Full-automatic Doppler weather radar speed defuzzification method

Abstract

The invention provides a full-automatic Doppler weather radar speed defuzzification method, which is carried out upwards and downwards from a second elevation angle in order to reduce the influence of factors such as clutter and beam blocking aiming at radar body scanning data. Aiming at the current elevation angle, carrying out good radial search, conditional good radial search and initial reference radial search, and carrying out multi-round omnibearing one-by-one radial deblurring processing on the unprocessed effective velocity library by taking the searched proper initial reference radial direction as a starting point. In order to optimize a speed deblurring effect in a severe horizontal wind shear environment, a horizontal neighborhood method and a radial deblurring step by radial processing and tangential processing are added between a first round of omnibearing radial deblurring step by radial processing and a second round of omnibearing radial deblurring step by tangential processing in an original algorithm; for a long-distance isolated echo region, a vertical neighborhood method deblurring is added after a horizontal neighborhood method deblurring. Compared with an original algorithm, the method has the advantage that the radial speed quality after speed deblurring is higher.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of velocity de-aliasing, and more particularly, to a full-automatic Doppler weather radar velocity de-aliasing method. BACKGROUND

[0002] Doppler weather radar can obtain high spatial and temporal resolution radial velocity in a large detection range, and is widely used in strong convective weather monitoring and early warning, numerical weather prediction model initialization, wind field inversion, ground clutter suppression and other tasks. However, the radial velocity observed by Doppler weather radar has quality problems such as range folding, velocity aliasing, noise point pollution, etc. In order to complete the above tasks with sufficient accuracy, strong quality control must be carried out on the radial velocity data.

[0003] Doppler weather radar has a maximum measurable distance (maximum unambiguous distance ) and a maximum measurable velocity (maximum unambiguous velocity ), which are inversely proportional and proportional to the radar pulse repetition frequency PRF (Pulse Repetition Frequency) respectively, and there is no same PRF that makes and are large, which is called the Doppler dilemma. When the actual wind radial component is outside the ± interval, the radar measured radial velocity is still shown within the ± interval, that is, velocity aliasing (folding) occurs, and the true radial velocity there is a relationship as follows: (1) where is an integer, generally taking 0, 1, 2.

[0004] Staggered or dual-pulse PRF technology (alternately transmitting 2 or 2 groups of pulses with different PRFs) is often used to expand while keeping unchanged. This technology is based on the assumption that the true radial velocities measured by 2 or 2 groups of pulses are equal. However, due to the influence of factors such as natural pulsation of actual wind, small-scale shear and noise, the true radial velocities measured by 2 or 2 groups of pulses with slight spatial differences are not necessarily equal, which leads to a decrease in the quality of radial velocity data obtained using this technology, and there are some large positive and negative velocity alternately jumping singular values, thereby making the application of radial velocity data difficult. For weather systems such as typhoon, tornado supercell, squall line, and strong hailstorm, some of the radial velocities may still exceed the expanded and appear aliasing.

[0005] Velocity ambiguity appears as nearly ± The velocity of the wind is suddenly changed, while the actual wind generally follows the principle of spatiotemporal continuity, although strong wind shear may occur sometimes. Therefore, most velocity deblurring software algorithms are based on the continuity of radial velocity in space and time, from one-dimensional deblurring along the radial or azimuth, two-dimensional deblurring along the radial and azimuth, to four-dimensional deblurring along the radial, azimuth, elevation and time series. The basic idea of ​​velocity deblurring is to: transform each observed radial velocity into and a reference speed When the difference between the two is greater than the preset wind shear threshold, the observed radial velocity is considered blurred and it is necessary to adjust the formula (1) value, so that and The difference is within the preset wind shear threshold. It can be seen that the key to velocity deblurring is to find a reliable Generally, the reference velocity is the average of the velocities of the multiple distance bins adjacent to the current velocity bin which have been processed. But for some boundaries, such as isolated storm or echo region edge surrounded by purple (distance folding failure area), there is no available processed velocity around it, in this case, some velocity de-aliasing algorithms use environmental wind from sounding or numerical model or VAD (velocity azimuth display) based on radar observation as reference velocity, but this environmental wind is not representative for small scale flow field, which easily leads to inappropriate velocity de-aliasing. The variational de-aliasing improves the effect of environmental wind de-aliasing, especially in the hurricane and typhoon environment, but the variational de-aliasing method is still limited for isolated storms far away from the radar. An automatic velocity de-aliasing algorithm independent of auxiliary wind field is proposed, which firstly determines the initial reference radial by automatically searching the weakest wind area, and then performs multi-round two-dimensional de-aliasing in clockwise and counterclockwise 180° directions based on the horizontal continuity of the velocity field, and in the multi-round de-aliasing process, the search range of the reference velocity is more relaxed. The algorithm module runs in the new generation weather radar business software system ROSE3.0 and the disaster weather short-term nowcasting business system SWAN3.0, although the overall velocity de-aliasing effect is good, but there are still some deficiencies, which are shown in the following three aspects: 1) for strong tornado, strong hailstorm, strong typhoon and other situations with strong horizontal wind shear, especially when the velocity aliasing area is connected with the non-aliasing area, inappropriate velocity de-aliasing may occur, which affects the subsequent application such as mesocyclone and tornado vortex feature recognition; 2) the de-aliasing effect of the velocity of the isolated echo area far away from the radar is not good enough, and some velocity aliasing areas are easily left; 3) the noise points are not processed, and the existence of the noise points has a great influence on the effective application of the radial velocity, especially on the small scale convergence, divergence and vortex feature recognition. SUMMARY

[0006] The present application provides a full-automatic Doppler weather radar velocity de-aliasing method, which overcomes the problem of sometimes poor quality of the radial velocity after de-aliasing processing in the existing algorithm, especially in the case of strong horizontal wind shear environment, isolated echo far away from the radar and a large number of noise points.

[0007] The present application provides a full-automatic Doppler weather radar velocity de-aliasing method, which overcomes the problem of sometimes poor quality of the radial velocity after de-aliasing processing in the existing algorithm, especially in the case of strong horizontal wind shear environment, isolated echo far away from the radar and a large number of noise points. Step 1, input the radial velocity data of the current elevation angle and the constraint initial reference radial, and mark the velocity bin, wherein the effective velocity bin is marked as 0 and the invalid velocity bin is marked as 3; Step 2, search for the first good radial for the velocity bin marked as 0 at the current elevation angle; Step 3, when the good radial number GN searched by the first time good radial search is equal to the total radial number BN, the current elevation angle has no velocity ambiguity, the velocity ambiguity elimination process of the current elevation angle is ended, and the velocity ambiguity elimination process of the next elevation angle is entered; otherwise, step 4 is executed; Step 4, conditional good radial search and velocity mean calculation are performed on the velocity library marked as 0 for the current elevation angle; Step 5, the constraint condition of the first initial reference radial search is set, the first initial reference radial search is performed on the velocity library marked as 0 for the current elevation angle based on the calculated velocity mean and the constraint condition, and the initial reference radial is found; Step 6, the velocity library marked as 0 for the initial reference radial and the two radials adjacent to the left and right thereof searched in step 5 is subjected to ambiguity elimination processing; Step 7, the initial three reference radials of the initial reference radial and the two radials adjacent to the left and right thereof searched in step 5 are taken as the initial three reference radials of the full-azimuth step-by-step radial ambiguity elimination processing, three rounds of full-azimuth step-by-step radial ambiguity elimination processing are performed on the velocity library marked as 0 for the current elevation angle, and the processed velocity library is marked as 1, wherein the initial reference radial is taken as the starting point, and the second round of full-azimuth step-by-step radial ambiguity elimination processing is performed on the velocity library marked as 0 for the current elevation angle by using the neighborhood method step-by-step radial ambiguity elimination and the radial-first-tangent-second step-by-step radial ambiguity elimination; Step 8, the second initial reference radial search is performed on the velocity library marked as 0 and 1 for the current elevation angle, the initial reference radial is determined, and steps 6 to 7 are repeatedly executed with the determined initial reference radial as the starting point; Step 9, singular value filtering processing is performed on the velocity library marked as 0 for the current elevation angle, and the radial velocity of the current elevation angle after the ambiguity elimination processing is obtained.

[0008] The full-automatic Doppler weather radar velocity ambiguity elimination method provided by the application is performed on radar volume scan data, and is performed upwards and downwards from the second elevation angle to reduce the influence of factors such as clutter and beam blockage. For the current elevation angle, good radial search and conditional good radial search are performed, and initial reference radial search is performed, and the appropriate initial reference radial searched is taken as the starting point, and multiple rounds of full-azimuth step-by-step radial ambiguity elimination processing are performed on the effective velocity library that has not been processed. In order to optimize the velocity ambiguity elimination effect in the environment of severe horizontal wind shear, the horizontal neighborhood method and the radial-first-tangent-second step-by-step radial ambiguity elimination step are added between the first round and the second round of full-azimuth step-by-step radial ambiguity elimination steps in the original algorithm. For the remote isolated echo area, vertical neighborhood method ambiguity elimination is added after the horizontal neighborhood method ambiguity elimination. Compared with the original algorithm, the radial velocity after the velocity ambiguity elimination of the application has higher quality. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1A flow chart of a full-automatic Doppler weather radar velocity de-ambiguity processing method is provided for the embodiment of the present application. Figure 2 A flow chart of the first initial reference radial search method of the embodiment of the present application is provided. Figure 3 A flow chart of the second initial reference radial search method of the embodiment of the present application is provided. Figure 4 A flow chart of the third initial reference radial search method of the embodiment of the present application is provided. Figure 5 A radial velocity and reflectivity factor chart of a tornado supercell case 1.5° elevation angle is provided as an example 1, wherein (a) is the original velocity, (b) is the velocity after de-ambiguity of the original algorithm, (c) is the velocity after de-ambiguity of the present application, and (d) is the reflectivity factor. Figure 6 A radial velocity and reflectivity factor chart of a typhoon case 0.5° elevation angle is provided as an example 2, wherein (a) is the original velocity, (b) is the good condition radial, (c) is the velocity after de-ambiguity of the original algorithm, (d) is the velocity after de-ambiguity of the present application for the first time and the second round, (e) is the velocity after de-ambiguity of the present application for the second time and the second round, (f) is the velocity after de-ambiguity of the present application for the third round, (g) is the velocity after noise filtering of the present application, and (h) is the reflectivity factor. Figure 7 A radial velocity chart of a 2.4° and 3.3° elevation angle of a scattered strong convection case is provided as an example 3, wherein (a) is the velocity after de-ambiguity of the present application for the 2.4° elevation angle, (b) is the original velocity for the 3.3° elevation angle, (c) is the velocity after de-ambiguity of the original algorithm for the 3.3° elevation angle, and (d) is the velocity after de-ambiguity of the present application for the 3.3° elevation angle. Figure 8 A radial velocity and reflectivity factor chart of a 2.4° elevation angle of a double PRF mode squall line case is provided as an example 4, wherein (a) is the original velocity, (b) is the velocity after de-ambiguity of the original algorithm, (c) is the velocity after de-ambiguity of the present application, and (d) is the reflectivity factor. DETAILED DESCRIPTION

[0010] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of the present application. In addition, the technical features in each of the embodiments or in a single embodiment provided by the present application can be combined with each other at will to form a feasible technical solution, and such combination is not restricted by the order of steps and / or structure mode, but should be based on the fact that it can be realized by those of ordinary skill in the art. When the combination of technical solutions appears to be contradictory or unfeasible, it should be considered that such combination of technical solutions does not exist and is not within the protection scope of the present application.

[0011] Figure 1 A flow chart of a full-automatic Doppler weather radar velocity de-aliasing method provided by one embodiment of the present application is shown in FIG. 1, which comprises the following steps: Figure 1 Step 1: input the radial velocity at the current elevation angle and the constraint initial reference radial, and mark the velocity library, wherein the valid velocity library is marked as 0 and the invalid velocity library is marked as 3.

[0012] It can be understood that the radar detects radial velocity data under multiple elevations, each azimuth and slant range library. Since there is velocity aliasing, the radial velocity under each elevation needs to be de-aliased. Before de-aliasing the radial velocity, a mark value is assigned to all velocity libraries, wherein the valid velocity library is marked as 0 and the invalid velocity library is marked as 3.

[0013] Each time the velocity library is de-aliased, only the velocity library marked as 0 is processed, and after the velocity library marked as 0 is de-aliased, the velocity library is marked as 1.

[0014] It should be noted that the radar collects radial velocity at different elevations, and the elevation sequence is usually 0.5°, 1.5°, 2.4°,.... Among them, since the radar observation data at the first elevation angle is easily affected by factors such as ground objects and beam blockage, the probability of not searching out the initial reference radial or searching out the inappropriate initial reference radial is large, and it is not suitable for being the constraint initial reference radial of the adjacent elevation, therefore, the present application starts from the second elevation angle for upward and downward velocity de-aliasing, i.e., the initial reference radial at the 1.5° elevation angle is taken as the constraint initial reference radial of the 0.5° and 2.4° elevation angles.

[0015] ​The deblurring process for each elevation angle is as follows: the radial velocity of the current elevation angle and the constrained initial reference radial are input. The constrained initial reference radial for each elevation angle is the initial reference radial found by searching for adjacent elevation angles using the first and second methods. Therefore, for the first elevation angle processed, its constrained initial reference radial is invalid. If no initial reference radial is found using the first and second initial reference radial search methods for adjacent elevation angles processed, the constrained reference radial for the current elevation angle is also invalid.

[0016] See Figure 1 After inputting the radial velocity at the current elevation angle, we remove isolated noise points. Specifically, we take an M×N (default value is 5×5) window centered on the current valid velocity library and calculate the ratio of invalid velocity libraries to the total number of grid points in the window. If the ratio is ≥ 70%, the velocity value in the current library is replaced with the invalid value. This step aims to reduce the impact of isolated noise on the reference velocity search.

[0017] Step 2: Perform the first good radial search for the velocity library with the current elevation angle marked as 0.

[0018] It is understandable that after the isolated noise points are removed, the first good radial search is performed on the velocity library with the current elevation angle marked as 0.

[0019] Among them, a good radial is defined as: there is no strong shear between all adjacent effective velocity bins marked as 0 or 1 on a radial.

[0020] The no strong shear condition is: (2) in, is the i-th effective radial velocity, is the i-1th effective radial velocity, =1, 2, … is the valid speed library number, is an adjustable coefficient (default value is 0.8), is the maximum unambiguous speed.

[0021] In this embodiment of the present invention, a radial direction in which the number of adjacent effective speed pairs is 0 is also regarded as a good radial direction.

[0022] Step 3: When the number of good radials GN found in the first good radial search is equal to the total number of radials BN, there is no velocity ambiguity at the current elevation angle, and the velocity deambiguation processing of the current elevation angle is ended, and the processing of the next elevation angle is entered; otherwise, step 4 is executed.

[0023] It can be understood that when the good radial number GN found by the first good radial search is equal to the total radial number BN, it indicates that there is no velocity ambiguity for the elevation angle, and the algorithm enters the next elevation angle, aiming to save processing time.

[0024] When the good radial number GN found by the first good radial search is less than the total radial number BN, subsequent de-ambiguity processing steps need to be performed, i.e., step 4 is performed.

[0025] Step 4, conditional good radial search and velocity average calculation are performed on the velocity file of the current elevation angle marked as 0.

[0026] Among them, the conditional good radial is defined as: the maximum value of the length of the continuous valid velocity file marked as 0 or 1 is greater than the set threshold value (the default value is 5 km) and the maximum velocity value < Vmax. The good radial is = 0.8.

[0027] For the conditional good radial, the following velocity average is calculated , : (3) Among them, (4) Among them, is the number of valid velocity files, is the number of radial velocity files satisfying , and is an adjustable coefficient (the default value is 0.3). The conditional good radial search is newly added on the basis of the original velocity de-ambiguity algorithm, and the subsequent initial reference radial is searched from the conditional good radial rather than the good radial.

[0028] Step 5, the constraint condition for the first initial reference radial search is set, and based on the calculated velocity average and the constraint condition, the first initial reference radial search is performed on the velocity file of the current elevation angle marked as 0 to find the initial reference radial. It can be understood that based on the velocity average , calculated by the conditional good radial of step 4 and some constraint conditions, the first initial reference radial search is performed on the velocity file of the current elevation angle marked as 0 to find the appropriate initial reference radial, and the subsequent de-ambiguity processing is performed on the velocity file marked as 0 with the appropriate initial reference radial found as the starting point.

[0029] Wherein, the successful search of the area without velocity ambiguity as the starting point of the de-ambiguity is crucial for the correct de-ambiguity. Generally, the weak wind area without strong shear is the ideal area for the starting point of the de-ambiguity, which is generally located near the real zero velocity line (wind direction perpendicular to the radial direction). Generally, there are two real zero velocity lines extending outward from the radar in the radial velocity diagram of one elevation angle, and when there is no large-scale convergent or divergent flow field, the two real zero velocity lines are distributed in nearly 180° symmetry, and when there is a large-scale convergent or divergent flow field, the small angle thereof will be <180°, but generally it will be relatively large. In the case of strong wind in a large area (such as a strong typhoon), it is possible that all the effective velocities in the radial direction are ambiguous and there are many velocity values near the zero velocity. The false zero velocity area caused by velocity ambiguity is generally located near the wind speed maximum area (wind direction parallel to the radial direction) of one elevation angle, and the small angle of the real zero velocity line is generally about 90°.

[0030] In order to avoid searching too many real zero velocity radials or false zero velocity radials as initial reference radials, the following three constraints are used in the first initial reference radial search: The first constraint: at most two initial reference radials are determined at the same elevation angle; The second constraint: the small angle of the two initial reference radials at the same elevation angle is >120°; The third constraint: the small angle of the initial reference radial at the current elevation angle and the initial reference radial at the adjacent elevation angle that has been processed (hereinafter referred to as the constraint initial reference radial) is <30° or >150°.

[0031] The following three progressive search methods are used in the first initial reference radial search, wherein the first and second methods are different from the original algorithm, and the third method is newly added, and only when the initial reference radial is found, the de-ambiguity step is continued.

[0032] Based on the three constraints, the first search method is used to search for at most two initial reference radials; If two initial reference radials are not searched out by the first search method, one or two initial reference radials are searched out by the second search method, so that the sum of the initial reference radials searched out by the first search method and the initial reference radials searched out by the second search method is two; If the initial reference radial is not found by the first search method and the second search method, the third search method is used to search for the initial reference radial; If the initial reference radial is not found by the three search methods, it is checked whether the direction corresponding to the radial of the initial reference radial of the adjacent elevation angle is a good radial, and if so, the radial is taken as the initial reference radial of the current elevation angle.

[0033] It is understandable that the three search methods are in a progressive relationship. First, the first search method is used to search for two initial reference radials as much as possible. If the first search method cannot find two initial reference radials, the second search method is used to search, so that the first and second search methods can find as many as two initial reference radials as possible. Of course, if two initial reference radials cannot be found, it is also acceptable to use the first and second search methods to find one initial reference radial. If the first and second search methods both fail to find an initial reference radial (not even one initial reference radial), the third search method is used to search for one initial reference radial.

[0034] Among them, the flowchart of the first search method is as follows Figure 2 As shown, for each elevation angle, search for continuous radial areas with good conditions, that is, search for continuous areas with good radial roots ≥ the set threshold (the default value is 5), and search in this area ( =0.3) value, for each positive-negative conversion area found (a positive-negative conversion area includes multiple radials), The larger radial is determined as the backup initial reference radial. The first search method is consistent with the original algorithm requirements. The sign of the radial direction changes from 3 consecutive radial directions to the next 3 consecutive radial directions. Compared with the original algorithm, the present invention only needs to find the adjacent radial direction in the radial area with good continuous conditions. The positive and negative conversion area of ​​the value makes it easier to find the initial reference radial in the zero velocity area. If N (N>1) spare initial reference radials are found, the constraints and Based on the minimum principle, at most two initial reference radials are determined and used as the constrained initial reference radials A0 and A1 for the next adjacent elevation angle.

[0035] For reference Figure 2 ,The specific steps of the first search method include: For the current elevation angle, search for continuous radial areas with good conditions, that is, search for continuous areas with good radial roots ≥ the set threshold, and search in this continuous area. The positive and negative conversion area of ​​the value The larger radial is determined as the backup initial reference radial; If multiple spare initial reference radials corresponding to multiple positive and negative conversion zones are found, the corresponding Sort from smallest to largest; If the constrained initial reference radial A0 is invalid, the first spare initial reference radial after sorting is determined as the first initial reference radial; if A0 is valid, the first spare initial reference radial that meets the third constraint condition is determined as the first initial reference radial, and the first radial that meets the second constraint condition and the third constraint condition is searched among the remaining spare initial reference radials as the second initial reference radial, and the two initial reference radials searched out are used as the constrained initial reference radials A0 and A1 of the next adjacent elevation angle.

[0036] If the number of initial reference radial roots found using the first search method is ≤ 1, the search is continued using the second search method.

[0037] The flowchart of the second search method is as follows Figure 3 As shown, in the radial with good conditions, search for all The radials on the left and right of the spare initial reference radials are both good radials. The difference between the second search method and the original algorithm is that the search starts from the good radials instead of the good radials, and the requirement is that the left and right radials of the spare initial reference radials are both good radials. If N (N>1) spare initial reference radials are found, the constraints and The minimum principle determines at most two initial reference radials, which serve as the constrained initial reference radials A0 and A1 for the next adjacent elevation angle.

[0038] refer to Figure 3 ,The specific steps of the second search method include: In the radial with good conditions, search for all The left and right radials are both good radials as spare initial reference radials. If multiple spare initial reference radials are found, the corresponding radials of each spare initial reference radial are used. Sort from smallest to largest; If an initial reference radial is found using the first search method, then based on the second and third constraints and The second initial reference radial is determined using the minimum principle. If the first search method fails to find an initial reference radial, determine whether A0 is valid. If A0 is invalid, select the first backup initial reference radial as the first initial reference radial. If A0 is valid, select the first backup initial reference radial that satisfies the third constraint as the first initial reference radial. Search the remaining backup initial reference radials for the first radial that satisfies both the second and third constraints as the second initial reference radial. These two initial reference radials serve as the constrained initial reference radials, A0 and A1, for the next adjacent elevation angle.

[0039] If the initial reference radial is not found by using the first and second search methods, the third search method is used to search for an initial reference radial.

[0040] The flow chart of the third search method can be seen in Figure 3 If the constraint initial reference radials A0 and A1 are invalid, search for and the minimum condition good radial as the initial reference radial; if the constraint initial reference radials A0 and A1 are valid, search for , and the minimum radial with a small angle < 30° with the constraint initial reference radials A0 and A1 as the initial reference radial. The initial reference radial searched by the third search method cannot be used as the constraint initial reference radial of the next elevation because the constraint condition of the third search method is relatively wide.

[0041] If the initial reference radial is not found by using the three search methods, check whether the radial corresponding to the azimuth of the initial reference radial of the adjacent elevation is a condition good radial, and if so, take the radial as the initial reference radial of the current elevation.

[0042] Step 6: Perform deblurring processing on the speed bins marked 0 on the initial reference radial and the two radials adjacent to the left and right of the initial reference radial searched in step 5.

[0043] It can be understood that after the initial reference radial is found by step 5, deblurring processing is performed on the speeds of the initial reference radial and the radials adjacent to the left and right of the initial reference radial. The deblurring processing is performed by adjusting the value of the formula (1) to find the value closest to If is the adjustable parameter and the default value is 0.2, the value of the speed bin is replaced by , the mark is set to 1, otherwise, the speed bin is not processed and the mark remains 0. Each round of deblurring is only for the speed bins marked 0. Specifically, first, use or as the reference speed

[0044] to perform deblurring processing on each speed bin marked 0 on the initial reference radial; then, use the speed value marked 1 on the initial reference radial as the reference speed to perform azimuth deblurring processing on each speed bin marked 0 on the two radials adjacent to the left and right of the initial reference radial. The initial reference radial and the radials adjacent to the left and right of the initial reference radial are taken as the initial reference radial of the full-azimuth deblurring of each radial. ​​​

[0045] Step 7, taking the initial three reference radials after step 6 as the starting point, the velocity database of the current elevation marked as 0 is processed by three rounds of full-directional radial deblurring, and the processed velocity database is marked as 1, wherein, taking the initial reference radials as the starting point, the velocity database of the current elevation marked as 0 is processed by the second round of full-directional radial deblurring by using the neighborhood method and the radial-first-tangential-second radial deblurring method.

[0046] It can be understood that step 6 processes the velocities on the initial reference radials and the adjacent left and right two radials, and then takes the initial three reference radials as the starting point to process the velocity database of the current elevation marked as 0 by multiple rounds of radial deblurring. In the embodiment of the present application, three rounds of full-directional radial deblurring are mainly performed, wherein the initial reference radials of the three rounds of full-directional radial deblurring are the initial three reference radials after step 6.

[0047] The first round of full-directional radial deblurring processing includes: The first round of full-directional radial deblurring processing includes:

[0048] In the first round of full-directional radial deblurring, the standard for determining the reference velocity is relatively strict, the horizontal wind shear around the reference velocity database is small, and the reference velocity only affects a small area around it. Therefore, the processed velocity is considered to be correct, has high reliability, and provides a good background field for the subsequent deblurring processing.

[0049] After the first round of velocity deblurring, the area with relatively large horizontal wind shear is not processed, so the second round of full-directional radial deblurring is needed, wherein the second round of full-directional radial deblurring is newly added on the basis of the original algorithm, and mainly includes the second round of full-directional radial deblurring of the velocity database of the current elevation marked as 0 by using the neighborhood method and the radial-first-tangential-second radial deblurring method.

[0050] In the horizontal neighborhood method, the de-aliasing is performed radially for each beam: starting from the initial reference radial, the de-aliasing is performed for 360° in both clockwise and counterclockwise directions. For each radial, the de-aliasing is performed once from far to near and once from near to far. For each velocity bin marked as 0, the de-aliasing is performed using the average of the 8 adjacent velocity bins when the number of velocity bins marked as 1 is greater than or equal to 3. Compared with the average of the 3 adjacent velocity bins satisfying the condition of no strong shear (equation (2)) used as the reference velocity in the tangential or radial de-aliasing in the first round of full azimuth radial de-aliasing, the reference velocity used in the horizontal neighborhood method has a smaller influence range, so the condition of no strong shear (equation (2)) is not considered, and the possibility of finding a usable reference velocity is greater, which is beneficial to the de-aliasing of the velocity in a small scale strong wind shear area.

[0051] In the vertical neighborhood method, the de-aliasing is performed radially for each beam: similar to the horizontal neighborhood method, the difference is that the reference velocity is not from the current elevation (since more than 3 velocity bins marked as 1 cannot be found for the current elevation), but is the average of the velocity bins marked as 1 in the 3x3 grid points centered on the current azimuth-range bin at a lower elevation. This is beneficial to the de-aliasing of the velocity in a distant isolated echo area. When the reference velocity at a short distance is not greater than the possible true value of the velocity at a distant place , the de-aliased velocity in the distant isolated echo area cannot be processed. The reason for using the average of the velocity bins marked as 1 at a lower elevation as the reference velocity is that the echo area at a lower elevation is generally larger than that at a higher elevation, and the velocity at a distant place has a greater possibility of being correctly de-aliased.

[0052] Since the reference velocity used in the neighborhood method has a small influence range, 3 velocity bins marked as 1 cannot be found for each velocity bin marked as 0, so the de-aliasing of the velocity is combined with the de-aliasing of the velocity in the first radial and then tangential processing.

[0053] In the first radial and then tangential processing, the de-aliasing is performed radially for each beam: similar to the first round of full azimuth radial de-aliasing, the difference is that for each radial, the de-aliasing is performed first in the radial direction and then in the tangential direction, and the de-aliasing is performed twice for each radial. In the second time, only the average of the 2 adjacent velocity bins marked as 1 and satisfying the condition of no strong shear (equation (2)) is used as the reference velocity, and in the third time, only the adjacent velocity bin marked as 1 is used as the reference velocity. This is beneficial to the de-aliasing of the velocity in a small scale and large horizontal wind shear area, because in these areas, it is difficult to find 3 or 2 consecutive velocity bins marked as 1 satisfying the condition of no strong shear (equation (2)) in the radial or tangential direction, so it is difficult to find a usable reference velocity.

[0054] By changing to ​Threshold, the second round of full-azimuth root radial deblurring processing is executed once, specifically, the neighborhood method and the root radial deblurring process of the first radial and then tangential processing in the second round of full-azimuth root radial deblurring processing method are repeated once, but the difference from the first time is that the deblurring processing is performed, The threshold is 0.3 instead of 0.2, which is good for the velocity deblurring of very strong horizontal wind shear areas (such as tornado vortex features).

[0055] After the first two rounds of velocity deblurring processing, there may still be some areas (such as scattered or isolated small strong wind areas) whose velocities have not been processed due to the lack of good reference velocities in the vicinity, and their indicators are still 0. Therefore, the algorithm designs a third round of full-azimuth root radial deblurring to solve this problem. The third round of full-azimuth root radial deblurring processing is similar to the first round of full-azimuth root radial deblurring, but the search range for finding reference velocities is more relaxed, and the reference velocities are searched within a limited radial threshold (the default value is 5) and a limited distance threshold (the default value is 5 km).

[0056] The third round of full-azimuth radial deblurring process is repeated three times (adjustable) with relaxed and thresholds (default values ; ), which is good for the velocity deblurring of isolated echo areas at a long distance or with a large azimuth interval.

[0057] Step 8: Perform the second initial reference radial search on the velocity database with the current elevation indicators of 0 and 1 to determine the initial reference radial. Take the determined initial reference radial as the starting point and repeat steps 6-7.

[0058] It can be understood that after the deblurring processing of the initial reference radial and its left and right adjacent 2 radials and the three rounds of full-azimuth root radial deblurring processing, there may still be some unprocessed velocity blur areas in the radial velocity of an elevation, and the most common case is isolated strong wind areas far from the radar. Therefore, after completing the third round of full-azimuth root radial deblurring processing, the second good radial search is performed again, and if the number of searched good radials GN is less than the total number of radials BN, the radial with the maximum number of valid velocity databases and its left and right adjacent radials are used as the initial reference radial for the full-azimuth root radial deblurring processing. Repeat the deblurring process of the previous steps 6-7. The difference between this step and the original algorithm is that the constraint condition that the left and right adjacent radials of the initial reference radial are both conditionally good radials is added, which improves the reliability of the initial reference radial.

[0059] Step 9: Perform singular value filtering on the velocity library with the current elevation angle marked as 0 to obtain the deblurred radial velocity of the current elevation angle.

[0060] It is understandable that after the aforementioned deblurring process, there may still be a velocity bin marked as 0 in the velocity field. This situation usually occurs in the noise interference area, where the noise interference may cause false strong wind shear, but it cannot reach the value close to 2 caused by velocity blur. Therefore, these velocities will not be treated as velocity blur. In this step, for each velocity bin marked as 0, the value of that bin is replaced with the average of the velocities marked as 1 in the eight adjacent velocity bins, and its flag is set to 1. Filtering is performed only on the velocity bin marked as 0, which will not affect the authenticity of other velocities and will not produce a smoothing effect.

[0061] The following example analysis is used to illustrate the effect of the fully automatic Doppler weather radar velocity defuzzification method provided by the present invention.

[0062] Specifically, an EF4 tornado case and a super typhoon case were used to compare and analyze the results of the velocity deambiguation process before and after the algorithm was improved; a scattered severe convection case and a dual PRF model squall line case were used to compare and analyze the final results of the velocity deambiguation before and after the algorithm was improved.

[0063] Case 1 was an EF4 supercell tornado observed by XX radar on June 23, 2016, which caused 99 deaths. Figure 5 This is the radial velocity diagram of the tornado supercell storm at an elevation angle of 1.5° observed by the XX radar at 14:20 on June 23, 2016 (see Figure 5 (a), (b), (c)) and reflectivity factor graph ( Figure 5 In (d), where Figure 5 (a) represents the original unprocessed speed data, (b) represents the speed data after deblurring by the original algorithm, and (c) represents the speed data after deblurring by the present invention. is 27.82 m / s).

[0064] Near the hook echo of the tornado supercell, the original radial velocity appears blurred (see Figure 5 The white arrow in (a) indicates -19 The bright blue area of ​​-27 m / s is not a closed positive velocity area. Looking clockwise from the radar, the positive velocity on the far right of the large velocity ambiguity area is continuous in magnitude with the unambiguous positive velocity adjacent to its right, indicating that the velocity ambiguity area is connected to the unambiguous area. After deblurring using the original algorithm, not only was the ambiguity not correctly processed, but inappropriate velocity deblurring also occurred at the far end of the velocity ambiguity area (see Figure 5the original algorithm, the unblurred velocity is mistaken as blurred velocity and is processed. After the improved algorithm is used to deblur, the velocity blur is successfully processed and no inappropriate deblurring phenomenon occurs Figure 5 In (c) of FIG. 1, the tornado vortex feature structure is clear.

[0065] Table 1 is Figure 5 In (b) of FIG. 1, the velocity values near the tornado vortex feature are shown, wherein different color shades represent intermediate results of the deblurring process. In the first round of deblurring process of the original algorithm, the velocity bins marked as 1 in the region are processed in the counterclockwise direction, and the velocity values do not change after deblurring, i.e., these velocities are not blurred. In the first and second deblurring processes of the second round of the original algorithm, the velocities in the black box, the blue and gray shaded regions are all processed inappropriately as unblurred velocities. In the third deblurring process, the blue numbers after the separator “|” are processed inappropriately as a result of using the inappropriately processed velocities as reference velocities. After two rounds of deblurring, from the edge of the region with large horizontal wind shear, there are many velocity bins marked as 0 which are not processed.

[0066] Table 1 Figure 5 In (b) of FIG. 1, the radial velocity values (unit: 0.1 m / s) in the tornado vortex feature region are shown (A is the azimuth angle, and R is the slant range)

[0067]

[0068] In Table 1 below, 1, 2, 3, and 4 represent the following meanings: 1, the first round of deblurring, 2, the first time of the second round of deblurring, 3, the second time of the second round of deblurring, and 4, the third time of the second round of deblurring (1-4 is a step-by-step process).

[0069] In Table 1, the black numbers represent velocities which are not deblurred or do not change after deblurring (unit: 0.1 m / s); the blue numbers after the separator “|” are inappropriately deblurred velocities; the numbers in the black box are velocities which are mistaken as unblurred, but in fact they are blurred; and the numbers without hatching are velocity bins marked as 0 which are not processed.

[0070] Table 2 is Figure 5Velocity values ​​near the tornado vortex feature in (c), where different color shading represents the intermediate results of the deblurring process. Because the 1.5° elevation angle is the starting elevation angle for deblurring the volume scan data, the vertical neighborhood method deblurring step is not executed due to the lack of valid reference velocity. As can be seen from the table, after the first round of deblurring, a new second round of deblurring using the neighborhood method (in the clockwise direction) and alternating radial and then tangential processing (in the counterclockwise direction) can correctly process the velocity in the strong wind shear area along the tangential or radial direction, avoiding inappropriate velocity deblurring and expansion. The present invention processes all valid velocity libraries in the area, unlike the original algorithm, which still has many unprocessed velocity libraries marked as 0 after the deblurring process.

[0071] Table 2 Figure 6 Radial velocity value of the tornado vortex characteristic area in (c) (unit: 0.1m / s) (A is the azimuth, R is the slant distance)

[0072]

[0073] Among them, the meanings of 1, 2, 3, and 4 at the bottom of Table 2 are: 1. First round of deblurring, 2. Second round of deblurring, first deblurring using the neighborhood method, 3. Second round of deblurring, first deblurring using the radial and posterior positions, 4. Second round of deblurring, second deblurring using the neighborhood method, 5. Second round of deblurring, first deblurring using the radial and posterior positions, 2nd time (1-5 are a step-by-step process).

[0074] In Table 2, the black numbers represent the speeds that were not deblurred or did not change after deblurring (unit: 0.1 m / s), and the red numbers after the separator “|” represent the speeds that were correctly deblurred.

[0075] Case 2 is the super typhoon "Haiyan" observed by XX radar on September 24, 2008. The maximum wind speed of the typhoon was 58m / s when it made landfall at 6:45 on the 24th. Figure 6 The radial velocity of Typhoon Hagupit at an elevation angle of 0.5° as observed by XX radar at 3:17 on September 24, 2009 ( Figure 6 (a)~(g)) and reflectivity factor ( Figure 6 (h) in which, Figure 6 In the figure, (a) represents the original velocity data, (b) represents the radial direction with good conditions, (c) represents the velocity data after deblurring using the original algorithm, (d) represents the velocity data after the second round of deblurring using the present invention, (e) represents the velocity data after the second round of deblurring using the present invention, (f) represents the velocity data after the third round of deblurring using the present invention, and (g) represents the velocity data after noise filtering using the present invention. The initial reference radial orientation is 333.3°. is 26.84m / s, (d)~(f) is a step process).

[0076] From the original radial velocity map ( Figure 6 As can be seen from (a), the weak wind area is located near 160° and 300°, and in the southeast quadrant, the velocity fuzzy area is connected to the non-fuzzy area ( Figure 6 From the reflectivity factor diagram ( Figure 6 As can be seen from (h) in the figure, there are relatively many clutters within 50km from the radar. Due to the influence of clutter, the radial area with good search conditions is not very continuous ( Figure 6 (b) in the figure). During the first initial reference radial search, the initial reference radial (the initial reference radial with an elevation angle of 1.5°) was constrained to 333.3° and 162.5°. The first and second search methods failed to find the initial reference radial. The third search method was used to search from the radial with good conditions ( Figure 6 The initial reference radial direction found in (b) is 333.3°. After the first round of all-round root-by-root radial deblurring starting from this radial direction, the radial velocity hardly changes (figure omitted). This is because the horizontal wind shear near the edge of the velocity ambiguity region is greater than the set shear threshold (0.2 After the second round of all-round root-by-root radial deblurring using the original algorithm, the deblurring results in the range of 162-277° are almost all inappropriate ( Figure 6 In (c) of the figure, this area should be in the positive velocity zone away from the radar, but after deblurring, it is almost all in the negative velocity zone towards the radar. The main reasons for the inappropriate velocity deblurring are: first, the velocity ambiguity zone and the unambiguous zone in the southeast quadrant are connected; second, the search range for the reference velocity in the second round of deblurring in the original algorithm is large. Due to the connection between the ambiguity zone and the unambiguity zone, the reference velocity found is inappropriate. After the second round of deblurring of the present invention ( Figure 6 In (d) above, most of the fuzzy velocities are correctly processed, and the connected areas between the fuzzy and unfuzzy velocity areas are also correctly processed. However, there are still some sporadic velocity ambiguities remaining. For example, there are some undefuzzified areas near the azimuths of 30°, 90°, 242°, and 266°. Possible reasons include partial beam blocking leading to false strong wind shear, real strong wind shear, no reference velocity, clutter, etc. After the second round of defuzzification by the present invention, the unprocessed velocity ambiguity area is smaller, and only significant velocity ambiguity remains near the beam blocking area due to the lack of reference velocity ( Figure 6 After three rounds of deblurring according to the present invention, the blur velocities near the three beam blocking areas are also correctly processed ( Figure 7(f) in the figure), but there are still some noise velocities in the figure. This is because the presence of clutter causes false strong wind shear, resulting in these velocities not being processed in the previous deblurring process and their labels are still 0. After filtering the velocity library marked as 0, the noise in the velocity field is significantly reduced ( Figure 7 (g) in the figure).

[0077] Case 3 is the scattered severe convection observed by the XX radar at 21:59 on March 22, 2023 ( Figure 7 ). Figure 7 The 2.4° ( Figure 7 (a) shows the deblurred radial velocity of the present invention) and the 3.3° elevation angle ( Figure 7 (b) in the figure represents the original speed data. Figure 7 (c) in the figure represents the deblurring result of the original algorithm. Figure 7 (d) in the figure represents the deblurred image of the present invention. is 27.73m / s).

[0078] After deblurring using the present invention, the radial velocity at an elevation angle of 2.4° is reasonable ( Figure 7 (a) in the figure), the zero velocity line across the radar is approximately a straight line, with all radial velocities on the left and all positive on the right. On the original radial velocity map at an elevation angle of 3.3°, there is a velocity blur in an isolated echo area 100 km west and 50 km east of the radar ( Figure 8 (b) in the figure). After using the original algorithm for velocity deblurring, the blurred velocity on the west side is not processed because there is no valid reference velocity ( Figure 8 (c) in the vertical neighborhood deblurring method of the present invention uses the deblurred velocity at an elevation angle of 2.4° as the reference velocity and successfully handles the blurred velocity on the west side ( Figure 8 (d) in the figure).

[0079] Case 4 is the squall line observed by radar at 11:00 on April 19, 2023 ( Figure 8 ), Figure 8 The radial velocity at an elevation angle of 2.4° observed by the XX radar at 11:00 on April 19, 2023 ( Figure 8 (a)~(c)) and reflectivity factor ( Figure 8 (d) in the equation, where Figure 8 In the figure, (a) represents the original velocity data, (b) represents the deblurred data of the original algorithm, and (c) represents the deblurred data of the present invention. It is 35.28m / s.

[0080] XX radar uses dual PRF to sample radial velocity from ​In (a) of FIG. 1, there is a bright yellow velocity ambiguity area in the range of 110-130 km at about 250-260° azimuth, and there are many singular values (or noise) with positive and negative velocity alternately jumping. After de-ambiguity using the original algorithm, the ambiguous velocity is correctly processed, but the velocity singular values almost have no change ​ In (b) of FIG. 1, their labels are still 0. After de-ambiguity using the present application, those velocity singular values are filtered out ​ In (c) of FIG. 1, but the velocity field does not appear smooth, because the velocities except the singular values before filtering have been processed and their labels are all 1, and the filtering only processes the velocity with label 0.

[0081] Evaluation of the present application: 3519 individual scan data of tornado, squall line, isolated severe storm, typhoon, thunderstorm gale and other types of severe weather events observed by S-band radar are used to test and evaluate the improved velocity de-ambiguity algorithm. The de-ambiguity accuracy is counted in PPI units, and through manual judgment, as long as there is obvious de-ambiguity error in one PPI, the de-ambiguity accuracy of this PPI is considered to be zero. Table 3 gives the radar data sources and velocity de-ambiguity accuracy, and from the table, it can be seen that the velocity de-ambiguity accuracy of the improved algorithm is almost 100%, and inappropriate velocity de-ambiguity or residual ambiguous velocity phenomenon occasionally occurs in the case of clutter interference or distant scattered echoes.

[0082] Table 3 Radar individual scan data sources and de-ambiguity accuracy

[0083] The present application compares and analyzes the velocity de-ambiguity effect before and after the algorithm improvement through four strong weather examples, and the improved algorithm is evaluated by using 3519 individual scan data observed by S-band radar, and some advantages of the de-ambiguity algorithm of the present application are as follows: (1) The horizontal neighborhood method and the full azimuth radial de-ambiguity step of first radial and then tangential processing are inserted between the first and second full azimuth radial de-ambiguity steps of the original algorithm, which can solve the velocity de-ambiguity problem in the case of severe horizontal wind shear. The reference velocity used by the horizontal neighborhood method has a smaller influence range, and there is no need to consider the second constraint condition, so the possibility of finding available reference velocity is greater; the radial de-ambiguity of first radial and then tangential processing in the second de-ambiguity process is more conducive to the velocity de-ambiguity in the strong azimuth shear area.

[0084] (2) When the reference velocity at a short distance and the possible true value of the velocity at a long distance do not satisfy In this case, the velocity of the isolated echo region at a long distance cannot be processed. The velocity deblurring problem of the isolated echo region at a long distance can be significantly improved by adding the vertical neighborhood method deblurring after the horizontal neighborhood method deblurring in the present application. The reason for using the velocity average value of the adjacent low elevation angle as the reference velocity is that the echo area of the low elevation angle is generally larger than that of the high elevation angle, and the velocity at a long distance has a greater possibility of being correctly deblurred.

[0085] (3) After all the multi-round velocity deblurring based on the continuity principle is completed, the neighborhood average filtering processing is performed on the velocity library still marked as 0, which can significantly reduce the velocity noise (especially the singular value of the positive and negative velocity alternately jumping in the double PRF velocity field), without affecting the authenticity of the velocity library marked as 1, and without producing the smoothing effect of the velocity field, because before filtering, except for the noise (singular value), other velocities have been processed and are marked as 1, and the filtering only processes the velocity library marked as 0.

[0086] (4) The velocity deblurring accuracy of the present application is almost 100%, and inappropriate velocity deblurring or velocity deblurring residual phenomenon occasionally occurs in the case of clutter interference or isolated echo at a long distance. If clutter suppression is performed before velocity deblurring, the velocity deblurring accuracy can be further improved, because too much clutter will affect the continuity of the velocity field, thereby affecting the correct search of the initial reference radial and reference velocity.

[0087] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0088] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications to the embodiments once they know the basic inventive concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0089] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A fully automatic Doppler weather radar velocity deblurring method, characterized in that: include: Step 1: Input the radial velocity data of the current elevation angle and the constrained initial reference radial, and mark the velocity library, where the valid velocity library is marked as 0 and the invalid velocity library is marked as 3; Step 2, perform the first good radial search for the velocity library with the current elevation angle marked as 0; Step 3: When the number of good radials GN obtained in the first good radial search is equal to the total number of radials BN, there is no velocity ambiguity at the current elevation angle, and the velocity deambiguation process for the current elevation angle is terminated, and the velocity deambiguation process for the next elevation angle is started; otherwise, step 4 is executed; Step 4: Perform a good radial search and velocity mean calculation on the velocity library with the current elevation angle marked as 0; Step 5: Set the constraints for the first initial reference radial search. Based on the calculated velocity mean and the constraints, perform the first initial reference radial search on the velocity library marked as 0 at the current elevation angle to find the initial reference radial. Step 6: Defuzzify the velocity library marked as 0 on the initial reference radial direction and the two radial directions adjacent to it on the left and right sides found in step 5; Step 7: Using the initial reference radial and its two adjacent radials found in step 5 as the initial three reference radials for omnidirectional root-by-root radial deblurring, three rounds of omnidirectional root-by-root radial deblurring are performed on the velocity library whose current elevation angle is marked as 0. The processed velocity library is marked as 1. Starting from the initial reference radial, a second round of omnidirectional root-by-root radial deblurring is performed on the velocity library whose current elevation angle is marked as 0 using the neighborhood method and radial-first then tangential root-by-root radial deblurring. Step 8: Perform a second initial reference radial search on the velocity library with the current elevation angles marked as 0 and 1 to determine the initial reference radial. Repeat steps 6 to 7 once, using the determined initial reference radial as the starting point. Step 9: Perform singular value filtering on the velocity library with the current elevation angle marked as 0 to obtain the radial velocity of the current elevation angle after deblurring.

2. The method according to claim 1, characterized in that The good radial direction in step 2 is defined as follows: there is no strong shear between all adjacent effective velocity bins marked as 0 or 1 on a radial direction, wherein the no strong shear condition is: in, is the i-th effective radial velocity, is the i-1th effective radial velocity, =1, 2, … is the valid speed library number, is the adjustable coefficient, is the maximum unambiguous speed.

3. The method according to claim 1, characterized in that The conditional good radial in step 4 is defined as: the maximum value of the length of the continuous effective velocity library marked as 0 or 1 is greater than the set threshold and the maximum velocity value is less than Good radial, is the adjustable coefficient; The speed mean value calculation in step 4 includes: For radial conditions with good conditions, calculate the following average velocity 、 : ,in in, is the number of effective speed libraries, To satisfy The number of radial velocity libraries, is the adjustable coefficient, is the maximum unambiguous speed.

4. The method according to claim 3, characterized in that Step 5, setting constraints for the first initial reference radial search, performing the first initial reference radial search for the velocity library marked with the current elevation angle as 0 based on the calculated velocity mean and the constraints, and finding the initial reference radial, includes: Set three constraints for the first initial reference radial search, where the three constraints are: The first constraint condition is that at most two initial reference radials can be determined for the same elevation angle; The second constraint condition is that the minimum angle between the two initial reference radials at the same elevation angle is greater than 120°; The third constraint condition is that the minimum angle between the initial reference radial of the current elevation angle and the initial reference radial of the processed adjacent elevation angle is less than 30° or greater than 150°, and the initial reference radial of the lowest second elevation angle is an invalid value; Based on three constraints, three search methods are used to search for the first initial reference radial direction and find the initial reference radial direction. The first initial reference radial search is performed using three search methods to find the initial reference radial direction, including: The first search method is used to search for up to two initial reference radials; If the first search method fails to find two initial reference radials, the second search method is used to search for one or two initial reference radials, so that the sum of the initial reference radials found by the first search method and the initial reference radials found by the second search method is two; If the initial reference radial direction is not found by the first search method and the second search method, the third search method is used to search for the initial reference radial direction; If the initial reference radial is not found by any of the three search methods, check whether the current elevation radial corresponding to the azimuth of the adjacent elevation initial reference radial is a good radial. If so, use the good radial as the initial reference radial for the current elevation. Among them, only the initial reference radials searched by the first and second search methods can be used as the constrained initial reference radials of adjacent elevation angles, otherwise the constrained initial reference radials are invalid values.

5. The method according to claim 4, characterized in that The first search method is used to search for at most two initial reference radials, including: For the current elevation angle, search for continuous radial areas with good conditions, that is, search for continuous areas with good radial roots ≥ the set threshold, and search in this continuous area. The positive and negative conversion area of ​​the value is The larger radial is determined as the backup initial reference radial; If multiple If there are multiple spare initial reference radials corresponding to the positive and negative conversion zones of the value, the value corresponding to each spare initial reference radial is Sort from smallest to largest; If the constrained initial reference radial A0 is invalid, the first spare initial reference radial after sorting is determined as the first initial reference radial; if A0 is valid, the first spare initial reference radial that satisfies the third constraint is determined as the first initial reference radial, and the first radial that satisfies the second constraint and the third constraint is searched for among the remaining spare initial reference radials as the second initial reference radial, and the two searched initial reference radials are used as the constrained initial reference radials A0 and A1 of the next adjacent elevation angle; The second search method is used to search for one or two initial reference radials, including: In the radial with good conditions, search for all The radials adjacent to it on the left and right are both good radials as spare initial reference radials. If multiple spare initial reference radials are found, the corresponding radials of each spare initial reference radial are used. Sort from smallest to largest; If an initial reference radial is found using the first search method, then based on the second and third constraints and Minimum principle, determine the second initial reference radial; if the initial reference radial is not found using the first search method, use the second constraint and the third constraint and Based on the minimum principle, at most two radials are determined as the initial reference radials. The two initial reference radials determined are used as the constrained initial reference radials A0 and A1 of the next adjacent elevation angle. The third search method for searching the initial reference radial direction includes: If the constrained initial reference radial of the current elevation angle is invalid, search and The smallest good radial is used as the initial reference radial; otherwise, search for the best radial in the condition. , and the small angle with the initial reference radial A0, A1 is less than 30° and The smallest radial direction is used as the initial reference radial direction.

6. The method according to claim 3, characterized in that The step 6 is to perform defuzzification processing on the velocity library marked as 0 on the initial reference radial direction and the two radial directions adjacent to it on the left and right sides, which are searched in step 5, including: use or As a reference speed Defuzzify each velocity library marked as 0 on the initial reference radial direction searched in step 5; Use the velocity value marked as 1 on the initial reference radial direction as the reference velocity , perform azimuth deambiguation on each velocity library marked as 0 on the two radial directions adjacent to the initial reference radial direction; Among them, deblurring is achieved by adjusting the formula in Value search and reference speed The closest real speed ,in, is the original velocity of the observation, if , If it is an adjustable parameter, the value of the speed library is used Instead, its flag is set to 1, otherwise, the speed library is not processed and its flag remains 0.

7. The method according to claim 4, characterized in that Step 7 uses the initial reference radial and the two adjacent radials found in step 5 as initial reference radials for omnidirectional root-by-root radial deblurring, and performs three rounds of omnidirectional root-by-root radial deblurring on the velocity library with the current elevation angle marked as 0, including: The first round of full-scale root-by-root radial deblurring includes: Starting from the two radials to the left and right of each initial reference radial, deblur the radials 360° clockwise and counterclockwise. For each radial, process it first along the tangential direction and then along the radial direction. The reference velocity is the average of three consecutive adjacent velocity bins that have been processed and are marked as 1 and meet the no strong shear condition. The second round of all-round root-by-root radial deblurring includes: Starting from each initial reference radial, deblur the image along 360° clockwise and counterclockwise direction. For each radial, velocity deblur is performed using the horizontal neighborhood method, the vertical neighborhood method, and the radial-first-then-tangential processing method. Among them, the root-by-root radial velocity deblurring based on the horizontal neighborhood method includes: Starting from the initial reference radial direction, deblur the image radially in a clockwise and counterclockwise direction for 360°. For each radial direction, process it once from far to near and once from near to far. For each velocity bin marked as 0, check its 8 adjacent velocity bins. If the number of velocity bins marked as 1 is ≥ 3, use their average velocity value as the reference velocity and deblur the current velocity bin. Root-by-root radial velocity deblurring based on the vertical neighborhood method, including: Starting from the initial reference radial, deblur the image in 360° clockwise and counterclockwise directions. For each radial, process it once from far to near and once from near to far. For each velocity bin marked as 0, take the average value of the velocity bins marked as 1 in the 3×3 grid points of the adjacent low elevation angle centered on the current azimuth-slant range bin as the reference velocity, and deblur the current velocity bin. Root-by-root radial deblurring with radial-first followed by tangential processing includes: Starting from each initial reference radial direction, perform root-by-root deblurring in a clockwise and counterclockwise direction of 360°. For each radial direction, perform velocity deblurring three times, first along the radial direction and then along the tangential direction. The first reference velocity is the average of three consecutive adjacent velocity bins marked as 1 and meeting the no-strong-shear condition. The second reference velocity is the average of two consecutive adjacent velocity bins marked as 1 and meeting the no-strong-shear condition. The third reference velocity is the velocity value of the adjacent velocity bin marked as 1. The second round of all-round root-by-root radial deblurring is repeated once more. The γ threshold in was changed from 0.2 to 0.3; The third round of all-round root-by-root radial deblurring includes: Starting from each initial reference radial direction, deblur the image in 360° clockwise and counterclockwise directions. For each radial direction, the image is processed along the tangential direction first and then along the radial direction. The finite radial number thresholds are set for the tangential direction and radial direction respectively. and finite distance threshold Search reference speed within range; When two initial reference radials are found in step 5, steps 6 and 7 are performed twice, respectively, with the two initial reference radials found in step 5 as starting points.

8. The method according to claim 1, characterized in that Step 8, performing a second initial reference radial search on the velocity library with the current elevation angles marked as 0 and 1, determining the initial reference radial, and repeating steps 6 to 7 based on the determined initial reference radial, includes: After completing three rounds of omnidirectional root-by-root radial deblurring, a good radial search is performed on the velocity libraries currently labeled 0 and 1. If the number of good radials searched, GN, is less than the total number of radials, BN, the good radial with the largest number of valid velocity libraries and whose adjacent radials are both conditionally good is used as the initial reference radial, and steps 6 to 7 are repeated.

9. The method according to claim 1, characterized in that Step 9, performing singular value filtering on the velocity library with the current elevation angle marked as 0 to obtain the deblurred radial velocity of the current elevation angle, includes: For each speed bin marked as 0, the average value of the speeds in the eight adjacent speed bins marked as 1 is used to replace the value of the speed bin, and its flag is set to 1.

10. The method according to claim 1, characterized in that The step 1 includes inputting the radial velocity of the current elevation angle and constraining the initial reference radial direction, and then removing the isolated effective velocity library of the current elevation angle, i.e., the noise points: Take an M×N window centered on the current valid speed library and calculate the ratio of the invalid speed library in the window to the total number of grid points in the window. If the ratio is greater than or equal to the threshold ratio, the speed value of the current valid speed library is replaced with the invalid value.

Images