Improved alternating polarization weather radar multi-lag mixed spectral width estimation method

Abstract

The invention discloses an improved multi-lag mixed spectral width estimation method for an alternating polarization weather radar. The method can be applied to an AHV-mode dual-polarization weather radar carried on a high-altitude platform. And approaching and continuous observation of meteorological targets are realized. The method comprises the following steps: firstly, introducing a cross-correlation function, and carrying out least square Gaussian fitting on an autocorrelation function and the cross-correlation function to obtain a multi-lag estimator; besides, the multi-lag estimator is easy to be saturated under a larger spectral width, so that the traditional estimator and the multi-lag estimator are combined to form a hybrid estimator by setting a spectral width threshold. The method can be applied to an AHV-mode dual-polarization weather radar carried on a high-altitude platform, and realizes approaching and continuous observation of meteorological targets. According to the invention, the performance of the spectral width estimator under the conditions of low SNR and small spectral width is improved, a larger spectral width measurement range can be ensured, and the reliability of spectral width data is improved.

Description

Technical Field

[0001] The invention relates to the technical field of meteorological radars, and in particular to an improved method for estimating multi-lag mixed spectrum width of an alternating polarization weather radar. Background Art

[0002] The three basic moments measured by weather radar are reflectivity, velocity, and spectral width. The spectral width represents the degree of dispersion of the velocity of meteorological targets. Spectral width has the potential to improve the interpretation of meteorological radar data and is crucial for severe weather warnings. The traditional time-domain pulse pair (PPP) algorithm usually estimates the spectral width by the ratio of the autocorrelation function of lag 0 (lag-0) and lag 1 (lag-1). Since the autocorrelation function of lag-0 contains noise, the traditional algorithm is susceptible to noise interference; especially in the case of low signal-to-noise ratio (SNR), the impact is more serious. In order to reduce noise interference, the researchers introduced a higher lag autocorrelation function to estimate the spectral width, constructed a multi-lag estimator, and improved the estimation performance of the spectral width at low SNR and small spectral width.

[0003] Polarimetric weather radar acquires more polarization parameter information by transmitting and receiving horizontal (H) and vertical (V) polarized waves, such as differential reflectivity, differential phase, correlation coefficient, differential phase shift rate, etc., which is of great significance for improving the accuracy of early warning and rainfall estimation. The alternating transmission (AHV) mode can be used in weather radar to achieve dual polarization function. Among them, the single-channel alternating transmission and reception mode (ATAR) weather radar has the advantages of lightweight, low power consumption and low cost, and can be applied to high-altitude platforms (HAPS) with limited load and power consumption (such as near-space airships, near-space solar drones, etc.).

[0004] In AHV mode, the radar transmits and receives H and V polarized waves alternately, so the time interval of the co-polarized echoes is twice the pulse repetition time (PRT), that is, the equivalent time interval is 2*PRT, which is twice the time interval of the co-polarized echoes in single polarization mode. This results in that some lagging correlation functions cannot be obtained in AHV mode (such as the autocorrelation functions of lag-1 and lag-3). Therefore, in AHV mode, the spectrum width estimation algorithm based on multi-lag autocorrelation fails. In AHV mode, the traditional estimation algorithm uses lag-0 autocorrelation and lag-1 cross-correlation functions to estimate the spectrum width. At present, there is a lack of estimators with higher lag correlation functions, which limits the spectrum width estimation performance in AHV mode at low SNR and small spectrum width.

[0005] In order to improve the spectral width estimation performance of the AHV mode weather radar, it is necessary to introduce higher lag autocorrelation and cross-correlation to construct the spectral width estimator, so as to make full use of the echo information of different lags. In addition, as the lag number contained in the spectral width estimator increases, the spectral width tends to a constant value (i.e. saturation) more easily. Therefore, in order to obtain a larger spectral width measurement range and reliable spectral width data at the same time, it is necessary to comprehensively utilize the traditional spectral width estimator in the AHV mode and the spectral width estimator with high-order lag to construct a multi-lag hybrid spectral width estimator. Summary of the invention

[0006] In a dual-polarization weather radar using the AHV mode, the traditional algorithm uses the ratio of lag-0 autocorrelation and lag-1 cross-correlation to estimate the spectral width. When the SNR is low and the spectral width is small, the traditional estimation algorithm has a large estimation error. This hinders the application of spectral width data in severe weather warnings. The present invention provides an improved multi-lag mixed spectral width estimation method for alternating polarization weather radar. First, a higher lag correlation function is introduced, and a multi-lag spectral width estimator in the AHV mode is constructed, which can improve the estimation performance of the spectral width when the SNR is low and the spectral width is small; secondly, by combining different spectrum extenders, a larger spectral width measurement range is obtained. The main technical problem to be solved is: to improve the spectral width estimation performance of the dual-polarization weather radar in the AHV mode.

[0007] To solve the above technical problems, the technical solution of the present invention is:

[0008] An improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar includes:

[0009] Step 1, construction of multi-lag spectral width estimator: Based on the autocorrelation and cross-correlation functions of different lags, construct spectral width estimators with maximum lag numbers 1, 2 and 3, which are denoted as lag-1, lag-2 and lag-3 spectral width estimators respectively.

[0010] Step 2: Create a lookup table of the maximum number of lags for different SNRs, correlation coefficients, and spectral widths based on radar parameters: For a given SNR, correlation coefficient, and spectral width, calculate the deviation, standard deviation, and invalid value ratio of different spectral width estimators. Construct an evaluation function by weighting the absolute value of the deviation, standard deviation, and invalid value ratio. Based on the minimum principle of the evaluation function, determine the optimal single estimator for a given SNR, correlation coefficient, and spectral width, and record the maximum number of lags contained in the optimal estimator. Finally, traverse different SNRs, correlation coefficients, and spectral widths, repeat the above process in step 2, and finally obtain the maximum number of lags for the optimal spectral width estimator for different SNRs, correlation coefficients, and spectral widths, and establish a lookup table of the maximum number of lags.

[0011] Step 3: Determine the decision threshold of the saturation spectral width according to the radar parameters: Simulate the standard deviation of the lag-2 spectral width estimator for different spectral widths. As the spectral width increases, this standard deviation tends to a fixed value, which is denoted as SDTH(lag-2).

[0012] Step 4: Estimate the SNR and correlation coefficient based on the radar H and V polarization echo sequences. According to the multi-lag estimator in Step 1, calculate the spectral width values using the lag-1, lag-2, and lag-3 spectral width estimators respectively.

[0013] Step 5: Search the look-up table according to the estimated SNR, correlation coefficient, and spectral width value of the lag-2 estimator to determine the maximum lag number of the optimal estimator; and calculate the standard deviation of the lag-2 spectral width estimator, denoted as SD(lag-2).

[0014] Step 6: Construct a hybrid estimator based on the maximum lag number and the threshold SDTH(lag-2): If the maximum lag number of the optimal estimator is greater than 2, the result of the lag-3 spectral width estimator is adopted for the hybrid estimator; if the maximum lag number of the optimal estimator is equal to 2 and SD(lag-2) < SDTH(lag-2) threshold, the result of the lag-2 spectral width estimator is adopted for the hybrid estimator; in other cases, the result of the lag-1 spectral width estimator is adopted for the hybrid estimator.

[0015] Optionally, Step 1 includes:

[0016] For an AHV-mode weather radar, the radar alternately transmits and receives H and V polarization waves within each radial. Denote H(2i) and V(2i + 1) as the 2i-th HH (transmit H and receive H) polarization echo and the (2i + 1)-th VV (transmit V and receive V) polarization echo respectively. Assume the radar first transmits the H polarization wave, then the echo sequence of the radar within each radial is H(0), V(1), H(2), V(3),..., H(2i), V(2i + 1)..., V(M - 1), where M represents the number of pulse accumulations, so the range of i is 0 to (M - 2) / 2. Then the autocorrelation functions of the H and V channels are respectively:

[0017]

[0018] The cross-correlation functions of HV and VH are respectively:

[0019]

[0020] Wherein, R and C represent the autocorrelation and cross-correlation functions, respectively. The subscripts h and v represent the estimation of the autocorrelation function using the echo signals of the H and V channels, respectively. The subscripts hv and vh represent the estimation of the cross-correlation function using the echo signals of the H and V channels at the same time. The symbol * represents conjugate calculation. In the AHV mode, m and n represent different lag numbers, where m is a positive even number or 0, n is a positive odd number, and T represents the pulse repetition time.

[0021] The lag-1 spectral width estimator with a maximum lag number of 1 is used to estimate the spectral width:

[0022]

[0023] The lag-2 spectral width estimator with a maximum lag number of 2 is used to estimate the spectral width:

[0024]

[0025] The lag-3 spectral width estimator with a maximum lag number of 3 is used to estimate the spectral width:

[0026]

[0027] Where λ is the wavelength, N h and N v are the noise power of the H channel and the V channel, respectively, and the R h (0) and R v (0) is the noise power of the H and V channels.

[0028] The step 2 comprises:

[0029] Step 21, calculation of deviation, standard deviation and invalid value ratio:

[0030] According to the radar wavelength, PRT and pulse accumulation number, the fixed SNR, correlation coefficient and true value of the spectrum width are set, and K Monte Carlo simulations are performed to obtain the simulated echo sequence; the echo sequence is substituted into formulas (3), (4) and (5), then the lag-1, lag-2 and lag-3 estimators can output K groups of estimated spectrum widths. For each estimator, the deviation of its estimated spectrum width is:

[0031]

[0032] Where W represents the estimated spectrum width, W true is the true value of the spectrum width of the simulation input. Similarly, for each estimator, the standard deviation of its estimated spectrum width is:

[0033]

[0034] in, is the estimated average spectral width.

[0035] In addition, when the logarithmic operation (ln) in formulas (3), (4) and (5) is less than 0, complex spectrum width will appear in the spectrum width estimator. The number of complex spectrum widths in K Monte Carlo is counted, denoted as Z, and the proportion of invalid values ​​is:

[0036] PIE=Z / K (8)

[0037] Step 22, evaluation index construction:

[0038] The evaluation index can be constructed based on the weighted sum of the absolute value of the deviation, the standard deviation and the proportion of invalid values:

[0039] INDEX=a*|Bias|+b*SD+c*PIE (9)

[0040] Among them, a, b and c represent the weights of deviation, standard deviation and invalid value proportion respectively.

[0041] Step 23, lookup table construction:

[0042] For the fixed SNR, correlation coefficient and spectrum width, calculate the INDEX of lag-1, lag-2 and lag-3 estimators, the estimator corresponding to the minimum INDEX value is recorded as the optimal estimator, and the maximum lag number of the optimal estimator is recorded.

[0043] Assume that the range of SNR is 0~SNR max When the correlation coefficient is measured in the range of CC, the SNR is traversed at an interval of 0.1 dB, and the above step 2 is repeated; similarly, when the correlation coefficient is measured in the range of CC min ~CC max When the spectral width measurement range is 0~W, each correlation coefficient is traversed with a correlation coefficient interval of 0.01, and the operation in step 2 is repeated; max m / s, and repeat the operation in step 2 above for each spectrum width at intervals of 0.1m / s; a lookup table of the maximum number of lags of the optimal estimator for different SNRs, correlation coefficients, and spectrum widths can be obtained.

[0044] The step 3 comprises:

[0045] According to the radar wavelength, PRT and pulse accumulation number, a fixed SNR, correlation coefficient and spectral width are set, and K Monte Carlo simulations are performed to obtain the simulated echo sequence. The echo sequence is substituted into formula (4); then the lag-2 estimator can output K groups of estimated spectral widths, and the standard deviation of the estimated spectral width is:

[0046]

[0047] in, is the estimated average spectral width.

[0048] When the spectrum width measurement range is 0~W max m / s, traverse the spectrum width range at intervals of 0.1m / s, repeat the operation in step 3, and you can get the standard deviation of the lag-2 estimator at different spectrum widths. As the spectrum width increases, the standard deviation of the lag-2 estimator tends to a constant value, which is recorded as the threshold of the standard deviation of the lag-2 estimator, that is, SDTH(lag-2).

[0049] The step 4 comprises:

[0050] In AHV mode, based on the echo sequence in step 1, the estimated SNR is:

[0051]

[0052] N h is the noise power of the H channel, and the R of the echo in each radial area without meteorological target is h (0) is the noise power of the H channel.

[0053] According to the autocorrelation and cross-correlation function expressions in step 1, the correlation coefficient can be expressed as:

[0054]

[0055] where ρ HV (1) and ρ HV (2) are:

[0056]

[0057] Where N h and N v are the noise power of the H channel and the V channel, respectively, and the R h (0) and R v (0) is the noise power of the H and V channels.

[0058] In addition, according to formulas (3), (4) and (5) in step 1, the spectral widths estimated by the lag-1, lag-2 and lag-3 estimators are:

[0059]

[0060] The step 5 comprises:

[0061] Step 51, determine the maximum hysteresis number:

[0062] Search the look-up table based on the SNR estimated by formula (11) in step 4, the correlation coefficient estimated by formula (12), and the spectral width value of the lag-2 estimator in formula (16) to determine the maximum lag number Lag-max of the optimal estimator.

[0063] Step 52, calculate the standard deviation of the lag-2 estimator:

[0064] For the p-th range bin in a certain radial direction, when calculating the standard deviation of the lag-2 spectral width estimator for this range bin, use the spectral widths of (2l + 1) range bins, which are the l range bins before and after this range bin, to calculate the standard deviation, denoted as SD(lag-2), that is:

[0065]

[0066] The said step 6 includes:

[0067] Construct a hybrid estimator based on the maximum lag number Lag-max of the optimal estimator in step 51 and the threshold SDTH(lag-2) in step 3. If Lag-max > 2, the hybrid estimator adopts the result of the lag-3 spectral width estimator; if Lag-max = 2 and the standard deviation in step 52 is less than the threshold in step 3, that is, SD(lag-2) < SDTH(lag-2), the hybrid estimator adopts the result of the lag-2 spectral width estimator; in other cases, the hybrid estimator adopts the result of the lag-1 spectral width estimator.

[0068] The beneficial effects of the present invention are:

[0069] (1) Based on the multi-lag correlation function, this paper constructs a multi-lag spectral width estimator in the AHV mode, which can improve the estimation performance of low SNR and small spectral width;

[0070] (2) This paper constructs a hybrid estimator using multiple lag estimators (i.e., lag-1, lag-2, and lag-3 estimators), which can obtain a larger spectral width measurement range while ensuring the improvement of the small spectral width estimation performance.

[0071] (3) Applying the algorithm of this paper can obtain reliable spectral width data and improve the forecasting and early warning ability of the AHV mode weather radar for severe weather. Description of the Drawings

[0072] Figure 1 It is a schematic diagram of the hybrid spectral width estimation for the AHV mode dual-polarization weather radar in Embodiment 1 of the present invention;

[0073] Figure 2 It is a diagram of the hybrid spectral width estimation result in Embodiment 1 of the present invention;

[0074] Figure 3This is a result diagram of a traditional spectrum width estimator according to the first embodiment of the present invention;

[0075] Figure 4 (a) is a deviation diagram of the hybrid estimator and the traditional estimator at different SNRs according to the first embodiment of the present invention; (b) is a standard deviation diagram of the hybrid estimator and the traditional estimator at different SNRs according to the first embodiment of the present invention;

[0076] Figure 5 (a) is a deviation diagram of the hybrid estimator and the traditional estimator at different spectral widths according to the first embodiment of the present invention; (b) is a standard deviation diagram of the hybrid estimator and the traditional estimator at different spectral widths according to the first embodiment of the present invention. DETAILED DESCRIPTION

[0077] In order to make the purpose and technical solution of the present invention more clearly understood, the present invention is further described in detail below through specific implementation methods in combination with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0078] Embodiment 1:

[0079] In this embodiment, the Ku-band weather radar parameters and measured data of the AHV mode are used to verify the content of the present invention.

[0080] This embodiment provides a spectrum width estimation method for AHV mode dual-polarization weather radar. Figure 1 , specifically including the following steps:

[0081] Table 1

[0082]

[0083] Step 1 specifically includes:

[0084] For the AHV mode weather radar, the radar transmits and receives H and V polarized waves alternately in each radial direction. H(2i) and V(2i+1) are respectively the 2ith HH (transmit H and receive H) polarized echo and the 2i+1th VV (transmit V and receive V) polarized echo. Assuming that the radar transmits the H polarized wave first, the echo sequence of the radar in each radial direction is H(0), V(1), H(2), V(3), ..., H(2i), V(2i+1) ..., V(M-1), where M = 64 represents the pulse accumulation number, so the range of i is 0 to 31. For other radar parameters, refer to Table 1. Then the autocorrelation functions of the H and V channels are:

[0085]

[0086] The cross-correlation functions of HV and VH are:

[0087]

[0088] Where R and C represent the autocorrelation and cross-correlation functions, respectively. The subscripts h and v represent the estimation of the autocorrelation function using the echo signals of the H and V channels, respectively. The subscripts hv and vh represent the estimation of the cross-correlation function using the echo signals of the H and V channels at the same time. The symbol * represents conjugate calculation. In the AHV mode, m and n represent different lag numbers, where m is a positive even number or 0, and n is a positive odd number. T represents the pulse repetition time (PRT).

[0089] The lag-1 spectral width estimator with a maximum lag number of 1 is used to estimate the spectral width:

[0090]

[0091] The lag-2 spectral width estimator with a maximum lag number of 2 is used to estimate the spectral width:

[0092]

[0093] The lag-3 spectral width estimator with a maximum lag number of 3 is used to estimate the spectral width:

[0094]

[0095] Where N h and N v are the noise power of the H channel and the V channel, respectively, and the R h (0) and R v (0) is the noise power of the H and V channels.

[0096] Step 2 specifically includes:

[0097] Step 21, calculation of deviation, standard deviation and invalid value ratio:

[0098] According to the radar wavelength, PRT and pulse accumulation number, the fixed SNR, correlation coefficient and true value of the spectrum width are set, and 500 Monte Carlo simulations are performed to obtain the simulated echo sequence. Substituting the echo sequence into formulas (21), (22) and (23), the lag-1, lag-2 and lag-3 estimators can output 500 sets of estimated spectrum widths. For each estimator, the deviation of its estimated spectrum width is:

[0099]

[0100] Where W represents the estimated spectrum width, W true is the true value of the spectrum width of the simulation input. Similarly, for each estimator, the standard deviation of its estimated spectrum width is:

[0101]

[0102] in, is the estimated average spectral width.

[0103] In addition, when the logarithmic operation (ln) in formulas (21), (22) and (23) is less than 0, complex spectral width will appear in the spectral width estimator. The number of complex spectral widths in 500 Monte Carlo calculations is counted, denoted as Z, and the proportion of invalid values ​​is:

[0104] PIE=Z / 500 (26)

[0105] Step 22, evaluation index construction:

[0106] The evaluation index can be constructed based on the weighted sum of the absolute value of the deviation, the standard deviation and the proportion of invalid values:

[0107] INDEX=0.3*|Bias|+0.2*SD+0.5*PIE (27)

[0108] Among them, 0.3, 0.2 and 0.5 represent the weights of deviation, standard deviation and invalid value ratio respectively.

[0109] Step 23, lookup table construction:

[0110] For the fixed SNR, correlation coefficient and spectrum width, calculate the INDEX of lag-1, lag-2 and lag-3 estimators, the estimator corresponding to the minimum INDEX value is recorded as the optimal estimator, and the maximum lag number of the optimal estimator is recorded.

[0111] Assuming that the range of SNR is 0-20, then traverse each SNR with an SNR interval of 0.1dB, and repeat the specific operation of step 2 above; similarly, when the measurement range of the correlation coefficient is 0.90-0.99, traverse each correlation coefficient with a correlation coefficient interval of 0.01, and repeat the specific operation in step 2 above; when the spectrum width measurement range is 0-10m / s, traverse each spectrum width with an interval of 0.1m / s, and repeat the specific operation in step 2 above; a lookup table of the maximum lag number of the optimal estimator for different SNRs, correlation coefficients and spectrum widths can be obtained. As shown in Table 2:

[0112] Table 2 Maximum hysteresis number lookup table for different SNR, correlation coefficient and spectrum width

[0113]

[0114] Step 3 specifically includes:

[0115] According to the radar wavelength of 0.0223m, PRF = 1860Hz and pulse accumulation number 64, SNR = 20dB, correlation coefficient = 0.995 and spectrum width of 1m / s, 500 Monte Carlo simulations are performed to obtain the simulated echo sequence. Substituting the echo sequence into formula (25), the lag-2 estimator can output 500 sets of estimated spectrum widths, and the standard deviation of the estimated spectrum width is:

[0116]

[0117] in, is the estimated average spectral width.

[0118] When the spectrum width measurement range is 0-10m / s, the spectrum width range is traversed at intervals of 0.1m / s, and the operation in step 3 is repeated to obtain the standard deviation of the lag-2 estimator at different spectrum widths. As the spectrum width increases, the standard deviation of the lag-2 estimator tends to a constant value of 0.4, which is recorded as the threshold of the standard deviation of the lag-2 estimator, that is, SDTH(lag-2)=0.4.

[0119] Step 4 specifically includes:

[0120] In AHV mode, based on the echo sequence in step 1, the estimated SNR is:

[0121]

[0122] N h is the noise power of the H channel, and the R of the echo in each radial area without meteorological target is h (0) is the noise power of the H channel.

[0123] According to the autocorrelation and cross-correlation function expressions in step 1, the correlation coefficient can be expressed as:

[0124]

[0125] where ρ HV (1) and ρ HV (2) are:

[0126]

[0127] Where N h and N v are the noise power of the H channel and the V channel, respectively, and the R h (0) and R v (0) is the noise power of the H and V channels.

[0128] In addition, according to formulas (21), (22) and (23) in step 1, the spectral widths estimated by the lag-1, lag-2 and lag-3 estimators are:

[0129]

[0130] Step 5 specifically includes:

[0131] Step 51, determine the maximum hysteresis number:

[0132] According to the SNR estimated by formula (29) in step 4, the correlation coefficient estimated by formula (30) and the spectrum width value of the lag-2 estimator in formula (34), the lookup table is searched to determine the maximum lag number Lag-max of the optimal estimator;

[0133] Step 52, calculate the standard deviation of the lag-2 estimator:

[0134] For the pth distance bin in a certain radial direction, when calculating the standard deviation of the lag-2 spectral width estimator of the distance bin, the spectral widths of the 2 distance bins before and after the distance bin, a total of 5 distance bins, are used to calculate the standard deviation, which is recorded as SD(lag-2), that is:

[0135]

[0136] Step 6 includes:

[0137] The hybrid estimator is constructed according to the maximum lag number Lag-max of the optimal estimator in step 51 and the threshold SDTH(lag-2) in step 3. If Lag-max>2, the hybrid estimator adopts the result of the lag-3 spectral width estimator; if Lag-max=2 and the standard deviation in step 52 is less than the threshold in step 3, that is, SD(lag-2)<0.4, the hybrid estimator adopts the result of the lag-2 spectral width estimator; in other cases, the hybrid estimator adopts the result of the lag-1 spectral width estimator.

[0138] The algorithm proposed in this paper is verified by using a stratiform precipitation echo measured by Ku weather radar. Figure 2 This is a graph of the spectrum width estimation results of the traditional estimator (i.e., lag-1 spectrum width estimator). Figure 3 This is the spectrum width estimation result of the hybrid estimator. Figure 2 and Figure 3 Invalid values ​​of the spectral width are shown in white. Figure 2 and Figure 3 In the red circle area, Figure 3 There are fewer invalid values. Therefore, the use of a hybrid estimator can reduce the proportion of invalid values ​​in the spectrum width.

[0139] In addition, after removing the abnormal data, the correlation coefficients of the data set are all greater than 0.99. Therefore, we evaluated the bias and standard deviation of the hybrid estimator and the traditional estimator at different SNRs and spectral widths. Similarly, as in step 6, the 5 distance libraries are also used to calculate the bias and standard deviation of the estimated spectral width.

[0140] Figure 4 In (a)(b), T represents the traditional estimator and HY represents the hybrid estimator. The results show that compared with the traditional estimator, the hybrid estimator proposed in this paper can reduce the estimation bias and standard deviation of the spectrum width when the SNR is less than 10dB. Figure 5 (a)(b) shows that compared with the traditional estimator, the hybrid estimator proposed in this paper can reduce the estimation bias and standard deviation of the spectrum width when the spectrum width is less than 2m / s. In summary, the hybrid spectrum width estimator proposed in this paper can improve the estimation performance at low SNR and small spectrum width.

[0141] The above contents are further detailed descriptions of the present invention in combination with specific implementation methods, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. An improved method for estimating multi-lag mixed spectral width of alternating polarization weather radar, characterized in that: Including: Step 1, construction of multi-lag spectral width estimators: Based on the autocorrelation and cross-correlation functions with different lags, construct spectral width estimators with maximum lags of 1, 2, and 3, denoted as lag-1, lag-2, and lag-3 spectral width estimators respectively; Step 2, create a lookup table of the maximum lag number for different SNRs, correlation coefficients, and spectral widths according to the radar parameters: For a given SNR, correlation coefficient, and spectral width, calculate the bias, standard deviation, and proportion of invalid values of different spectral width estimators respectively; Step 3, determine the decision threshold for the saturated spectral width according to the radar parameters: Simulate the standard deviation of the lag-2 spectral width estimator for different spectral widths; Step 4, estimate the SNR and correlation coefficient according to the radar H and V polarization echo sequences, and calculate the spectral width values using the lag-1, lag-2, and lag-3 spectral width estimators respectively according to the multi-lag estimators in Step 1; Step 5, search the lookup table according to the estimated SNR, correlation coefficient, and spectral width value of the lag-2 estimator to determine the maximum lag number of the optimal estimator; and calculate the standard deviation of the lag-2 spectral width estimator, denoted as SD(lag-2); Step 6, construct a hybrid estimator according to the maximum lag number and the SDTH(lag-2) threshold: If the maximum lag number of the optimal estimator is greater than 2, the result of the lag-3 spectral width estimator is used for the hybrid estimator; if the maximum lag number of the optimal estimator is equal to 2 and SD(lag-2) < SDTH(lag-2) threshold, the result of the lag-2 spectral width estimator is used for the hybrid estimator; in other cases, the result of the lag-1 spectral width estimator is used for the hybrid estimator.

2. The improved method for estimating multi-lag mixed spectral width of alternating polarization weather radar according to claim 1, characterized in that: In the said Step 1, for the AHV mode weather radar, the radar alternately transmits and receives H and V polarization waves within each radial. Denote H(2i) and V(2i + 1) as the 2i-th HH (transmit H and receive H) polarization echo and the (2i + 1)-th VV (transmit V and receive V) polarization echo respectively; assuming the radar first transmits the H polarization wave, the echo sequence of the radar within each radial is: H(0), V(1), H(2), V(3),..., H(2i), V(2i + 1)..., V(M - 1) where M represents the number of pulse accumulations, so the range of i is 0 to (M - 2) / 2; the autocorrelation functions of the H and V channels are respectively: The cross-correlation functions of HV and VH are respectively: where R and C respectively represent the autocorrelation and cross-correlation functions, the subscripts h and v respectively represent estimating the autocorrelation function using the H and V channel echo signals, the subscripts hv and vh represent estimating the cross-correlation function using the echo signals of both the H and V channels simultaneously, the symbol * represents conjugate calculation. In the AHV mode, m and n represent different lag numbers, where m is a positive even number or 0, n is a positive odd number, and T represents the pulse repetition time.

3. The improved method for estimating multi-lag mixed spectral width of alternating polarization weather radar according to claim 1, characterized in that: In the said Step 1, the spectral width estimated by the lag-1 spectral width estimator with a maximum lag number of 1 is: The spectral width estimated by the lag-2 spectral width estimator with a maximum lag number of 2 is: The spectral width estimated by the lag-3 spectral width estimator with a maximum lag number of 3 is: Where λ is the wavelength, N h and N v are the noise power of the H channel and the V channel, respectively, and the R h (0) and R v (0) is the noise power of the H and V channels.

4. The improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar according to claim 1, characterized in that: In step 2, an evaluation index is constructed based on the weighted sum of the absolute value of the deviation, the standard deviation, and the proportion of invalid values: INDEX = a * Bias + b * SD + c * PIE where a, b, and c represent the weights of the deviation, the standard deviation, and the proportion of invalid values, respectively.

5. The improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar according to claim 1, characterized in that: In step 2, for the set fixed SNR, correlation coefficient, and spectral width, calculate the INDEX of the lag-1, lag-2, and lag-3 estimators. The estimator corresponding to the minimum INDEX is denoted as the optimal estimator, and the maximum lag number of the optimal estimator is recorded.

6. The improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar according to claim 1, characterized in that: In step 3, according to the radar wavelength, PRT, and number of pulse integrations, set fixed SNR, correlation coefficient, and spectral width, and perform K Monte Carlo simulations to obtain the simulated echo sequence; the lag-2 estimator can output K groups of estimated spectral widths, and the standard deviation of its estimated spectral width is: in, is the average value of the estimated spectral width.

7. The improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar according to claim 1, characterized in that: In step 3, when the spectral width measurement range is 0 to Wmax m / s, traverse the spectral width range at intervals of 0.1 m / s, and repeat the operations in step 3. The standard deviation of the lag-2 estimator at different spectral widths can be obtained; as the spectral width increases, the standard deviation of the lag-2 estimator tends to a fixed value, and this fixed value is denoted as the threshold of the standard deviation of the lag-2 estimator, i.e., SDTH(lag-2).

8. The improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar according to claim 1, characterized in that: In step 4, in the AHV mode, according to the echo sequence in step 1, the estimated SNR is: Among them, N h is the noise power of the H channel, and the R of the echo in each radial area without meteorological target is h (0) is the noise power of the H channel.

9. The improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar according to claim 1, characterized in that: In step 4, the spectral widths estimated by the lag-1, lag-2, and lag-3 estimators are:

10. The improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar according to claim 1, characterized in that: In step 5, the method for calculating the standard deviation of the lag-2 estimator is as follows: for the p-th range bin in a certain radial direction, when calculating the standard deviation of the lag-2 spectral width estimator for this range bin, use the spectral widths of a total of (2l + 1) range bins, including l range bins before and after this range bin, to calculate the standard deviation, denoted as SD(lag-2), i.e.:

11. The improved method for estimating multi-lag mixed spectrum width of alternating polarization weather radar according to claim 1, characterized in that: In step 6, if Lag-max > 2, the result of the lag-3 spectral width estimator is used for the hybrid estimator; if Lag-max = 2 and the standard deviation in step 52 is less than the threshold in step 3, i.e., SD(lag-2) < SDTH(lag-2), the result of the lag-2 spectral width estimator is used for the hybrid estimator; in other cases, the result of the lag-1 spectral width estimator is used for the hybrid estimator.

Images