatio-temporal resolution, denoted as T_space_time_high.

[0098] Apply the above scheme to the Q_space_high and P_space_high datasets generated in step 5, and respectively obtain three-dimensional atmospheric relative humidity and atmospheric pressure datasets with high spatio-temporal resolution, denoted as Q_space_time_high and P_space_time_high.

[0099] Step 7: As Figure 6 shown, first, input T_space_time_high, Q_space_time_high, and P_space_time_high generated in step 6 into the atmospheric refractive index calculation model and run the model. The output of the model is three-dimensional atmospheric refractive index data with high spatio-temporal resolution.

[0100] Secondly, input the radar antenna azimuth and elevation angles collected in step 1 and the three-dimensional atmospheric refractive index data generated in this step into the radar ray propagation path model and run the model. The output of the model is the propagation paths of the radar beam center, upper edge, and lower edge rays starting from the radar antenna and ending at any point in the high spatio-temporal resolution grid.

[0101] Step 8: As Figure 7 shown, first, according to the spatial distribution data of landform types with high spatial resolution collected in step 1, screen out the data belonging to types such as mountains and tall buildings, mark them as obstacle data, and extract the longitude and latitude data of the locations of such obstacle data, denoted as [lon_c, lat_c].

[0102] Secondly, according to the high spatial resolution DEM data collected in step 1 and [lon_c, lat_c], screen out the DEM dataset of the obstacle data, denoted as DEM_c.

[0103] After that, input [lon_c, lat_c], the altitude, longitude, and latitude of the radar antenna collected in step 1 into the geographic information processing system and run the system. The output of the system is the azimuth range [AZ_c1, AZ_c2] of the distribution of the obstacle data in the polar coordinate system with the radar antenna as the origin.

[0104] After that, using the radar antenna beam width collected in Step 1 as the azimuth division interval, the space within [AZ_c1, AZ_c2] is divided into several azimuth beam resolution volumes.

[0105] After that, take any one of the azimuth beam resolution volumes. Using the spatial resolution of the DEM_c dataset as the step size, in the azimuth direction, further divide this azimuth beam resolution volume into several azimuth ray beam blocks, and the number is denoted as N_c2. Combining with the DEM_c dataset, screen out the maximum value of the DEM within each azimuth ray beam block to form the dataset DEM_c_max. At the same time, record the longitude and latitude where each data in the DEM_c_max dataset appears to form the dataset lon_lat_c_max.

[0106] After that, input DEM_c_max and lon_lat_c_max into the propagation paths of the beam center, upper edge, and lower edge rays of the radar beam generated in Step 7 at any point in space, and obtain the azimuth range of the obstacle relative to the radar beam center line and the maximum elevation angle of occlusion in each azimuth ray beam block, which are denoted as [AZ_c3, AZ_c4] and ELE_c_max respectively. After that, input [AZ_c3, AZ_c4], ELE_c_max, and the radar transmit power and antenna gain collected in Step 1 into the beam occlusion power integration calculation model, and run the model. The output of the model is the power occluded by the obstacle in each azimuth ray beam block, denoted as Power_ci, where ci = 1, 2, …, N_c2. Then superimpose these N_c2 power values to obtain the sum of the radar transmit powers occluded by the obstacle within this azimuth beam resolution volume, denoted as Power_C.

[0107] Finally, divide Power_C by the radar transmit power collected in Step 1 to obtain the calculation result of the radar beam occlusion rate in this azimuth beam resolution volume.

[0108] The above is only the preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications, and improvements, and can be changed within the scope of the concept described herein through the above teachings or the technology or knowledge in related fields. And the changes and alterations made by those skilled in the art that do not depart from the spirit and scope of the present invention shall fall within the protection scope of the appended claims of the present invention.

Claims

1. A method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution, characterized in that: The calculation method includes: S1. Collect data, input the collected data into the geographic coordinate system generation model to obtain the first data, determine the spatial search and matching area range according to the collected data, and optimize the time matching maximum threshold by combining the first data, the initial classification scheme and the time matching threshold optimization scheme. Extract all products within the time matching range and the spatial search area according to the collected data and the optimized time matching maximum threshold; S2. Construct the first data set according to the collected data and the results obtained in S1. Input the first data set and the collected data into the geodetic coordinate spatial distance calculation model to obtain the second data. Input the first data set and the results obtained in S1 into the stereoscopic grid center difference calculation model to obtain the third data. Process the second data and the third data to obtain the second data set; S3. Input the second data set into the architecture model to obtain the third data set. Input the third data set and the second data set into the self-attention mechanism model to output the fourth data set. Construct a refined prediction model according to the fourth data set, the second data set and the collected data. Obtain the fifth data set according to the collected data, the second data set and the refined model, and obtain the sixth data set based on the fifth data set and the multivariate nonlinear fitting model; S4. Obtain the fourth data through the sixth data set, the atmospheric refractive index calculation model and the collected data. Screen and extract the collected data to obtain the seventh data set with the extracted data. Input the extracted data and the collected data into the geographic information processing system to obtain the fifth data. Obtain the eighth data set and the ninth data set through the collected data and the seventh data set. Input the eighth data set and the ninth data set into the fourth data to obtain the sixth data. Input the sixth data and the collected data into the beam occlusion power integration calculation model to obtain the seventh data. Process the seventh data to obtain the calculation result of the radar beam occlusion rate; The obtaining of the fourth data through the sixth data set, the atmospheric refractive index calculation model and the collected data in S4 includes: D1. Input the generated sixth data set into the atmospheric refractive index calculation model and run the model to obtain three-dimensional atmospheric refractive index data with high spatio-temporal resolution; D2. Input the collected radar antenna azimuth and elevation angles, and the generated three-dimensional atmospheric refractive index data into the radar ray propagation path model and run the model to obtain the propagation paths of the radar beam center, upper edge and lower edge rays starting from the radar antenna and ending at any point in the high spatio-temporal resolution grid, that is, the fourth data.

2. The method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution according to claim 1, wherein: The data collection in S1 includes: Collect real-time forecast products and auxiliary data of the European Centre for Medium-Range Weather Forecasts; Collect meteorological data and auxiliary data of meteorological unmanned aerial vehicles with high-frequency down-dropped radiosondes; Collect weather echo intensity data detected by weather radars and auxiliary data; Collect spatial distribution data of landform types with high spatial resolution and digital elevation data.

3. A method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution according to claim 2, characterized in that: The real-time forecast products of the European Centre for Medium-Range Weather Forecasts include: three-dimensional atmospheric temperature datasets, atmospheric relative humidity datasets, and atmospheric pressure datasets at different longitudes, latitudes, and altitudes at different times, as well as surface pressure data. Among them, the atmospheric temperature dataset is denoted as the T_EC dataset, the atmospheric relative humidity dataset is denoted as the Q_EC dataset, the atmospheric pressure dataset is denoted as the P_EC dataset, and the surface pressure dataset is denoted as the P0_EC dataset; The auxiliary data of the European Centre for Medium-Range Weather Forecasts include: time resolution, spatial resolution, longitude, latitude, and altitude information; The meteorological data of the high-frequency downward radiosonde of the meteorological unmanned aerial vehicle include: atmospheric temperature datasets, atmospheric relative humidity datasets, and atmospheric pressure datasets at different longitudes, latitudes, and altitudes at different times. Among them, the atmospheric temperature dataset is denoted as the T_SO dataset, the atmospheric relative humidity dataset is denoted as the Q_SO dataset, and the atmospheric pressure dataset is denoted as the P_SO dataset; The auxiliary data of the high-frequency downward radiosonde of the meteorological unmanned aerial vehicle include: the longitude, latitude, and altitude corresponding to the meteorological data, as well as the observation time data; The auxiliary data detected by the weather radar include: the height, longitude, latitude, azimuth angle, and elevation angle of the radar antenna, the radar radial distance resolution, the radial distance, azimuth angle, elevation angle, and observation time of the weather echo intensity, the radar transmission power, antenna gain, and antenna beam width.

4. A method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution according to claim 3, characterized in that: The specific content of S1 is as follows: S101. Input the height, longitude, and latitude data of the radar antenna in the auxiliary data detected by the weather radar, as well as the radial distance, azimuth angle, and elevation angle data of the weather echo intensity into the geographic coordinate system generation model, and run the model to output the longitude, latitude, and altitude information of the weather echo intensity data, denoted as [lon_wea, lat_wea, hei_wea], that is, the first data; S102. Determine the spatial search and matching interval range, denoted as [lon_so, lat_so, hei_so], according to the maximum and minimum values of the longitude, latitude, and altitude information of the observation data of the high-frequency downward radiosonde of the meteorological unmanned aerial vehicle collected. Combine [lon_wea, lat_wea, hei_wea] obtained in S101, first extract the weather echo intensity data within the range of [lon_so, lat_so, hei_so], and then calculate its average value, denoted as Z_ave; S103. Set the maximum initial threshold for time matching of different datasets as Δt_ori. Classify the weather echoes in the spatial search and matching area according to the initial classification scheme based on the weather echo intensity, and then optimize Δt_ori according to the time matching threshold optimization scheme for the weather echo category. The optimized maximum time matching threshold is denoted as Δt_new; S104. Taking the moment when the meteorological data of the high-frequency down-dropping radiosonde of the meteorological UAV are collected as the benchmark, adding and subtracting Δt_new / 2 at this moment to form a time matching range, and using [lon_so, lat_so, hei_so] as the spatial search matching area, extracting all the real-time forecast products of the European Centre for Medium-Range Weather Forecasts within the time matching range and the spatial search matching area, denoted as T1_EC, Q1_EC, and P1_EC.

5. A method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution according to claim 4, characterized in that: The specific content of S2 is as follows: S201. Selecting any point in the collected T_SO dataset, denoted as the T_SO_j point. Taking the spatial position of the T_SO_j point as the origin and using T1_EC as a three-dimensional grid, selecting all the vertices of the cube grid where the T_SO_j point is located and the closest three-dimensional grids, thereby constituting an extended domain dataset around the T_SO_j point, denoted as T2_EC_j_i, that is, the first dataset, where i starts from 1, the setting step is 1, and the maximum value of i is the total number of grid points in the T2_EC_j_i dataset; S202. Inputting the longitude, latitude, and altitude information of both the T_SO_j point and T2_EC_j_i into the geodetic coordinate space distance calculation model to obtain the extended domain space distance from the T_SO_j point to each point in the T2_EC_j_i dataset, denoted as T2_R_j_i, that is, the second data; S203. Inputting T1_EC and T2_EC_j_i into the three-dimensional grid central difference calculation model to obtain the gradients of each grid point in the T2_EC_j_i dataset in the longitude, latitude, and altitude directions, denoted as [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i], that is, the third data; S204. Summing and averaging [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i] for each grid point to obtain the gradient average values in multiple dimensions, denoted as T2_G_j. Taking the second derivatives of [T2_G_lon_j_i, T2_G_lat_j_i, T2_G_hei_j_i] in the longitude, latitude, and altitude directions respectively, inputting the generated second derivatives into the Hessian matrix calculation model to construct the curvature tensor matrix for each grid point, and summing all the elements in the curvature tensor matrix for each grid point to obtain the average value of the curvature tensor matrix elements, denoted as T2_H_j; S205. Repeating S201 - S204 until all points in the T_SO dataset are selected, obtaining the extended domain dataset T3_EC dataset, the extended domain space distance T3_R dataset, the multi-dimensional gradient average T3_G dataset, and the average value of the curvature tensor matrix elements T3_H dataset that are spatio-temporally matched with the T_SO dataset; S206. Apply the solutions of S201 - S205 to the Q_SO and P_SO data respectively to obtain the [Q3_EC, Q3_R, Q3_G, Q3_H] and [P3_EC, P3_R, P3_G, P3_H] data sets. The [T3_EC, T3_R, T3_G, T3_H] data set, the [Q3_EC, Q3_R, Q3_G, Q3_H] data set, and the [P3_EC, P3_R, P3_G, P3_H] data set are the second data sets.

6. A method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution according to claim 5, characterized in that: In S3, input the second data set into the architecture model to obtain the third data set, input the third data set and the second data set into the self - attention mechanism model to output the fourth data set. Constructing the refined prediction model based on the fourth data set, the second data set, and the collected data includes: A1. Denote the generated [T3_EC, T3_R, T3_G, T3_H] data set as T3_DS, and input T3_DS into the encoder - decoder architecture model with self - attention mechanism. Run the model to obtain the preliminary spatial feature data set of T3_DS, denoted as T3_DS_FEA1, which is the third data set. A2. Input T3_DS_FEA1 and T3_DS together into the self - attention mechanism model integrated with physical perception. Run the model to obtain the refined spatial feature data set of T3_DS after physical perception, denoted as T3_DS_FEA2, which is the fourth data set. A3. Use T3_DS_FEA2 and T3_DS as the input data set, and the collected T_SO data set as the output data set. Construct a dynamic convolutional neural model based on thermodynamic consistency constraints, design a joint loss function, and conduct model training to obtain the refined spatial prediction model of atmospheric temperature. A4. Apply the solutions of A1 - A3 to the [Q3_EC, Q3_R, Q3_G, Q3_H] and [P3_EC, P3_R, P3_G, P3_H] data sets respectively to obtain the refined spatial prediction models of atmospheric relative humidity and atmospheric pressure.

7. A method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution according to claim 6, characterized in that: In S3, obtaining the fifth data set based on the collected data, the second data set, and the refined model includes: B1. Set the spatial resolution the same as the obtained radar radial distance resolution, denoted as ΔR_high, and construct a three - dimensional grid with high spatial resolution, denoted as Grid_high. B2. Replace the T_SO data set with the Grid_high grid, replace the T1_EC data set with the collected T_EC data set, repeat the processing flow of S2 steps and A1 - A4 steps to obtain the three - dimensional atmospheric temperature data set with high spatial resolution, denoted as T_space_high. B3. Replace the T1_EC dataset with the Q_EC and P_EC datasets respectively to obtain high-spatial-resolution three-dimensional atmospheric relative humidity and atmospheric pressure datasets, denoted as Q_space_high and P_space_high. T_space_high, Q_space_high, and P_space_high are the fifth dataset.

8. A method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution according to claim 7, characterized in that: The obtaining of the sixth dataset based on the fifth dataset and the multi-variable non-linear fitting model in S3 includes: C1. For the generated T_space_high dataset, take the observational data of the dataset varying with time in the three-dimensional directions of longitude, latitude, and altitude, denoted as T_time_lon, T_time_lat, T_time_hei. The corresponding observational time vector is denoted as time1, and the corresponding longitude, latitude, and altitude vectors are denoted as lon1, lat1, and height1 respectively. C2. Input [T_time_lon, time1, lon1], [T_time_lat, time1, lat1], and [T_time_hei, time1, height1] into the multi-variable non-linear fitting model respectively. Run the model to obtain non-linear trend models of high-spatial-resolution atmospheric temperature and observational time in the longitude, latitude, and altitude directions, denoted as Model1, Model2, and Model3. C3. Set the high-time resolution as Δt_high, and construct a high-time-resolution observational time vector, denoted as time_high = [0, Δt_high, 2×Δt_high, …, n×Δt_high], where n is the length of the time vector. C4. Input time_high and the corresponding longitude, latitude, and altitude data into Model1, Model2, and Model3 respectively, and combine the outputs of the three models in the three-dimensional directions of longitude, latitude, and altitude to obtain a high-spatiotemporal-resolution three-dimensional atmospheric temperature dataset, denoted as T_space_time_high. C5. Apply the solutions in C1 - C4 to the Q_space_high and P_space_high datasets to obtain high-spatiotemporal-resolution three-dimensional atmospheric relative humidity and atmospheric pressure datasets, denoted as Q_space_time_high and P_space_time_high. T_space_time_high, Q_space_time_high, and P_space_time_high are the sixth dataset.

9. A method for calculating the beam occlusion rate of a weather radar based on improving spatio-temporal resolution according to claim 1, characterized in that: The screening and extraction of the collected data in S4 is carried out to screen the collected data with the extracted data to obtain the seventh data set. The extracted data and the collected data are input into the geographic information processing system to obtain the fifth data. The eighth data set and the ninth data set are obtained through the collected data and the seventh data set. The eighth data set and the ninth data set are input into the fourth data to obtain the sixth data. The sixth data and the collected data are input into the beam occlusion power integration calculation model to obtain the seventh data. The processing of the seventh data to obtain the calculation result of the radar beam occlusion rate includes: E1. According to the collected spatial distribution data of high-spatial-resolution landform types, the data belonging to the mountain and tall building types are screened out, marked as occlusion object data, and the longitude and latitude data of the locations of such occlusion object data are extracted, denoted as [lon_c, lat_c]; E2. According to the collected high-spatial-resolution DEM data and [lon_c, lat_c], the DEM data set of the occlusion object data is screened out, denoted as DEM_c, that is, the seventh data set, where DEM represents digital elevation; E3. [lon_c, lat_c], the altitude, longitude, and latitude of the collected radar antenna are input into the geographic information processing system. Running the system obtains the azimuth range [AZ_c1, AZ_c2] of the distribution of the occlusion object data in the polar coordinate system with the radar antenna as the origin, that is, the fifth data; E4. With the beam width of the collected radar antenna as the azimuth division interval, the space within [AZ_c1, AZ_c2] is divided into several azimuth beam resolution volumes. Taking any one azimuth beam resolution volume, with the spatial resolution of the DEM_c data set as the step size, in the azimuth direction, the azimuth beam resolution volume is further divided into several azimuth ray beam blocks, and the number is denoted as N_c2; E5. Combining the DEM_c data set, the maximum value of the DEM in each azimuth ray beam block is screened out to form the data set DEM_c_max, that is, the eighth data set. At the same time, the longitude and latitude where each data in the DEM_c_max data set appears are recorded to form the data set lon_lat_c_max, that is, the ninth data set; E6. DEM_c_max and lon_lat_c_max are input into the propagation paths of the beam center, upper edge, and lower edge rays of the generated radar beam at any point in space to obtain the occlusion object azimuth range and the maximum occlusion elevation angle relative to the radar beam center line in each azimuth ray beam block, denoted as [AZ_c3, AZ_c4] and ELE_c_max respectively, that is, the sixth data; E7. [AZ_c3, AZ_c4], ELE_c_max, and the collected radar transmit power and antenna gain are input into the beam occlusion power integration calculation model. Running the model obtains the power occluded by the occlusion object in each azimuth ray beam block, denoted as Power_ci, which is the seventh data, where ci = 1, 2,..., N_c2; E8. These N_c2 power values are superimposed to obtain the sum of the radar transmission powers blocked by the occluding objects within the azimuth beam resolution volume, denoted as Power_C. Divide Power_C by the collected radar transmission power to obtain the calculation result of the radar beam occlusion rate in the azimuth beam resolution volume.

Images