A tracking method of single convective cloud artificial rainfall enhancement using optical flow method

Abstract

The present invention belongs to the field of artificial rainfall enhancement technology, and in particular, is a tracking method for single convective cloud artificial rainfall enhancement using an optical flow method, comprising step 1: interpolating the intensity field data of the S-band Doppler weather radar of the detected area into a three-dimensional rectangular coordinate system. The optical flow method is used for tracking single convective cloud artificial rainfall enhancement, and the precipitation echo is tracked by setting the optical flow method. The significance test values ​​of the evaluation results of the maximum echo intensity, echo top height and rainfall rate of the entire layer calculated by the optical flow method are generally higher than the calculation results of the cross-correlation algorithm. When the optical flow method tracks the precipitation echo, the physical parameter estimation will not be overestimated or underestimated due to the appearance of redundant echo areas, thereby increasing the tracking accuracy, and solving the technical problem that the existing convective cloud artificial rainfall enhancement adopts a cross-correlation algorithm for tracking, and cannot track the merging or splitting of echoes, resulting in reduced tracking accuracy and a significant increase in tracking failures.

Description

Technical Field

[0001] The invention relates to the technical field of artificial rainfall enhancement, and in particular to a tracking method for single convective cloud artificial rainfall enhancement using an optical flow method. Background Art

[0002] The effect of artificial rainfall refers to the changes that occur in the cloud body after artificial catalysis. It contains two meanings: one is the changes in the microphysical processes in the cloud and the macroscopic and microscopic physical quantities in the cloud (such as cloud thickness, width, rising air velocity, cloud body temperature, ice crystal concentration, large cloud droplet concentration, etc.) after artificial influence, which is the direct effect or physical effect of artificial influence; the other is the changes in precipitation before and after the influence, which is the indirect effect of cloud seeding and the final result of a series of physical changes in the cloud. The vitality of artificial weather modification science depends on the effect of its application and its role in disaster reduction and disaster relief. Therefore, effect testing is one of the important links that cannot be avoided in artificial rainfall operations and occupies an important position in the entire operation project. Scientific and objective effect testing is of great significance to promoting the progress of artificial weather modification. It is very important and necessary to improve the theory and methods of cloud and precipitation disciplines, improve the scientific and technological level and economic benefits of artificial weather modification operations, and obtain the support of society and the public.

[0003] However, due to the uncertainty of the evaluation object and the huge natural variation of clouds and precipitation, it is very difficult to effectively test the rainfall enhancement effect in these variations, especially for the evaluation of the effect of artificial rainfall enhancement for a single convective cloud. In the existing evaluation of the effect of a single artificial rainfall enhancement operation, the traditional radar echo extrapolation tracking method is used: the cross-correlation algorithm is used for calculation. When tracking precipitation echoes, the cross-correlation algorithm cannot track the merging or splitting of echoes, resulting in reduced tracking accuracy and a significant increase in tracking failures. Summary of the invention

[0004] Based on the technical problem that the existing convective cloud artificial rainfall enhancement adopts a cross-correlation algorithm for tracking, which cannot track the merging or splitting of echoes, resulting in reduced tracking accuracy and a significant increase in tracking failures, the present invention proposes a tracking method for single convective cloud artificial rainfall enhancement using an optical flow method.

[0005] The present invention proposes a method for tracking single convective cloud artificial rainfall enhancement using an optical flow method, comprising the steps of: interpolating intensity field data of an S-band Doppler weather radar in a detected area into a three-dimensional rectangular coordinate system;

[0006] Step 2: Select the radar echo CAPPI data at the cloud base height and determine the moving target unit in the CAPPI data layer;

[0007] Step 3: Use the local smoothness constrained optical flow field calculation method to extrapolate and track the moving target unit, and record the echo characteristic parameter value of each moving target unit;

[0008] Step 4: Determine the influencing unit and the comparison unit in the mobile target unit, make a statistical analysis of the operation effect based on the echo characteristic parameter values ​​of the influencing unit and the comparison unit, and make a comparative analysis with the effect evaluation results based on Trec algorithm tracking.

[0009] Preferably, in step 1, the Cressman interpolation method is used for interpolation processing, and the interpolated data is 30 layers from 0.5km to 15km, with a layer of 0.5km every other layer, and the interpolation grid of each layer of CAPPI data is 1km×1km;

[0010] Through the above technical solution, the detected areas are Tanggu, Beijing and Qinhuangdao. By reading the intensity field data scanned by three new generation S-band Doppler weather radars in Tanggu, Beijing and Qinhuangdao, it is interpolated into a three-dimensional rectangular coordinate system.

[0011] Preferably, in step 2, the precision of precipitation is quantitatively measured by radar, and the formula for measuring precipitation by radar is:

[0012] Z=300R 1.4

[0013] In the formula, Z represents the reflectivity factor, unit: mm 6 / m 3 , the relationship between the reflectivity factor and dbZ is R represents the precipitation intensity, in mm / h;

[0014] Through the above technical solution, a major difficulty in using radar data for effect evaluation lies in the accuracy of radar quantitative measurement of precipitation. Since precipitation is related to many factors such as season, region, and precipitation nature, the estimated precipitation often has a large deviation from the actual precipitation. Even a small estimated precipitation deviation is still large compared with the effect of artificial rainmaking, so that the rainmaking effect is submerged in this deviation. However, according to Cunning's research conclusion: the raindrop spectrum distribution at the cloud base height of cumulus clouds is not affected by artificial catalytic operations, that is, the ZR relationship used to estimate precipitation is not affected by catalytic operations. Therefore, it can be considered that the error in radar measurement of precipitation will not affect the results of rainmaking effect evaluation, because this error exists in both catalytic and uncatalyzed clouds. Of course, this conclusion has two restrictions: one is for convective clouds, and the other refers to the ZR relationship at the cloud base height. Since the error in estimated precipitation does not affect the evaluation result of rainmaking effect, the radar quantitative measurement precipitation formula suitable for Tianjin area is no longer calculated separately, but the formula for measuring precipitation by the new generation of weather radar is directly used.

[0015] Preferably, in the step 3, the selection of the optical flow algorithm is: when Z(x, y, t) is the radar reflectivity value at the pixel point (x, y) at time t, after a time interval of dt, the reflectivity value of the corresponding pixel point is Z(x+dx, y+dy, t+dt), when dt→0, it is considered that the reflectivity values ​​of the two points are approximately unchanged, that is, Z(x+dx, y+dy, t+dt)=Z(x, y, t), if the reflectivity value changes with position and time, the left side of the above formula is Taylor expanded

[0016] Ignoring the second-order infinitesimal term E, we get

[0017]

[0018] Through the above technical solution, optical flow is the instantaneous speed of pixel movement of a space moving object on the observation imaging plane; the optical flow field refers to the surface movement of the image grayscale pattern. The optical flow method is applied to radar echo tracking.

[0019] Preferably, in step 3, when the moving speeds of the echo in the x and y directions are respectively The gradients of reflectivity values ​​in the x, y, and t directions are Substituting into (3) we get z x u+z y v+z t = 0, which is the optical flow constraint equation, where z x ,z y is the spatial gradient, z t is the time rate of change, (u, v) is the optical flow, and the optical flow of all points is the optical flow field.

[0020] Preferably, in step 3, when the optical flows of N points in a small area with P as the center are the same, and the motion between two frames of images can be approximately linear within a short time interval, different points in the area are given different weights, and the closer to point P, the higher the weight. The calculation of the optical flow can be converted into

[0021] In the above formula, it is a small area centered on P, W 2 (x) is the window function, representing the weight of each point in the region, and is the optical flow of point P;

[0022] Through the above technical scheme, the constraint methods for calculating the optical flow field include global smoothness constraint (Horn et al., 1981) and local smoothness constraint (Lucas et al., 1981). The global constraint method assumes that the optical flow satisfies certain constraints in the entire image area, and the local constraint method assumes that the optical flow satisfies certain conditions in a small area around a given point. In the single convective artificial rainfall operation effect evaluation method, a certain moving target unit is tracked, and the focus is on a small area. Therefore, the optical flow field calculation method with local smoothness constraint is adopted.

[0023] Preferably, in the step three,

[0024] make

[0025]

[0026]

[0027] Then the solution of the equation is:

[0028] in

[0029] Preferably, in step three, the change in the radar echo reflectivity between two adjacent time periods is relatively weak, so the radar image satisfies the optical flow constraint equation.

[0030] I x u+I y v+I t =0;

[0031] Preferably, in step 3, the first-order difference is used to replace the derivative of the reflectivity in each direction, then:

[0032]

[0033]

[0034]

[0035] Calculate the optical flow field.

[0036] Preferably, in step three, the control of the optical flow field quality is as follows: vectors whose deviation from the average vector size of a certain point and several surrounding points exceeds 5m / s and whose angle exceeds 25° are eliminated, and compared with the overall average value of the moving vector size of the entire echo area, individual vectors with a deviation of more than 15m / s are eliminated.

[0037] The beneficial effects of the present invention are:

[0038] 1. The optical flow method is set to track precipitation echoes. The significance test values ​​of the evaluation results of the maximum echo intensity of the entire layer, echo top height and rainfall rate calculated by the optical flow method are generally higher than those calculated by the cross-correlation algorithm. When the optical flow method tracks precipitation echoes, the physical parameter estimation will not be overestimated or underestimated due to the presence of redundant echo areas. Therefore, the results of evaluating the effects of these physical parameters as variables are more accurate, thereby increasing the accuracy of tracking and solving the technical problem that the existing convective cloud artificial rainfall enhancement adopts the cross-correlation algorithm for tracking, which cannot track the merging or splitting of echoes, resulting in reduced tracking accuracy and a significant increase in tracking failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of an operation site for a method for tracking single convective cloud artificial rainfall enhancement using an optical flow method proposed by the present invention;

[0040] Figure 2 The catalytic incident azimuth diagrams of three stations and four time periods in radar echoes of a tracking method for single convective cloud artificial rainfall enhancement using an optical flow method proposed by the present invention;

[0041] Figure 3 A radar position diagram of a tracking method for a single convective cloud artificial rainfall enhancement using an optical flow method proposed by the present invention;

[0042] Figure 4 A comparison diagram of precipitation echo tracking results in operation area 1 of a tracking method for single convective cloud artificial rainfall enhancement using an optical flow method proposed by the present invention;

[0043] Figure 5 This is a comparison diagram of precipitation echo tracking results in operation area 2 of a tracking method for single convective cloud artificial rainfall enhancement using an optical flow method proposed in the present invention. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0045] Reference Figure 1-5 , a tracking method for single convective cloud artificial rainfall enhancement using an optical flow method, comprising the steps of: interpolating the intensity field data of the S-band Doppler weather radar in the detected area into a three-dimensional rectangular coordinate system;

[0046] Furthermore, in step 1, the Cressman interpolation method was used for interpolation processing. The interpolated data were 30 layers from 0.5km to 15km, with a layer of 0.5km every other layer. The interpolation grid of each layer of CAPPI data was 1km×1km.

[0047] The inspected areas are Tanggu, Beijing and Qinhuangdao. The intensity field data scanned by three new-generation S-band Doppler weather radars in Tanggu, Beijing and Qinhuangdao are read and interpolated into a three-dimensional rectangular coordinate system.

[0048] Step 2: Select the radar echo CAPPI data at the cloud base height and determine the moving target unit in the CAPPI data layer;

[0049] Furthermore, in step 2, the precision of precipitation is quantitatively measured by radar, and the formula for measuring precipitation by radar is:

[0050] Z=300R 1.4

[0051] In the formula, Z represents the reflectivity factor, with the unit of mm6 / m3. The relationship between the reflectivity factor and dbZ is: R represents the precipitation intensity, in mm / h;

[0052] One of the major difficulties in using radar data for effect evaluation is the accuracy of radar quantitative measurement of precipitation. Since precipitation is related to many factors such as season, region, and precipitation nature, the estimated precipitation often has a large deviation from the actual precipitation. Even a small estimated precipitation deviation is still large compared with the effect of artificial rainmaking, so that the rainmaking effect is submerged in this deviation. However, according to Cunning's research conclusion: the raindrop spectrum distribution at the cloud base height of cumulus clouds is not affected by artificial catalytic operations, that is, the ZR relationship used to estimate precipitation is not affected by catalytic operations. Therefore, it can be considered that the error in radar measurement of precipitation will not affect the results of rainmaking effect evaluation, because this error exists in both catalytic and uncatalyzed clouds. Of course, this conclusion has two restrictions: one is for convective clouds, and the other refers to the ZR relationship at the cloud base height. Since the error in estimated precipitation does not affect the evaluation results of rainmaking effects, the radar quantitative measurement precipitation formula suitable for Tianjin area is no longer calculated separately, but the formula for measuring precipitation by the new generation of weather radar is directly used.

[0053] Step 3: Use the local smoothness constrained optical flow field calculation method to extrapolate and track the moving target unit, and record the echo characteristic parameter value of each moving target unit;

[0054] Furthermore, in step 3, the selection of the optical flow algorithm is: when Z(x, y, t) is the radar reflectivity value at the pixel point (x, y) at time t, after a time interval of dt, the reflectivity value of the corresponding pixel point is Z(x+dx, y+dy, t+dt). When dt→0, it is considered that the reflectivity values ​​of the two points are approximately unchanged, that is, Z(x+dx, y+dy, t+dt)=Z(x, y, t). If the reflectivity value changes with position and time, the Taylor expansion is performed on the left side of the above formula:

[0055] Ignoring the second-order infinitesimal term E, we get

[0056]

[0057] Optical flow is the instantaneous speed of the pixel movement of a moving object in space on the observation imaging plane; the optical flow field refers to the surface movement of the image grayscale pattern. The optical flow method is applied to radar echo tracking.

[0058] Furthermore, in step 3, when the echo moving speeds in the x and y directions are The gradients of reflectivity values ​​in the x, y, and t directions are Substituting into (3) we get z x u+z y v+z t = 0, which is the optical flow constraint equation, where z x ,z y is the spatial gradient, z t is the time rate of change, (u, v) is the optical flow, and the optical flow of all points is the optical flow field;

[0059] Furthermore, in step 3, when the optical flows of N points in a small area with P as the center are the same, and the motion between two frames of images can be approximated as linear in a short time interval, different points in the area are given different weights. The closer to point P, the higher the weight. The calculation of optical flow can be transformed into

[0060] In the above formula, it is a small area centered on P, W 2 (x) is the window function, representing the weight of each point in the region, and is the optical flow of point P;

[0061] The constraint methods for calculating the optical flow field include global smoothness constraint (Horn et al., 1981) and local smoothness constraint (Lucas et al., 1981). The global constraint method assumes that the optical flow satisfies certain constraints in the entire image area, and the local constraint method assumes that the optical flow satisfies certain conditions in a small area around a given point. The single convective artificial rainfall operation effect evaluation method tracks a certain moving target unit and focuses on a small area. Therefore, the optical flow field calculation method with local smoothness constraint is adopted.

[0062] Furthermore, in step three,

[0063] make

[0064]

[0065]

[0066] Then the solution of the equation is:

[0067] in

[0068] Furthermore, in step 3, the change in the radar echo reflectivity between two adjacent time periods is relatively weak, so the radar image satisfies the optical flow constraint equation;

[0069] I x u+I y v+I t =0;

[0070] Furthermore, in step 3, the first-order difference is used to replace the derivative of the reflectivity in each direction, then:

[0071]

[0072]

[0073]

[0074] Calculate the optical flow field;

[0075] Furthermore, in step three, the quality of the optical flow field is controlled as follows: vectors whose deviation from the average vector size of a certain point and several surrounding points exceeds 5m / s and whose angle exceeds 25° are eliminated, and compared with the overall average value of the moving vector size in the entire echo area, individual vectors with a deviation of more than 15m / s are eliminated.

[0076] Step 4: Determine the influencing unit and the comparison unit in the mobile target unit, perform statistical analysis on the operation effect according to the echo characteristic parameter values ​​of the influencing unit and the comparison unit, and perform comparative analysis with the effect evaluation result based on the Trec algorithm tracking. The embodiment of the comparative analysis is as follows:

[0077] The intensity field data of three Doppler weather radars in Tanggu, Beijing and Qinhuangdao were interpolated to obtain multi-layer CAPPI data. Then, the optical flow method was used to replace the cross-correlation algorithm in the original scheme. The mobile target units determined in the echo field of this layer were extrapolated and tracked, and the radar physical parameter values ​​of each target unit were recorded. The mobile target unit information obtained by the new tracking method was used to evaluate the effect of a single convective precipitation enhancement operation, and a comparative analysis was made with the results of the original evaluation scheme.

[0078] The optical flow method is added to the "Tianjin Weather Modification Effect Evaluation System" and applied in the effect evaluation work; through the analysis of the case on August 2, 2017:

[0079] Taking the rocket and antiaircraft artillery rainmaking operations carried out by three artillery stations in Jizhou District on August 2, 2017 as an example, (see Table 1 for specific operation information and the corresponding station locations for Figure 1 ) Analysis of ground operation information and rationality of catalytic operations:

[0080] Table 1 Ground operation information in Tianjin on August 2, 2017

[0081]

[0082]

[0083] To ensure the validity of the evaluation results, the rationality of the catalytic operation is first analyzed; Figure 2 These are the catalytic incident positions in the radar echo at three stations and four times. It can be seen from the figure that the precipitation echoes at the rocket and anti-aircraft gun incident positions at the Xilonghuyu Artillery Station and Luozhuangzi Artillery Station at three times are in the development stage, and the catalytic operation is relatively reasonable; the precipitation echoes at the rocket catalytic incident position at the Liangzhuangzi Artillery Station are in the dissipation stage, and the catalytic operation is unreasonable;

[0084] According to the results of rationality analysis, the catalytic operation of Liangzhuangzi Artillery Station is unreasonable, so the operation at that time will not be evaluated. The evaluation sites only include Xilonghuyu Artillery Station and Luozhuangzi Artillery Station. The operating areas are divided into two according to the locations of the two artillery stations and the operating periods: Xilonghuyu Artillery Station is operating area 1, and Luozhuangzi Artillery Station is operating area 2.

[0085] Through data and processing: Using the volume scanning data of three new generation S-band Doppler weather radars (CINRAD / SA) located in Tianjin Tanggu, Beijing and Qinhuangdao, the radar locations are as follows: Figure 3As shown in the figure, A represents Tanggu radar, B represents Beijing radar, and C represents Qinhuangdao radar. The effective detection radius of the three radars is 230km. In the VCP21 precipitation detection mode, the radar volume scan time interval is about 6 minutes, the lowest elevation angle is about 0.5° (0.4°-0.6°), the highest elevation angle is about 19.5°, and there are 9 elevation angles in total. This case selected the radar volume scan data of Tianjin, Beijing and Hebei from 06:00 (Beijing time) to 10:30 on August 2, 2017, read the intensity field data of three new generation S-band Doppler weather radars in Tanggu, Beijing and Qinhuangdao, and then interpolated them. The interpolated data is 20 layers from 0.5km to 15km, with a layer of 0.5km every 0.5km, and the interpolation grid of each layer of CAPPI data is 1km×1km. To ensure data validity, it is stipulated that the effective detection radius at an altitude of 0.5km is 95km, the effective detection radius at an altitude of 1km is 130km, the effective detection radius at an altitude of 1.5km is 160km, the effective detection radius at an altitude of 2km is 180km, the effective detection radius at an altitude of 2.5km is 210km, and the effective detection radius at other altitudes is 230km.

[0086] The steps for evaluating the effectiveness of a single operation based on the new generation weather radar data can be briefly summarized as follows:

[0087] (a) The optical flow method and cross-correlation algorithm are used to identify and track the moving target unit on the CAPPI map, and the precipitation rate of each unit, the maximum echo intensity on the CAPPI map of this layer, and the corresponding echo top height physical parameters are calculated;

[0088] Figure 4 The following is a comparison of precipitation echo tracking results in Operation Area 1. The tracking period is 07:30-09:30 on August 2, 2018. The left figure is the tracking result of the cross-correlation algorithm, and the right figure is the tracking result of the optical flow method. It can be seen from the figure that the tracking paths of the optical flow method and the cross-correlation algorithm are basically the same, but from the tracking results of the optical flow method, it can be seen that the initial echo splits into two echoes during the movement, while the echo tracked by the cross-correlation algorithm is always of a fixed size and cannot track the splitting of the precipitation echo;

[0089] Figure 5 This is a comparison of precipitation echo tracking results for Operation Area 2. The tracking period is 08:30-10:30 on August 2, 2018. The left picture shows the tracking result of the cross-correlation algorithm, and the right picture shows the tracking result of the optical flow method. The tracking paths of the optical flow method and the cross-correlation algorithm are basically the same, but it can be seen from the tracking results of the optical flow method that the initial echo area is shrinking, and the cross-correlation algorithm cannot track this information. Therefore, it will inevitably lead to overestimation or underestimation in the estimation of physical parameters, thus affecting the final evaluation results.

[0090] (b) determining the catalytic unit among all the identified and tracked mobile target units, i.e., the mobile target unit affected by the catalytic operation;

[0091] (c) determining a comparison unit among the remaining moving target units (except the catalytic unit);

[0092] (d) Conduct effectiveness evaluation.

[0093] Table 2 and Table 3 are the evaluation results of the cross-correlation algorithm and the optical flow method respectively:

[0094] Table 2 Cross-correlation algorithm evaluation results

[0095]

[0096] Table 3 Optical flow method evaluation results

[0097]

[0098] Note: TD1 represents the average value of the physical parameter of the catalytic unit before operation, TD2 represents the average value of the physical parameter of the catalytic unit after operation, CD1 represents the average value of the physical parameter of the comparison unit before operation, CD2 represents the average value of the physical parameter of the comparison unit before operation, the ratio calculation formula is: (TD2 / TD1) / (CD2 / CD1), P value is the significance test value.

[0099] It can be seen from the table that the significance test values ​​of the evaluation results of the maximum echo intensity of the entire layer, echo top height and rainfall rate calculated by the optical flow method are generally higher than those calculated by the cross-correlation algorithm. When the optical flow method and the cross-correlation algorithm are used to track precipitation echoes, the tracking paths of the two are basically the same, but the cross-correlation algorithm cannot track the merging or splitting of echoes. The optical flow method can make up for this defect and has higher tracking accuracy. Compared with the cross-correlation algorithm, the optical flow method will not be overestimated or underestimated in the estimation of physical parameters due to the appearance of redundant echo areas. Therefore, the results of using these physical parameters as variable effect evaluation are more accurate. The significance test values ​​of the effect evaluation results of the optical flow method are generally higher than those of the cross-correlation algorithm.

[0100] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for tracking optical flow clouds for single convective cloud artificial rainfall enhancement, characterized in that: include Step 1: interpolate the intensity field data of the S-band Doppler weather radar in the detected area into a three-dimensional rectangular coordinate system; Step 2: Select the radar echo CAPPI data of cloud base height and determine the moving target unit in the CAPPI data; Step 3: Use the local smoothness constrained optical flow field calculation method to extrapolate and track the moving target unit, and record the echo characteristic parameter value of each moving target unit; Step 4: Determine the influencing unit and the comparison unit in the mobile target unit, make a statistical analysis of the operation effect based on the echo characteristic parameter values ​​of the influencing unit and the comparison unit, and make a comparative analysis with the effect evaluation results based on Trec algorithm tracking.

2. The optical flow method for tracking single convective cloud artificial rainfall according to claim 1, characterized in that: In the step 1, the Cressman interpolation method is used for interpolation processing. The interpolated data is 30 layers from 0.5km to 15km, with a layer of 0.5km every other layer. The interpolation grid of each layer of CAPPI data is 1km×1km.

3. The optical flow method for tracking single convective cloud artificial rainfall according to claim 1, characterized in that: In the step 2, the precision of precipitation is quantitatively measured by radar, and the formula for measuring precipitation by radar is: , In the formula, Z represents the reflectivity factor, unit: mm 6 / m 3 , the relationship between the reflectivity factor and dbZ is , R represents precipitation intensity, unit is mm / h.

4. The method for tracking single convective cloud artificial rainfall enhancement using an optical flow method according to claim 1, characterized in that: In the step 3, the selection of the optical flow algorithm is: when Z (x, y, t) is the radar reflectivity value at the pixel point (x, y) at time t, after a time interval of dt, the reflectivity value of the corresponding pixel point is Z (x + dx, y + dy, t + dt), when dt → 0, it is considered that the reflectivity values ​​of the two points are approximately unchanged, that is, Z (x + dx, y + dy, t + dt) = Z (x, y, t), if the reflectivity value changes with position and time, then the left side of the above formula is Taylor expanded, omitting the second-order infinitesimal term E, and the result is , where Z represents the radar reflectivity value, x, y represent the coordinate values, and t represents the time.

5. The method for tracking single convective cloud artificial rainfall enhancement using an optical flow method according to claim 4, characterized in that: In step 3, when the moving speeds of the echo in the x and y directions are , , the gradients of reflectivity values ​​in the x, y, and t directions are , substituting into the formula we get , which is the optical flow constraint equation, where z x， z y is the spatial gradient, z t is the time rate of change, (u, v) is the optical flow velocity vector, and the optical flow of all points is the optical flow field.

6. The optical flow method for tracking single convective cloud artificial rainfall according to claim 5, characterized in that: In step 3, when the optical flows of N points in a small area with P as the center are the same, and the motion between two frames of images can be approximated as linear in a short time interval, different points in the area are given different weights. The closer to point P, the higher the weight. The calculation of optical flow can be converted into , where Ω is a small area centered on P, is the window function, representing the weight of each point in the region, is the optical flow at point P, u and v are the optical flow velocity vectors, I represents the radar reflectivity value Z in the above, x is the x-direction coordinate value, and t represents time.

7. The optical flow method for tracking single convective cloud artificial rainfall according to claim 6, characterized in that: In the step three, make , , , Then the solution of the equation is: ,in , , , Used to replace the formula on the right side of the equal sign and simplify the solution of the equation.

8. The optical flow method for tracking single convective cloud artificial rainfall according to claim 7, characterized in that: In the step 3, the change in the radar echo reflectivity between two adjacent time periods is relatively weak, so the radar image satisfies the optical flow constraint equation; .

9. The method for tracking single convective cloud artificial rainfall enhancement using an optical flow method according to claim 8, characterized in that: In step 3, the first-order difference is used to replace the derivative of the reflectivity in each direction, then: ， ， , calculate the optical flow field, where I x , I y , I t Represent the partial derivatives of reflectivity at x, y, and t, respectively. i , I j , I k Respectively represent the reflectivity sequence marks in different directions.

10. The optical flow method for tracking single convective cloud artificial rainfall according to claim 9, characterized in that: In step three, the quality of the optical flow field is controlled as follows: vectors whose deviation from the average vector size of a certain point and several surrounding points exceeds 5m / s and whose angle exceeds 25° are eliminated, and compared with the overall average value of the moving vector size of the entire echo area, individual vectors with deviations of more than 15m / s are eliminated.

Images