A Method for Constructing High-Precision Three-Dimensional Echo Intensity Fields in Mountainous Areas Based on Phased Array Radar

Abstract

This invention relates to a method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar, belonging to the field of radio. First, by calculating the local density estimate of spatial points and combining it with a sliding window and range variance to filter dense neighborhoods, the filtering capability for recovering sparse data is enhanced while preserving structural edge details. Second, a radar echo removal model based on terrain occlusion perception is constructed; by simulating the intersection of the beam propagation path and the three-dimensional surface of the terrain, pseudo-echoes on the occluded path are accurately identified and removed. Third, an interpolation strategy matching method incorporating terrain complexity indices is proposed. Based on features such as elevation gradient, terrain standard deviation, and local range, a clustering algorithm is used to determine typical terrain categories, and these categories are mapped one-to-one with the interpolation strategy, constructing a high-precision coordinate transformation and three-dimensional mesh mapping chain. This, combined with DEM terrain data, constructs an effective radar observation space, improving the spatial constraints and terrain adaptability of three-dimensional modeling.

Description

Technical Field

[0001] This invention relates to the field of radio, and more particularly to a method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar. Background Technology

[0002] In mountainous and complex terrain areas, while phased array radars possess advantages such as rapid pointing and high temporal resolution, the generation of their three-dimensional echo intensity fields still faces numerous challenges. First, terrain obstruction and beam jamming are extremely common. Mountains and valleys hinder radar beam propagation, particularly affecting low-elevation beams, resulting in undetectable "blind spots" in the radar observation space. Furthermore, even if some beams are not completely blocked, power attenuation due to reflection or scattering from mountains can cause discontinuities and non-physical attenuation in the echo signal, affecting the quantitative estimation of echo intensity. Second, complex terrain disturbs the radar beam propagation path. Considering the Earth's curvature and electromagnetic wave refraction effects, terrain undulations further lead to nonlinear changes in the beam trajectory, significantly reducing the accuracy of spatial projection and registration based on simple geometric models, severely impacting the spatial consistency of the three-dimensional modeling results.

[0003] The complex terrain and diverse surface reflections in mountainous regions enhance clutter and reduce the signal-to-noise ratio. High-reflectivity targets such as buildings, mountains, and forests easily generate strong clutter and non-meteorological echoes, interfering with the identification of accurate meteorological signals and increasing the difficulty of data quality control and filtering. Furthermore, due to terrain obstruction and observational limitations, radar often fails to acquire sufficient echo observations in key low-level areas, resulting in uneven spatial sampling point distribution and significant data sparsity and discontinuity. Traditional spatial interpolation methods perform poorly in such sparse data fields, easily introducing interpolation errors, leading to overly smoothed reconstructions, blurred edges, and even structural breaks, thus limiting the ability to reconstruct strong convective structures and the accuracy of precipitation estimation.

[0004] To address the above issues, the mainstream methods for generating 3D radar echo intensity fields currently include the CAPPI (Constant Altitude Plan Position Indicator) method, vertical profile inference methods (such as VPR), and spatial interpolation methods based on 3D meshes. Among these, the CAPPI method interpolates PPI data at different elevation angles to a fixed height, constructs horizontal slices, and then stacks them to generate a 3D field. It boasts advantages such as simplicity and high computational efficiency, making it suitable for real-time applications in operational systems. However, it ignores the actual beam propagation path, resulting in severe low-level blind spots, and is almost ineffective, especially in terrain-obstructed areas. Vertical profile inference methods extrapolate near-surface data in obstructed areas by constructing a regional average vertical profile model, partially filling in the gaps. However, because it assumes a uniform vertical structure across the region, it cannot reconstruct the true structure of the local convection system, and errors are easily amplified. Spatial interpolation construction methods project radar data at various elevation angles onto a three-dimensional grid and use interpolation techniques such as IDW, Kriging, or cubic splines. This method can preserve structural details well, but the interpolation strategy is fixed and lacks adaptability to terrain features. Furthermore, the interpolation error is significant in sparse areas, making it difficult to combine occlusion information for quality judgment. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for constructing a high-precision three-dimensional echo intensity field in multi-mountain areas based on phased array radar, thus solving the deficiencies of the prior art.

[0006] The objective of this invention is achieved through the following technical solution: a method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar, the method comprising:

[0007] S1. Collect phased array radar echo intensity dataset, radar operation status data and high spatial resolution digital elevation data (DEM), and input the collected data into the coordinate system transformation model to obtain radar echo intensity geographic dataset;

[0008] S2. Input the radar echo intensity geographic dataset and the data collected in S1 into the spatial distance calculation model, obtain the spatial distance set through calculation and filtering, and obtain the spatial local density dataset by filtering and calculating the spatial distance set.

[0009] S3. Process a spatial point to be filtered in the radar echo intensity geographic dataset according to the spatial distance relationship, and input the processing result and the spatial location information of the point to be filtered into the spatial distance calculation model. Calculate the filtered dataset by using the sparse data spatial distance weight calculation formula and the sparse data spatial filtering formula.

[0010] S4. Generate a three-dimensional mesh and combine it with the data collected in S1 to generate an effective three-dimensional spatial target mesh for radar echoes. Input the data collected in S1 into the beam unit direction vector construction model. Input the results and the Cartesian coordinates of the radar station obtained in S1 into the beam ray propagation model. Combine the data obtained in S3 to calculate and obtain the spatial information dataset.

[0011] S5. Spatial matching is performed between the filtered dataset and the spatial information dataset to form a spatial matching dataset, and then the dataset is filtered to obtain a high-quality radar echo intensity dataset. The data collected in S1 is input into the elevation gradient calculation model, and the results are input into the standard deviation calculation model along with the data collected in S1. The results are then processed to finally obtain the GRAD dataset, STD dataset, and EXT dataset.

[0012] S6. Merge the three datasets obtained in S5 into a single feature vector and input it into the clustering model. Process and calculate the output of the clustering model to construct a spatial interpolation strategy. Calculate and process the three datasets obtained in S5 to obtain a terrain classification map. Process the effective three-dimensional spatial target grid of radar echoes in conjunction with a high-quality radar echo intensity dataset and input it into the spatial distance calculation model. Filter the output results and determine the interpolation method for the interpolation points by combining the terrain classification map and the spatial interpolation strategy. Finally, generate a three-dimensional radar echo intensity field dataset based on the spatial interpolation model.

[0013] The phased array radar echo intensity dataset includes echo intensity datasets obtained under PPI scanning modes with different antenna elevation angles El, P_Echo_radial=(Z_Echo,R_Echo,Az_Echo,El_Echo), Z_Echo,R_Echo,where Az_Echo,El_Echo are echo intensity, echo range, echo azimuth and antenna elevation angle, respectively;

[0014] The radar operating status data includes the spatial location information of the radar station as P_radar = (Lon_radar, Lat_radar, H_radar), where Lon_radar, Lat_radar and H_radar are the longitude, latitude and altitude of the radar station and the antenna feed, respectively, as well as the radar's radial range resolution delt_R and beamwidth delt_sita.

[0015] The process of inputting the collected data into the coordinate system transformation model to obtain the radar echo intensity geographic dataset includes:

[0016] A1. Input the collected P_radar = (Lon_radar, Lat_radar, H_radar) into the geographic coordinate system to Cartesian coordinate system model, run the model, and output the Cartesian coordinates of the radar station (x_radar, y_radar, H_radar).

[0017] A2. Input the Cartesian coordinates (x_radar, y_radar, H_radar) of the radar site and the collected P_Echo_radial = (Z_Echo, R_Echo, Az_Echo, El_Echo) into the spherical coordinate to Cartesian coordinate conversion model, run the model, and output the Cartesian coordinates (Z_Echo, x_Echo, y_Echo, z_Echo) of the radar echo intensity.

[0018] A3. Input the Cartesian coordinates (Z_Echo, x_Echo, y_Echo, z_Echo) of the radar echo intensity into the inverse projection transformation model from Cartesian coordinates to latitude and longitude, run the model, and output the geographic dataset of radar echo intensity P_Echo=(Z_Echo, Lon_Echo, Lat_Echo, Hig_Echo), where Lon_Echo, Lat_Echo, and Hig_Echo represent longitude, latitude, and altitude, respectively.

[0019] S2 specifically includes the following:

[0020] S201. Based on the calculated P_Echo = (Z_Echo, Lon_Echo, Lat_Echo, Hig_Echo), set any one of the echo points P. O The information is (Z_Echo_O, Lon_Echo_O, Lat_Echo_O, Hig_Echo_O), and the collected radar station spatial information P_radar = (Lon_radar, Lat_radar, H_radar), and P O The information (Z_Echo_O, Lon_Echo_O, Lat_Echo_O, Hig_Echo_O) is input together into the spatial distance calculation model, and the model output is P. O Radial distance R from the point to the radar station O Z_Echo_O, Lon_Echo_O, Lat_Echo_O, and Hig_Echo_O represent echo points P, respectively. O The echo intensity, longitude, latitude, and altitude;

[0021] S202, according to R OAnd collect delt_R and delt_sita, calculate the maximum radius r_max for searching the neighboring spatial domain = m × [delt_R × delt_R + (R O ×delt_sita) 2 ] 0.5 m is a set multiple, with P O Let P be the center of the sphere. O Let P be the N1 neighboring points of a given point. i = (Z_Echo_i, Lon_Echo_i, Lat_Echo_i, Hig_Echo_i), i=1,2,3,…,N1, set P O and P i The information is input together into the spatial distance calculation model, and the output of the model is N1 P. i Distance P O The spatial distance dataset, denoted as d Oi Where Z_Echo_i, Lon_Echo_i, Lat_Echo_i, and Hig_Echo_i represent echo points P, respectively. i The echo intensity, longitude, latitude, and altitude;

[0022] S203, Filter out d Oi Collect all neighboring points less than r_max and count their number, denoted as N2, where N2 ≤ N1. Combine the spatial distances of these N2 points into P. O The set of neighboring spatial distances, denoted as d Oj , j=1,2,3,…,N2, d Oj Sort the dataset from smallest to largest to generate dataset d. Oj_sort Set the sliding window size to N_w, and let the sliding window traverse all possible d. Oj_sort Sub-intervals are defined, and the variance of each window is calculated and denoted as Var_d. Oj_sort ;

[0023] S204, retrieve Var_d Oj_sort The minimum value is located at the center point d of the window. _dense , will d _dense As d Oj_sort The distance P of the densest region in the dataset O The representative value of spatial distance, denoted by P. O With the center of the sphere as d _dense Let d be the radius of the sphere, then select the appropriate values. Oj Less than d _dense Collect all neighboring points and count their number, denoted as N3, N3≤N2, and generate P. O The set of nearest neighbors is denoted as P. kk=1,2,3,…,N3, and the distance of these N3 points from P O The set of spatial distances consisting of spatial distances is denoted as d. Ok ;

[0024] S205, Based on the obtained spatial distance set d Ok Filter out the maximum value in the dataset, denoted as R. Omax , with P O With R as the center of the sphere Omax Let P be the radius of the sphere. Calculate the radius of the sphere using the formula for calculating the radius of a sphere. O The volume V of the spatial region defined by its N3 nearest neighbors O P is calculated based on the spatial local density calculation formula. O Density (P) of the spatial domain formed by N3 nearest neighbors O );

[0025] S206, Set P O Update the dataset to include all data points in the generated P_Echo, and repeat steps S201-S205 to obtain a spatial local density dataset Density(P_Echo) centered on a sphere with each point in the dataset P_Echo as the center. m ), m=1,2,3,…,N E N E This represents the total number of points in the dataset P_Echo.

[0026] S3 specifically includes the following:

[0027] S301. For a spatial point in the generated P_Echo data to be filtered, let this point be Z. f_b Based on the spatial distance, extract the distance from Z. f_b The N4 points that are closest in spatial distance to each other are denoted as Z. n n=1,2,3,…,N4, and according to Z n Spatial location from the spatial local density dataset Density(P) m Extract the spatial local density of these N4 points from the dataset to form the dataset Density(Z). n );

[0028] S302, Transfer the dataset Density(Z) n Sort the data sequences from smallest to largest, defining the data with percentages Per1 before sorting and Per2 after sorting as low-density and high-density regions respectively. Calculate the average value of each low-density and high-density region, denoted as Density_Z. L and Density_Z HBased on the formula for calculating the local density standard deviation parameter, Density_Z is calculated respectively. L and Density_Z H The corresponding local density standard deviation parameter is denoted as and ;

[0029] S303, will and Combined into a local density standard deviation parameter diffusion range sequence, denoted as and will Divide into N4 equal parts, and generate each Z n The local density standard deviation parameter corresponding to the point is denoted as . ;

[0030] S304, Z f_b and Z n The spatial location information is input into the spatial distance calculation model, the model is run, and the output is N4 Z-axis distances. n Distance Z f_b The spatial distance dataset, denoted as d On , then d On With the calculated The values ​​are input together into the sparse data space distance weight calculation formula, and each Z is calculated separately. n The weight value of the point is denoted as Wn. The point Z to be filtered is calculated using the sparse data space filtering formula. f_b The filtered value Z f_a ;

[0031] S305, Z in S301-S304 f_b Update each point in the P_Echo data to obtain the filtered dataset P_Echo, denoted as P_Echo_filter.

[0032] S4 specifically includes the following:

[0033] S401. Set the spatial resolution of the horizontal grid of latitude and longitude in two-dimensional space to lon_lat_reso, set the spatial resolution of the vertical height in one-dimensional space to Hig_reso, and generate a three-dimensional grid TAR_Grid with a spatial resolution of lon_lat_reso×lon_lat_reso×Hig_reso.

[0034] S402. The collected high spatial resolution digital elevation data (DEM) is used to generate a three-dimensional terrain surface continuity function, denoted as DEM_F, by using a three-dimensional spatial spline interpolation method. The latitude and longitude horizontal grid in the two-dimensional space of TAR_Grid is input into the three-dimensional terrain surface continuity function DEM_F to obtain the ground height corresponding to each grid point in the latitude and longitude horizontal grid in the two-dimensional space of TAR_Grid. Combining the height information corresponding to each grid point in the height dimension of TAR_Grid, all grid points in the vertical direction of each latitude and longitude horizontal grid point in TAR_Grid that are lower than the ground height are removed to generate the radar echo effective three-dimensional spatial target grid TAR_Grid_1.

[0035] S403. Based on the obtained radar echo intensity spatial location information P_Echo_radial=(Z_Echo,R_Echo,Az_Echo,El_Echo), input it into the beam unit direction vector construction model, run the model, and generate the beam unit direction vector d=[cos(El_Echo)*sin(Az_Echo), cos(El_Echo)*cos(Az_Echo), sin(El_Echo)];

[0036] S404. Input the Cartesian coordinates (x_radar, y_radar, H_radar) of the acquired radar station and the beam unit direction vector d into the beam ray propagation model that considers the curvature of the earth and electromagnetic refraction. Run the model to generate a radar beam ray trajectory point set, denoted as Path_Beam. For any radar beam ray, calculate the height difference between the trajectory point set Path_Beam and the generated three-dimensional terrain surface continuity function DEM_F at the latitude and longitude grid points corresponding to the beam propagation path. Based on the principle of minimizing the height difference, determine the spatial information of the first intersection point between the beam and the terrain surface.

[0037] S405. Apply the method of step S404 to all radar beam paths to obtain a spatial information dataset of all radar beam paths intersecting with the terrain surface, denoted as P_proj.

[0038] S5 includes the following:

[0039] S501. Based on the obtained P_Echo_filter and P_proj, perform spatial matching according to the principle of shortest spatial distance to form a spatial matching dataset, denoted as Match. Based on the height information of the matching points in the Match dataset, check whether the echo points on each radar ray path are located before the terrain intersection of the beam ray. Remove all echo points on the radar beam ray with a height greater than the height of the intersection point, and generate a high-quality radar echo intensity dataset P_Echo_filter_clear after occlusion judgment.

[0040] S502. Input the collected high-resolution DEM dataset into the elevation gradient calculation model, run the model, and the model output is the elevation gradient dataset GRAD(Lon,Lat) for the latitude and longitude point (Lon,Lat). Set the spatial sliding window size to N_slid×N_slid, and input it together with the collected high-resolution DEM dataset into the standard deviation calculation model. Run the model, and the model output is the elevation standard deviation dataset STD(Lon,Lat) for each sliding window at the latitude and longitude point (Lon,Lat).

[0041] S503. Find the maximum value DEMmax and minimum value DEMmin of the DEM in the spatial sliding window, generate the DEM range DEMmax-DEMmin in the spatial sliding window, and use this range value as the local range value EXT(Lon,Lat) of the center point (Lon,Lat) of the spatial sliding window.

[0042] S504. Apply the steps of S501-S503 to all spatial points in the DEM dataset to obtain the GRAD dataset, STD dataset, and EXT dataset.

[0043] The process of merging the three datasets obtained in S5 into a single feature vector and inputting it into the clustering model, and then processing and calculating the output of the clustering model to construct a spatial interpolation strategy, specifically includes the following:

[0044] B1. Merge the acquired GRAD, STD, and EXT datasets into a single feature vector Fe = [GRAD, STD, EXT]. Input the feature vector Fe into the K-means clustering model and set the number of terrain categories to kind. Run the model. The model outputs the elevation gradient GRADc_p, elevation standard deviation STDc_p, and local extreme value EXTc_p corresponding to each cluster center, p = 1, 2, 3, ..., kind, as well as the maximum elevation gradient GRAD_max, the maximum elevation standard deviation STD_max, and the maximum local extreme value EXT_max across all clusters.

[0045] B2. Based on the eigenvalue clustering weight calculation formula, calculate the weight coefficients of the three features, namely elevation gradient, elevation standard deviation, and local extrema, in each cluster and normalize them, denoted as W1_p, W2_p, and W3_p respectively.

[0046] B3. Calculate the terrain complexity index TRI_p for each typical terrain type according to the terrain complexity index calculation formula; based on the calculated TRI_p, construct d spatial interpolation strategies that map one-to-one with different typical terrain regions: Inter_str = [bilinear interpolation, inverse distance weighted interpolation, least squares interpolation, ..., cubic spline interpolation].

[0047] The process of calculating and processing the three datasets obtained from S5 to obtain the terrain classification map specifically includes the following:

[0048] C1. Based on S5, obtain the GRAD, STD, and EXT data at any point Q in the DEM dataset. Divide the GRAD, STD, and EXT data by the maximum value in the corresponding dataset to obtain the normalized GRAD_q, STD_q, and EXT_q data.

[0049] C2. Substitute the normalized GRAD_q data, STD_q data, and EXT_q data into the terrain complexity index calculation formula to generate the terrain complexity index value TRI_Q for this point. Calculate the absolute difference between TRI_Q and the terrain complexity index TRI_p of kind typical terrain categories. According to the principle of minimum absolute deviation, classify point Q into the terrain category closest to its terrain complexity index, that is, the terrain type to which it belongs is the terrain category corresponding to the smallest deviation of TRI_q.

[0050] C3. Apply steps C1 and C2 to all spatial points in the DEM to obtain a terrain classification map for each point in the DEM, denoted as Terrain_C.

[0051] The process involves combining the effective 3D spatial target mesh of radar echoes with a high-quality radar echo intensity dataset, inputting it into a spatial distance calculation model, filtering the output results, and determining the interpolation method for the interpolation points by combining terrain classification maps and spatial interpolation strategies. Finally, the 3D radar echo intensity field dataset generated based on the spatial interpolation model specifically includes the following:

[0052] D1. For any interpolation point in the generated effective 3D spatial target grid TAR_Grid_1 of radar echo, the spatial information is (lon_inter, lat_inter, hig_inter), and the corresponding echo intensity is assumed to be Z_inter. Then, the interpolation point P_Echo_inter = (Z_inter, lon_inter, lat_inter, hig_inter). The generated high-quality radar echo intensity dataset P_Echo_filter_clear after occlusion judgment is input together with the spatial information of the interpolation point P_Echo_inter into the spatial distance calculation model. Run the model, and the output is the spatial distance dataset from each spatial point in P_Echo_filter_clear to the interpolation point P_Echo_inter, denoted as Distance_P_inter.

[0053] D2. Filter all spatial points smaller than the maximum radius r_max of the neighboring spatial domain from Distance_P_inter to form the neighboring point dataset of the point to be interpolated, denoted as P_Echo_inter_near. Based on the longitude and latitude of the point to be interpolated, combined with the generated terrain classification map Terrain_C, query the terrain category of the point to be interpolated, and then determine the interpolation method of the point based on the obtained spatial interpolation strategy Inter_str.

[0054] D3. Input the neighboring point dataset P_Echo_inter_near of the interpolation point P_Echo_inter and the interpolation point P_Echo_inter together into the spatial interpolation model, combine the determined interpolation method, run the model, and generate the echo intensity Z_inter of the interpolation point P_Echo_inter.

[0055] D4. Apply steps D1-D4 to all points in the effective three-dimensional spatial target grid TAR_Grid_1 of the radar echo to generate a high spatial resolution and high precision three-dimensional radar echo intensity field dataset for mountainous areas.

[0056] The present invention has the following advantages:

[0057] 1. Spatially Local Density-Driven Adaptive Filtering Mechanism to Improve Sparse Region Reconstruction Capability: Addressing the data sparsity problem caused by low-level observation blind spots in mountainous areas, a filtering strategy based on spatially neighboring point density estimation is proposed. Through adaptive adjustment of the sliding window and density weights, structural restoration of sparse regions is achieved while avoiding over-smoothing and preserving echo edge details.

[0058] 2. Combining terrain occlusion perception with echo quality control to improve data credibility: A calculation model of the intersection of radar beam propagation path and three-dimensional terrain surface is established, which can accurately identify and eliminate echo points blocked by terrain, avoid false echo interference, and effectively improve the physical consistency and authenticity of the three-dimensional echo field, which is particularly suitable for mountainous environments with drastic terrain undulations.

[0059] 3. Enhance terrain adaptability by integrating interpolation strategy matching with terrain complexity index: A terrain complexity index integrating elevation gradient, terrain standard deviation and range was constructed, and typical terrain categories were divided by clustering algorithm to match the optimal interpolation method for each terrain category. This realizes spatial interpolation strategy matching for terrain perception and effectively improves the interpolation accuracy and robustness in complex terrain areas.

[0060] 4. High-precision multi-stage coordinate transformation to ensure consistency in 3D modeling space: A continuous transformation chain is constructed from polar coordinates to Cartesian coordinates, then to geographic coordinates, and finally to a 3D regular mesh. Combined with beam propagation model and digital elevation data (DEM), the distribution of radar echoes in geographic space is accurately located, ensuring the geometric accuracy of echo projection and the consistency of the 3D mesh. Attached Figure Description

[0061] Figure 1 This is a schematic diagram of the process of the present invention;

[0062] Figure 2 A flowchart illustrating the implementation of the nearest neighbor set and spatial distance set in this invention;

[0063] Figure 3 This is a flowchart illustrating the implementation of generating the spatially filtered phased array radar echo intensity dataset of the present invention.

[0064] Figure 4 This is a flowchart illustrating the implementation of the effective three-dimensional spatial target mesh generation for radar echoes in this invention.

[0065] Figure 5 A flowchart illustrating the implementation process of generating a spatial information dataset of the intersection points of radar ray paths and the Earth's surface in this invention.

[0066] Figure 6 A flowchart illustrating the implementation process of calculating the terrain complexity index and generating the corresponding spatial interpolation strategy for each typical terrain type in this invention.

[0067] Figure 7 This is a flowchart illustrating the implementation process of generating a high spatial resolution and high precision three-dimensional radar echo intensity field dataset for mountainous areas, as described in this invention. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application provided below with reference to the accompanying drawings is not intended to limit the scope of protection of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. The present invention will be further described below with reference to the accompanying drawings.

[0069] like Figure 1 As shown, this invention specifically relates to a method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar, which specifically includes the following:

[0070] Step 1: Collect phased array radar echo intensity dataset and radar operation status data; collect high spatial resolution digital elevation data (DEM);

[0071] Among them, the phased array radar echo intensity dataset mainly refers to the echo intensity dataset P_Echo_radial=(Z_Echo, R_Echo, Az_Echo, El_Echo) obtained under the PPI scanning mode at different antenna elevation angles El, with echo intensity, range, azimuth and antenna elevation information being Z_Echo, R_Echo, Az_Echo and El_Echo respectively;

[0072] The radar operation status data mainly includes: the spatial location information of the radar station as P_radar = (Lon_radar, Lat_radar, H_radar), where Lon_radar, Lat_radar and H_radar are the longitude, latitude and altitude of the radar station and the antenna feed, respectively; the radial range resolution of the radar delt_R = 30 meters and the beamwidth delt_sita = 0.0174 rad (1 degree).

[0073] Step 2: First, input the P_radar = (Lon_radar, Lat_radar, H_radar) collected in Step 1 into the geographic coordinate system to Cartesian coordinate system model, run the model, and output the Cartesian coordinates (x_radar, y_radar, H_radar) of the radar station; Second, combine the Cartesian coordinates (x_radar, y_radar, H_radar) of the radar station with the P_Echo_radial = (Z_Echo, R_Echo, Az_Echo, El_Ec) collected in Step 1. The radar echo intensity (Z_Echo, x_Echo, y_Echo, z_Echo) is input into the Cartesian coordinate conversion model. The model is run, and the output is the Cartesian coordinates of the radar echo intensity (Z_Echo, x_Echo, y_Echo, z_Echo). Finally, the Cartesian coordinates of the radar echo intensity (Z_Echo, x_Echo, y_Echo, z_Echo) are input into the Cartesian coordinate to latitude and longitude inverse projection conversion model. The model is run, and the output is the radar echo intensity geographic dataset P_Echo = (Z_Echo, Lon_Echo, Lat_Echo, Hig_Echo).

[0074] Step 3: As Figure 2 As shown, firstly, based on the calculation in step 2, P_Echo = (Z_Echo, Lon_Echo, Lat_Echo, Hig_Echo), assume any one of the echo points P O The information is (Z_Echo_O, Lon_Echo_O, Lat_Echo_O, Hig_Echo_O). The radar station spatial information P_radar = (Lon_radar, Lat_radar, H_radar) collected in step 1, and P... O The information (Z_Echo_O, Lon_Echo_O, Lat_Echo_O, Hig_Echo_O) is input together into the spatial distance calculation model, and the model output is P. O Radial distance R from the point to the radar station O Secondly, according to R O Following step 1, we collect delt_R and delt_sita, and calculate the maximum radius r_max for searching the neighboring spatial domain: r_max = m × [delt_R × delt_R + (R O ×delt_sita) 2 ] 0.5 m=10 is a set multiple; then, with P O Let P be the center of the sphere. O Let P be a point with N1 = 26 neighboring points. i= (Z_Echo_i, Lon_Echo_i, Lat_Echo_i, Hig_Echo_i) (i=1,2,3,…,N1), put P O and P i The information is input together into the spatial distance calculation model, and the output of the model is N1 P. i Distance P O The spatial distance dataset, denoted as d Oi Then, filter out d. Oi Collect all neighboring points less than r_max and count their number, denoted as N2 (N2<=N1). Combine the spatial distances of these N2 points into P. O The set of neighboring spatial distances, denoted as d Oj (j=1,2,3,…,N2). Let d Oj Sort the dataset from smallest to largest to generate dataset d. Oj_sort Set the sliding window size to N_w, and let the sliding window traverse all possible d. Oj_sort Sub-intervals are defined, and the variance of each window is calculated and denoted as Var_d. Oj_sort Get Var_d Oj_sort The minimum value is located at the center point d of the window. _dense , will d _dense As d Oj_sort The distance P of the densest region in the dataset O The representative value of spatial distance; finally, P O With the center of the sphere as d _dense Let d be the radius of the sphere, then select the appropriate values. Oj Less than d _dense Collect all neighboring points and count their number, denoted as N3 (N3 <= N2), and generate P. O The set of nearest neighbors is denoted as P. k (k=1,2,3,…,N3), and the distance of these N3 points from P O The set of spatial distances consisting of spatial distances is denoted as d. Ok (k=1,2,3,…,N3).

[0075] Step 4: First, based on the spatial distance set d obtained in Step 3... Ok Filter out the maximum value in the dataset, denoted as R. Omax Secondly, with P O With R as the center of the sphere Omax Let P be the radius of the sphere. Calculate the radius of the sphere using the formula for calculating the radius of a sphere. O The volume V of the spatial region defined by its N3 nearest neighbors O Then, P was calculated according to the spatial local density calculation formula. ODensity (P) of the spatial domain formed by N3 nearest neighbors O ).

[0076] The formula for calculating a sphere is: V O =4 / 3× ×R Omax 3 The formula for calculating local spatial density is: Density(P) O )=N3 / V O .

[0077] Step 5: As Figure 3 As shown, firstly, the P assumed in step 3... O Update all data points in P_Echo generated in step 2, and repeat the processing steps 3 and 4 to obtain the spatial local density dataset Density(P) centered on each point in the dataset P_Echo. m (m=1,2,3,…,N) E ), N E Let Z be the total number of points in the dataset P_Echo; secondly, for a specific spatial point in the P_Echo data generated in step 2 that needs to be filtered, let this point be denoted as Z. f_b Based on the spatial distance, extract the distance from Z. f_b The N4=14 points that are closest in spatial distance are denoted as Z. n (n=1,2,3,…,N4), and according to Z n Spatial location from dataset Density(P) m Extract the spatial local density of these N4 points from the dataset to form the dataset Density(Z). n Then, the dataset Density(Z) n Sort the data in ascending order. Define the data sequences with Per1=20% before sorting and Per2=20% after sorting as low-density and high-density regions, respectively. Calculate the average value of the low-density and high-density regions, denoted as Density_Z. L and Density_Z H Then, based on the formula for calculating the local density standard deviation parameter, Density_Z was calculated. L and Density_Z H The corresponding local density standard deviation parameter is denoted as and .Will and Combined into a local density standard deviation parameter diffusion range sequence, denoted as and will Divide into N4 equal parts, and generate each Zn The local density standard deviation parameter corresponding to point (n=1,2,3,…,N4) is denoted as After that, Z f_b and Z n The spatial location information is input into the spatial distance calculation model. The model is run, and the output is N4 Z-axis distances. n Distance Z f_b The spatial distance dataset, denoted as d On After that, d On and the calculated The data is input together into the formula for calculating the spatial distance weights of sparse data, and each Z value is calculated separately. n The weight value of the point is denoted as Wn; finally, according to the sparse data space filtering formula, the Z of the point to be filtered can be calculated. f_b The filtered value Z f_a In step 5, Z... f_b By updating each point in the P_Echo data, we can obtain the filtered dataset P_Echo, denoted as P_Echo_filter.

[0078] The formula for calculating the local density standard deviation parameter is as follows: , ,in This is the initial setting value when the density is 1;

[0079] The formula for calculating the spatial distance weight of sparse data is: ;

[0080] The formula for spatial filtering of sparse data is: ;

[0081] Step 6: As Figure 4As shown, firstly, the spatial resolution of the two-dimensional latitude and longitude horizontal grid is set to lon_lat_reso = 0.0001 degrees, and the spatial resolution of the one-dimensional vertical height is set to Hig_reso = 10 meters, generating a three-dimensional grid TAR_Grid with a spatial resolution of lon_lat_reso × lon_lat_reso × Hig_reso. Secondly, the high spatial resolution digital elevation data (DEM) collected in step 1 is used to generate a three-dimensional terrain surface continuity function, denoted as DEM_F, using a three-dimensional spatial spline interpolation method. Then, the two-dimensional latitude and longitude horizontal grid in TAR_Grid is input into the three-dimensional terrain surface continuity function DEM_F to obtain the ground height corresponding to each grid point in the two-dimensional latitude and longitude horizontal grid of TAR_Grid. Combining the height information corresponding to each grid point in the height dimension of TAR_Grid, all grid points in the vertical direction of each latitude and longitude horizontal grid point in TAR_Grid that are lower than the ground height are removed, generating the radar echo effective three-dimensional spatial target grid TAR_Grid_1.

[0082] Step 7: As Figure 5 As shown, firstly, the spatial location information of the radar echo intensity P_Echo_radial = (Z_Echo, R_Echo, Az_Echo, El_Echo) obtained in step 1 is input into the beam unit direction vector construction model. The model is run to generate the beam unit direction vector d = [cos(El_Echo)*sin(Az_Echo), cos(El_Echo)*cos(Az_Echo), sin(El_Echo)]. Secondly, the Cartesian coordinates (x_rada) of the radar site obtained in step 2 are used... The radar beam propagation model, which considers the Earth's curvature and electromagnetic refraction, is input along with the radar beam unit direction vector d (r, y_radar, H_radar). The model is run to generate a radar beam trajectory point set, denoted as Path_Beam. Then, for any radar beam, the height difference between the trajectory point set Path_Beam and the 3D terrain surface continuity function DEM_F generated in step 6 at the latitude and longitude grid points corresponding to the beam propagation path is calculated. The spatial information of the first intersection point between the beam and the terrain surface is determined based on minimizing the height difference. This method is applied to all radar beam paths to obtain a dataset of spatial information of all radar beam paths intersecting with the terrain surface, denoted as P_proj.

[0083] Step 8: First, based on P_Echo_filter obtained in Step 5 and P_proj obtained in Step 7, perform spatial matching according to the principle of shortest spatial distance to form a spatial matching dataset, denoted as Match; Second, based on the height information of the matching points in the Match dataset, check one by one whether the echo points on each radar ray path are located before the terrain intersection of the beam ray, remove all echo points on the radar beam ray whose height is greater than the height of the intersection point, and generate a high-quality radar echo intensity dataset P_Echo_filter_clear after occlusion judgment.

[0084] Step 9: First, input the high-resolution DEM dataset collected in Step 1 into the elevation gradient calculation model, run the model, and the model output is the elevation gradient dataset GRAD(Lon, Lat) for the latitude and longitude point (Lon, Lat). Second, set the spatial sliding window size to N_slid×N_slid=3×3, and input it together with the high-resolution DEM dataset collected in Step 1 into the standard deviation calculation model, run the model, and the model output is the elevation standard deviation dataset STD(Lon, Lat) for each sliding window at the latitude and longitude point (Lon, Lat). Find the maximum value DEMmax and minimum value DEMmin of the DEM in the spatial sliding window, generate the DEM range DEMmax-DEMmin in the spatial sliding window, and use this range value as the local range value EXT(Lon, Lat) of the center point (Lon, Lat) of the spatial sliding window. Apply the above steps to all spatial points in the DEM dataset to obtain the GRAD dataset, STD dataset, and EXT dataset.

[0085] Step 10: As Figure 6As shown, firstly, the GRAD, STD, and EXT datasets obtained in step 9 are merged into a single feature vector Fe = [GRAD, STD, EXT]; secondly, the feature vector Fe is input into the K-means algorithm. In the clustering model, the number of terrain categories is set to kind=3 (flat areas, hilly areas, mountainous areas). The model is run, and the output is the elevation gradient GRADc_p, elevation standard deviation STDc_p, and local extreme value EXTc_p (p=1,2,3) corresponding to each cluster center, as well as the maximum elevation gradient GRAD_max, the maximum elevation standard deviation STD_max, and the maximum local extreme value EXT_max in all clusters. Then, according to the cluster weight calculation formula of feature value, the weight coefficients of the three features of elevation gradient, elevation standard deviation, and local extreme value in each cluster are calculated and normalized, and denoted as W1_p, W2_p, and W3_p respectively. Finally, according to the terrain complexity index calculation formula, the terrain complexity index TRI_p of each typical terrain category is calculated. Based on the calculated TRI_p, three spatial interpolation strategies Inter_str=[bilinear interpolation, inverse distance weighted interpolation, cubic spline interpolation] are constructed to map one-to-one with different typical terrain areas.

[0086] The formulas for calculating the cluster weights of the eigenvalues ​​are: W1_p=GRADc_p / GRAD_max, W2_p=STDc_p / STD_max, W3_p=EXTc_p / EXT_max;

[0087] The formula for calculating the terrain complexity index is: TRI_p = W1_p * GRAD + W2_p * STD + W3_p * EXT_max;

[0088] Step 11: As Figure 7 As shown, firstly, according to step 9, the GRAD, STD, and EXT data at any point Q in the DEM dataset are obtained. The GRAD, STD, and EXT data are then divided by the maximum value in the corresponding dataset to obtain normalized GRAD_q, STD_q, and EXT_q data. Secondly, the normalized GRAD_q, STD_q, and EXT_q data are substituted into the terrain complexity index calculation formula in step 10 to generate the terrain complexity index value TRI_Q for that point. Then, the absolute difference between TRI_Q and the terrain complexity index TRI_p for the three typical terrain categories (kind=3) obtained in step 10 is calculated. Based on the principle of minimum absolute deviation, point Q is assigned to the terrain category closest to its terrain complexity index, i.e., the terrain type it belongs to is the terrain category corresponding to the smallest deviation of TRI_q. Finally, this scheme is applied to all spatial points in the DEM to obtain the terrain classification map for each point in the DEM, denoted as Terrain_C.

[0089] Step 12: First, for any interpolation point in the effective 3D spatial target grid TAR_Grid_1 generated in Step 6, the spatial information is (lon_inter, lat_inter, hig_inter), and the corresponding echo intensity is assumed to be Z_inter. Then, the interpolation point P_Echo_inter = (Z_inter, lon_inter, lat_inter, hig_inter). Second, input the high-quality radar echo intensity dataset P_Echo_filter_clear after occlusion judgment generated in Step 8 and the spatial information of the interpolation point P_Echo_inter together into the spatial distance calculation model, run the model, and output the spatial distance dataset from each spatial point in P_Echo_filter_clear to the interpolation point P_Echo_inter, denoted as Distance_P_int. Next, all spatial points smaller than the maximum radius r_max found in the neighboring spatial domain set in step 3 are selected from Distance_P_inter to form the neighboring point dataset of the interpolation point P_Echo_inter, denoted as P_Echo_inter_near. Then, based on the longitude and latitude of the interpolation point P_Echo_inter and the terrain classification map Terrain_C generated in step 11, the terrain category of the interpolation point is queried, and then the interpolation method for the point is determined according to the spatial interpolation strategy Inter_str obtained in step 10. Finally, the neighboring point dataset P_Echo_inter_near and the interpolation point P_Echo_inter are input into the spatial interpolation model, and the model is run using the interpolation method determined in this step to generate the echo intensity Z_inter of the interpolation point P_Echo_inter. Applying the above methods in this step to all points in the effective three-dimensional spatial target grid TAR_Grid_1 of radar echoes can generate a high spatial resolution and high precision three-dimensional radar echo intensity field dataset for mountainous areas.

[0090] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and improvements, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar, characterized by: The method includes: S1. Collect phased array radar echo intensity dataset, radar operation status data and high spatial resolution digital elevation data (DEM), and input the collected data into the coordinate system transformation model to obtain radar echo intensity geographic dataset; S2. Input the radar echo intensity geographic dataset and the data collected in S1 into the spatial distance calculation model, obtain the spatial distance set through calculation and filtering, and obtain the spatial local density dataset by filtering and calculating the spatial distance set. S3. Process a spatial point to be filtered in the radar echo intensity geographic dataset according to the spatial distance relationship, and input the processing result and the spatial location information of the point to be filtered into the spatial distance calculation model. Calculate the filtered dataset by using the sparse data spatial distance weight calculation formula and the sparse data spatial filtering formula. S4. Generate a three-dimensional mesh and combine it with the data collected in S1 to generate an effective three-dimensional spatial target mesh for radar echoes. Input the data collected in S1 into the beam unit direction vector construction model. Input the results and the Cartesian coordinates of the radar station obtained in S1 into the beam ray propagation model. Combine the data obtained in S3 to calculate and obtain the spatial information dataset. S5. Spatial matching is performed between the filtered dataset and the spatial information dataset to form a spatial matching dataset, and then the dataset is filtered to obtain a high-quality radar echo intensity dataset. The data collected in S1 is input into the elevation gradient calculation model, and the results are input into the standard deviation calculation model along with the data collected in S1. The results are then processed to finally obtain the GRAD dataset, STD dataset, and EXT dataset. S6. Merge the three datasets obtained in S5 into a single feature vector and input it into the clustering model. Process and calculate the output of the clustering model to construct a spatial interpolation strategy. Calculate and process the three datasets obtained in S5 to obtain a terrain classification map. Process the effective three-dimensional spatial target grid of radar echoes in conjunction with a high-quality radar echo intensity dataset and input it into the spatial distance calculation model. Filter the output results and determine the interpolation method for the interpolation points by combining the terrain classification map and the spatial interpolation strategy. Finally, generate a three-dimensional radar echo intensity field dataset based on the spatial interpolation model.

2. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 1, characterized in that: The phased array radar echo intensity dataset includes echo intensity datasets obtained under PPI scanning modes with different antenna elevation angles El, P_Echo_radial=(Z_Echo, R_Echo, Az_Echo, El_Echo), where Z_Echo, R_Echo, Az_Echo, and El_Echo are echo intensity, echo range, echo azimuth, and antenna elevation angle, respectively. The radar operating status data includes the spatial location information of the radar station as P_radar = (Lon_radar, Lat_radar, H_radar), where Lon_radar, Lat_radar and H_radar are the longitude, latitude and altitude of the radar station and the antenna feed, respectively, as well as the radar's radial range resolution delt_R and beamwidth delt_sita.

3. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 2, characterized in that: The process of inputting the collected data into the coordinate system transformation model to obtain the radar echo intensity geographic dataset includes: A1. Input the collected P_radar = (Lon_radar, Lat_radar, H_radar) into the geographic coordinate system to Cartesian coordinate system model, run the model, and output the Cartesian coordinates of the radar station (x_radar, y_radar, H_radar). A2. Input the Cartesian coordinates (x_radar, y_radar, H_radar) of the radar site and the collected P_Echo_radial = (Z_Echo, R_Echo, Az_Echo, El_Echo) into the spherical coordinate to Cartesian coordinate conversion model, run the model, and output the Cartesian coordinates (Z_Echo, x_Echo, y_Echo, z_Echo) of the radar echo intensity. A3. Input the Cartesian coordinates (Z_Echo, x_Echo, y_Echo, z_Echo) of the radar echo intensity into the inverse projection transformation model from Cartesian coordinates to latitude and longitude, run the model, and output the geographic dataset of radar echo intensity P_Echo=(Z_Echo, Lon_Echo, Lat_Echo, Hig_Echo), where Lon_Echo, Lat_Echo, and Hig_Echo represent longitude, latitude, and altitude, respectively.

4. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 3, characterized in that: S2 specifically includes the following: S201. Based on the calculated P_Echo = (Z_Echo, Lon_Echo, Lat_Echo, Hig_Echo), set any one of the echo points P. O The information is (Z_Echo_O, Lon_Echo_O, Lat_Echo_O, Hig_Echo_O), and the collected radar station spatial information P_radar = (Lon_radar, Lat_radar, H_radar), and P O The information (Z_Echo_O, Lon_Echo_O, Lat_Echo_O, Hig_Echo_O) is input together into the spatial distance calculation model, and the model output is P. O Radial distance R from the point to the radar station O Z_Echo_O, Lon_Echo_O, Lat_Echo_O, and Hig_Echo_O represent echo points P, respectively. O The echo intensity, longitude, latitude, and altitude; S202, according to R O And collect delt_R and delt_sita, calculate the maximum radius r_max for searching the neighboring spatial domain = m × [delt_R × delt_R + (R O ×delt_sita) 2 ] 0.5 m is a set multiple, with P O Let P be the center of the sphere. O Let P be the N1 neighboring points of a given point. i = (Z_Echo_i, Lon_Echo_i, Lat_Echo_i, Hig_Echo_i), i=1,2,3,…,N1, set P O and P i The information is input together into the spatial distance calculation model, and the output of the model is N1 P. i Distance P O The spatial distance dataset, denoted as d Oi Where Z_Echo_i, Lon_Echo_i, Lat_Echo_i, and Hig_Echo_i represent echo points P, respectively. i The echo intensity, longitude, latitude, and altitude; S203, Filter out d Oi Collect all neighboring points less than r_max and count their number, denoted as N2, where N2 ≤ N1. Combine the spatial distances of these N2 points into P. O The set of neighboring spatial distances, denoted as d Oj , j=1,2,3,…,N2, d Oj Sort the dataset from smallest to largest to generate dataset d. Oj_sort Set the sliding window size to N_w, and let the sliding window traverse all possible d. Oj_sort Sub-intervals are defined, and the variance of each window is calculated and denoted as Var_d. Oj_sort ; S204, retrieve Var_d Oj_sort The minimum value is located at the center point d of the window. _dense , will d _dense As d Oj_sort The distance P of the densest region in the dataset O The representative value of spatial distance, denoted by P. O With the center of the sphere as d _dense Let d be the radius of the sphere, then select the appropriate values. Oj Less than d _dense Collect all neighboring points and count their number, denoted as N3, N3≤N2, and generate P. O The set of nearest neighbors is denoted as P. k k=1,2,3,…,N3, and the distance of these N3 points from P O The set of spatial distances consisting of spatial distances is denoted as d. Ok ; S205, Based on the obtained spatial distance set d Ok Filter out the maximum value in the dataset, denoted as R. Omax , with P O With R as the center of the sphere Omax Let P be the radius of the sphere. Calculate the radius of the sphere using the formula for calculating the radius of a sphere. O The volume V of the spatial region defined by its N3 nearest neighbors O P is calculated based on the spatial local density calculation formula. O Density (P) of the spatial domain formed by N3 nearest neighbors O ); S206, Set P O Update the dataset to include all data points in the generated P_Echo, and repeat steps S201-S205 to obtain a spatial local density dataset Density(P_Echo) centered on a sphere with each point in the dataset P_Echo as the center. m ), m=1,2,3,…,N E N E This represents the total number of points in the dataset P_Echo.

5. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 4, characterized in that: S3 specifically includes the following: S301. For a spatial point in the generated P_Echo data to be filtered, let this point be Z. f_b Based on spatial distance, extract the distance from Z. f_b The N4 points that are closest in spatial distance to each other are denoted as Z. n n=1,2,3,…,N4, and according to Z n Spatial location from the spatial local density dataset Density(P) m Extract the spatial local density of these N4 points from the dataset to form the dataset Density(Z). n ); S302, Transfer the dataset Density(Z) n Sort the data sequences from smallest to largest, defining the data with percentages Per1 before sorting and Per2 after sorting as low-density and high-density regions respectively. Calculate the average value of each low-density and high-density region, denoted as Density_Z. L and Density_Z H Based on the formula for calculating the local density standard deviation parameter, Density_Z is calculated respectively. L and Density_Z H The corresponding local density standard deviation parameter is denoted as and ; S303, will and Combined into a local density standard deviation parameter diffusion range sequence, denoted as and will Divide into N4 equal parts, and generate each Z n The local density standard deviation parameter corresponding to the point is denoted as . ; S304, Z f_b and Z n The spatial location information is input into the spatial distance calculation model, the model is run, and the output is N4 Z-axis distances. n Distance Z f_b The spatial distance dataset, denoted as d On , then d On With the calculated The values ​​are input together into the sparse data space distance weight calculation formula, and each Z is calculated separately. n The weight value of the point is denoted as Wn. The point Z to be filtered is calculated using the sparse data space filtering formula. f_b The filtered value Z f_a ; S305, Z in S301-S304 f_b Update each point in the P_Echo data to obtain the filtered dataset P_Echo, denoted as P_Echo_filter.

6. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 5, characterized in that: S4 specifically includes the following: S401. Set the spatial resolution of the horizontal grid of latitude and longitude in two-dimensional space to lon_lat_reso, set the spatial resolution of the vertical height in one-dimensional space to Hig_reso, and generate a three-dimensional grid TAR_Grid with a spatial resolution of lon_lat_reso×lon_lat_reso×Hig_reso. S402. The collected high spatial resolution digital elevation data (DEM) is used to generate a three-dimensional terrain surface continuity function, denoted as DEM_F, by using a three-dimensional spatial spline interpolation method. The latitude and longitude horizontal grid in the two-dimensional space of TAR_Grid is input into the three-dimensional terrain surface continuity function DEM_F to obtain the ground height corresponding to each grid point in the latitude and longitude horizontal grid in the two-dimensional space of TAR_Grid. Combining the height information corresponding to each grid point in the height dimension of TAR_Grid, all grid points in the vertical direction of each latitude and longitude horizontal grid point in TAR_Grid that are lower than the ground height are removed to generate the radar echo effective three-dimensional spatial target grid TAR_Grid_1. S403. Based on the obtained radar echo intensity spatial location information P_Echo_radial=(Z_Echo,R_Echo,Az_Echo,El_Echo), input it into the beam unit direction vector construction model, run the model, and generate the beam unit direction vector d=[cos(El_Echo)*sin(Az_Echo), cos(El_Echo)*cos(Az_Echo), sin(El_Echo)]; S404. Input the Cartesian coordinates (x_radar, y_radar, H_radar) of the acquired radar station and the beam unit direction vector d into the beam ray propagation model that considers the curvature of the earth and electromagnetic refraction. Run the model to generate a radar beam ray trajectory point set, denoted as Path_Beam. For any radar beam ray, calculate the height difference between the trajectory point set Path_Beam and the generated three-dimensional terrain surface continuity function DEM_F at the latitude and longitude grid points corresponding to the beam propagation path. Based on the principle of minimizing the height difference, determine the spatial information of the first intersection point between the beam and the terrain surface. S405. Apply the method of step S404 to all radar beam paths to obtain a spatial information dataset of all radar beam paths intersecting with the terrain surface, denoted as P_proj.

7. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 6, characterized in that: S5 includes the following: S501. Based on the obtained P_Echo_filter and P_proj, perform spatial matching according to the principle of shortest spatial distance to form a spatial matching dataset, denoted as Match. Based on the height information of the matching points in the Match dataset, check whether the echo points on each radar ray path are located before the terrain intersection of the beam ray. Remove all echo points on the radar beam ray with a height greater than the height of the intersection point, and generate a high-quality radar echo intensity dataset P_Echo_filter_clear after occlusion judgment. S502. Input the collected high-resolution DEM dataset into the elevation gradient calculation model, run the model, and the model output is the elevation gradient dataset GRAD(Lon,Lat) for the latitude and longitude point (Lon,Lat). Set the spatial sliding window size to N_slid×N_slid, and input it together with the collected high-resolution DEM dataset into the standard deviation calculation model. Run the model, and the model output is the elevation standard deviation dataset STD(Lon,Lat) for each sliding window at the latitude and longitude point (Lon,Lat). S503. Find the maximum value DEMmax and minimum value DEMmin of the DEM in the spatial sliding window, generate the DEM range DEMmax-DEMmin in the spatial sliding window, and use this range value as the local range value EXT(Lon,Lat) of the center point (Lon,Lat) of the spatial sliding window. S504. Apply the steps of S501-S503 to all spatial points in the DEM dataset to obtain the GRAD dataset, STD dataset, and EXT dataset.

8. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 7, characterized in that: The process of merging the three datasets obtained in S5 into a single feature vector and inputting it into the clustering model, and then processing and calculating the output of the clustering model to construct a spatial interpolation strategy, specifically includes the following: B1. Merge the acquired GRAD, STD, and EXT datasets into a single feature vector Fe = [GRAD, STD, EXT]. Input the feature vector Fe into the K-means clustering model and set the number of terrain categories to kind. Run the model. The model outputs the elevation gradient GRADc_p, elevation standard deviation STDc_p, and local extreme value EXTc_p corresponding to each cluster center, p = 1, 2, 3, ..., kind, as well as the maximum elevation gradient GRAD_max, the maximum elevation standard deviation STD_max, and the maximum local extreme value EXT_max across all clusters. B2. Based on the eigenvalue clustering weight calculation formula, calculate the weight coefficients of the three features, namely elevation gradient, elevation standard deviation, and local extrema, in each cluster and normalize them, denoted as W1_p, W2_p, and W3_p respectively. B3. Calculate the terrain complexity index TRI_p for each typical terrain type according to the terrain complexity index calculation formula; based on the calculated TRI_p, construct d spatial interpolation strategies that map one-to-one with different typical terrain regions: Inter_str = [bilinear interpolation, inverse distance weighted interpolation, least squares interpolation, ..., cubic spline interpolation].

9. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 7, characterized in that: The process of calculating and processing the three datasets obtained from S5 to obtain the terrain classification map specifically includes the following: C1. Based on S5, obtain the GRAD, STD, and EXT data at any point Q in the DEM dataset. Divide the GRAD, STD, and EXT data by the maximum value in the corresponding dataset to obtain the normalized GRAD_q, STD_q, and EXT_q data. C2. Substitute the normalized GRAD_q data, STD_q data, and EXT_q data into the terrain complexity index calculation formula to generate the terrain complexity index value TRI_Q for this point. Calculate the absolute difference between TRI_Q and the terrain complexity index TRI_p of kind typical terrain categories. According to the principle of minimum absolute deviation, classify point Q into the terrain category closest to its terrain complexity index, that is, the terrain type to which it belongs is the terrain category corresponding to the smallest deviation of TRI_q. C3. Apply steps C1 and C2 to all spatial points in the DEM to obtain a terrain classification map for each point in the DEM, denoted as Terrain_C.

10. The method for constructing a high-precision three-dimensional echo intensity field in mountainous areas based on phased array radar according to claim 9, characterized in that: The process involves combining the effective 3D spatial target mesh of radar echoes with a high-quality radar echo intensity dataset, inputting it into a spatial distance calculation model, filtering the output results, and determining the interpolation method for the interpolation points by combining terrain classification maps and spatial interpolation strategies. Finally, the 3D radar echo intensity field dataset generated based on the spatial interpolation model specifically includes the following: D1. For any interpolation point in the generated effective 3D spatial target grid TAR_Grid_1 of radar echo, the spatial information is (lon_inter, lat_inter, hig_inter), and the corresponding echo intensity is assumed to be Z_inter. Then, the interpolation point P_Echo_inter = (Z_inter, lon_inter, lat_inter, hig_inter). The generated high-quality radar echo intensity dataset P_Echo_filter_clear after occlusion judgment is input together with the spatial information of the interpolation point P_Echo_inter into the spatial distance calculation model. Run the model, and the output is the spatial distance dataset from each spatial point in P_Echo_filter_clear to the interpolation point P_Echo_inter, denoted as Distance_P_inter. D2. Filter all spatial points smaller than the maximum radius r_max of the neighboring spatial domain from Distance_P_inter to form the neighboring point dataset of the point to be interpolated, denoted as P_Echo_inter_near. Based on the longitude and latitude of the point to be interpolated, combined with the generated terrain classification map Terrain_C, query the terrain category of the point to be interpolated, and then determine the interpolation method of the point based on the obtained spatial interpolation strategy Inter_str. D3. Input the neighboring point dataset P_Echo_inter_near of the interpolation point P_Echo_inter and the interpolation point P_Echo_inter together into the spatial interpolation model, combine the determined interpolation method, run the model, and generate the echo intensity Z_inter of the interpolation point P_Echo_inter. D4. Apply steps D1-D4 to all points in the effective three-dimensional spatial target grid TAR_Grid_1 of the radar echo to generate a high spatial resolution and high precision three-dimensional radar echo intensity field dataset for mountainous areas.

Images