Weather radar signal processing device and weather radar signal processing program
The weather radar signal processing device and program address the challenges of clutter removal and beam resolution by applying a window function and adjusting Doppler velocity widths, enhancing clutter removal and rainfall recovery while reducing computational burden.
Patent Information
- Application Number
- JP2023082291
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2023-05-18
- Publication Date
- 2026-11-11
- Estimated Expiration
- 2043-05-18
AI Technical Summary
Existing weather radar signal processing techniques face challenges in effectively removing surface clutter components while maintaining azimuth beam resolution and computational efficiency, leading to incomplete or excessive removal of ground and tree clutter, and irrecoverable rainfall signal components.
A weather radar signal processing device and program that applies a window function to time-series data, adjusts Doppler velocity widths based on observation points, and stores filtering characteristics prior to observation, to enhance clutter removal and rainfall signal recovery, reducing computational burden.
Achieves effective removal of surface clutter, improves azimuth beam resolution, reduces excessive or incomplete clutter removal, and minimizes irrecoverable rainfall signal components, while lowering computational processing load during weather observations.
Smart Images

Figure 0007947697000017 
Figure 0007947697000018 
Figure 0007947697000019
Abstract
Description
[Technical Field]
[0001] This disclosure relates to a technique for removing surface clutter components from weather radar signals. [Background technology]
[0002] Two techniques are known for removing surface clutter components from weather radar signals: a technique that performs frequency-domain filtering (MTI: Moving Target Indicator) and a technique that performs time-domain filtering (GMAP-TD: Gaussian Model Adaptive Processing in Time Domain). GMAP-TD is disclosed in Non-Patent Document 1.
[0003] In MTI processing, the time-series data signal is Fourier transformed into a Doppler velocity spectrum, removing the ground surface clutter component near a Doppler velocity of 0 and extracting the rainfall signal component with a finite Doppler velocity. However, applying an infinite-length data approximation and performing the Fourier transform degrades the Doppler velocity spectrum. Furthermore, even if the main lobe of the ground surface clutter component is removed, if the side lobes of the ground surface clutter component remain, it becomes difficult to extract the rainfall signal component. Moreover, if the MTI processing is performed after multiplying by a window function, the side lobes of the ground surface clutter component are suppressed, but the rainfall signal component is also suppressed, thus reducing the gain of the MTI processing.
[0004] In the GMAP-TD processing of Non-Patent Document 1, the autocorrelation matrix of the time-series data signal is multiplied by a filtering matrix to remove the ground surface clutter component near zero Doppler velocity, and extract the precipitation signal component with a finite Doppler velocity. Here, the Doppler velocity spectrum of the ground surface clutter component is set to a Gaussian that spreads its tail from the main lobe to the side lobe, and the filtering matrix is calculated. Accordingly, the problems of MTI processing (deterioration of the Doppler velocity spectrum, difficulty in extracting the precipitation signal component, and reduction in the gain of MTI processing) are resolved. However, if the ground surface clutter component near zero Doppler velocity is removed, part of the precipitation signal component near zero Doppler velocity will also be removed. Therefore, the Doppler velocity spectrum of part of the precipitation signal component with a finite Doppler velocity is fitted with a Gaussian whose distribution center is the average Doppler velocity, to recover the part of the precipitation signal component near zero Doppler velocity. [Prior Art Documents] [Non-Patent Documents]
[0005] [Non-Patent Document 1] Cuong M. Nguyen and V. Chandrasekar, "Gaussian Model Adaptive Processing in Time Domain (GMAP-TD) for Weather Radars", Journal of Atmospheric and Oceanic Technology, American Meteorological Society, November 2013, Volume 30, pp.2571~2584. [Non-Patent Document 2] Sebastian Torres and Christopher Curtis, "Design considerations for improved tornado detection using super-resolution data on the NEXRAD network", Preprints, Third European Conference on Radar Meteorology and Hydrology (ERAD), Barcelona, Spain, Copernicus. 2006. SUMMARY OF THE INVENTION PROBLEMS TO BE SOLVED BY THE INVENTION
[0006] In the beam resolution enhancement processing of Non-Patent Document 2, a window function is applied to time-series data of weather radar signals in a weather radar apparatus equipped with an antenna that rotates in the azimuth direction.
[0007] However, if the beam resolution enhancement processing of Non-Patent Document 2 is applied in addition to the GMAP-TD processing of Non-Patent Document 1, the ground surface clutter component cannot be properly removed, and it is impossible to achieve both the effect of removing the ground surface clutter component and the effect of improving the azimuth beam resolution.
[0008] In the GMAP-TD processing of Non-Patent Document 1, the Doppler velocity width of the ground surface clutter component is set to a fixed width when calculating the filtering matrix.
[0009] Therefore, when removing ground clutter components that inherently have a small Doppler velocity width, if the fixed Doppler velocity width is too wide, components other than the ground clutter components may be excessively removed. Further, when removing tree clutter components that inherently have a large Doppler velocity width, if the fixed Doppler velocity width is too narrow, the tree clutter components may not be completely removed.
[0010] Furthermore, when extracting the rainfall signal component where the Doppler velocity near 0 is the average Doppler velocity, if the Doppler velocity range of the rainfall signal component is narrow and the fixed Doppler velocity range is too wide, there is a possibility that the rainfall signal component may be excessively removed, potentially leading to irrecoverable rainfall signal components. Also, when extracting the rainfall signal component where the Doppler velocity near 0 is the average Doppler velocity, if the Doppler velocity range of the ground surface clutter component is wide and the fixed Doppler velocity range is too narrow, there is a possibility that the remaining ground surface clutter component may be extracted as a spurious rainfall signal component, potentially leading to excessive recovery of the rainfall signal component.
[0011] In the GMAP-TD processing described in Non-Patent Document 1, a filtering matrix is calculated during meteorological observations. Here, the calculation of the filtering matrix includes the calculation of the inverse matrix, as described later in Equation 14. Therefore, the computational burden increases during meteorological observations.
[0012] Therefore, in order to solve the aforementioned problems, this disclosure aims to (1) achieve both the effect of removing the ground surface clutter component and the effect of improving the azimuth beam resolution when performing time-domain filtering on weather radar signals, (2) reduce the possibility of excessive or incomplete removal of ground surface clutter, and reduce the possibility of the rainfall signal component becoming unrecoverable or over-recovered, and (3) reduce the computational processing burden during weather observation. [Means for solving the problem]
[0013] To solve the aforementioned problem, a window function is applied to the time-series data of weather radar signals, and the same window function is also applied to the time-domain filtering process.
[0014] Specifically, the present disclosure is a weather radar signal processing device characterized by comprising: a weather radar signal acquisition unit that acquires time-series data of a weather radar signal to which a window function is applied from a weather radar device equipped with an antenna that rotates in the azimuth direction; a ground clutter removal unit that removes a ground clutter component from the time-series data of the weather radar signal to which the window function is applied and extracts a rainfall signal component; and a rainfall signal recovery unit that recovers the rainfall signal component removed by the ground clutter removal unit along with the ground clutter component by calculating the intensity distribution of received power with respect to the Doppler velocity of the rainfall signal component extracted by the ground clutter removal unit based on statistical properties.
[0015] Furthermore, this disclosure is a weather radar signal processing program that causes a computer to sequentially execute the following steps: a weather radar signal acquisition step of acquiring time-series data of a weather radar signal to which a window function is applied from a weather radar device equipped with an antenna that rotates in the azimuth direction; a ground clutter removal step of removing a ground clutter component from the time-series data of the weather radar signal to which the window function is applied and extracting a rainfall signal component; and a rainfall signal recovery step of recovering the rainfall signal component that was removed along with the ground clutter component in the ground clutter removal step by calculating the intensity distribution of received power with respect to the Doppler velocity of the rainfall signal component extracted in the ground clutter removal step based on statistical properties.
[0016] Applying a window function to time-series data of weather radar signals changes the statistical properties of the time-series data of the weather radar signals, such as the Gaussian. Therefore, as described in Non-Patent Documents 1 and 2, if the same window function is not applied to the time-domain filtering process, the ground surface clutter component cannot be properly removed, and it is not possible to achieve both the effect of removing the ground surface clutter component and the effect of improving the azimuth beam resolution. However, as described in this disclosure, if the same window function is applied to the time-domain filtering process, the ground surface clutter component can be properly removed, and it is possible to achieve both the effect of removing the ground surface clutter component and the effect of improving the azimuth beam resolution.
[0017] Furthermore, this disclosure relates to a weather radar signal processing device characterized in that the ground surface clutter removal unit sets the Doppler velocity width of the ground surface clutter component to be set to a variable width according to the weather observation point in the time domain filtering process to which the window function applied to the time series data of the weather radar signal is applied.
[0018] With this configuration, when removing the ground clutter component, which inherently has a small Doppler velocity range, the possibility of extra components other than the ground clutter component being removed can be reduced by appropriately narrowing the variable Doppler velocity range. Furthermore, when removing the tree clutter component, which inherently has a large Doppler velocity range, the possibility of the tree clutter component not being completely removed can be reduced by appropriately widening the variable Doppler velocity range.
[0019] Furthermore, when extracting a rainfall signal component where the Doppler velocity near 0 is the average Doppler velocity, even when the Doppler velocity range of the rainfall signal component is narrow, appropriately narrowing the variable Doppler velocity range can reduce the possibility of excessive removal of the rainfall signal component, thereby reducing the possibility of the rainfall signal component becoming irrecoverable. Also, when extracting a rainfall signal component where the Doppler velocity range near 0 is the average Doppler velocity, even when the Doppler velocity range of the ground surface clutter component is wide, appropriately widening the variable Doppler velocity range can reduce the possibility of the remaining ground surface clutter component being extracted as a spurious rainfall signal component, thereby reducing the possibility of excessive recovery of the rainfall signal component.
[0020] Furthermore, this disclosure provides a weather radar signal processing device characterized in that, in the time domain filtering process to which the window function applied to the time series data of the weather radar signal is applied, the ground surface clutter removal unit sets the Doppler velocity width of the ground surface clutter component to be set to a fixed width regardless of the weather observation point as an initial process, and the weather radar signal processing device further comprises a discontinuity point extraction unit that extracts discontinuity observation points in which at least one of the received power, average Doppler velocity, and Doppler velocity width of the rainfall signal component recovered by the rainfall signal recovery unit has a discontinuity within the weather observation space, and the ground surface clutter removal unit changes the Doppler velocity width of the ground surface clutter component to be set to a variable width different from the fixed width at the discontinuity observation points extracted by the discontinuity point extraction unit as a post-process.
[0021] According to this configuration, meteorological observation points where the parameters of the rainfall signal component, which should normally be continuous, are incorrectly discontinuous (discontinuous observation points) are considered as meteorological observation points where the rainfall signal component is irrecoverable or excessively recovered. Therefore, it is possible to distinguish between meteorological observation points where the rainfall signal component is irrecoverable and meteorological observation points where the rainfall signal component is normally low. Furthermore, it is possible to distinguish between meteorological observation points where the rainfall signal component is excessively recovered and meteorological observation points where the rainfall signal component is normally high.
[0022] Furthermore, at most meteorological observation points (continuous observation points) where the rainfall signal component can be recovered, the variable Doppler velocity range is set to a fixed range only. At some meteorological observation points (discontinuous observation points) where the rainfall signal component cannot be recovered or is recovered excessively, the variable Doppler velocity range is changed to a variable range. Thus, the processing burden of setting the variable Doppler velocity range can be reduced.
[0023] Furthermore, this disclosure provides a weather radar signal processing device characterized in that the discontinuity point extraction unit extracts as discontinuity points weather observation points weather observation points where the Doppler velocity average of the rainfall signal component recovered by the rainfall signal recovery unit has a sign inverted compared to surrounding weather observation points, and where the received power of the rainfall signal component recovered by the rainfall signal recovery unit or the power difference before and after recovery is smaller than a predetermined difference compared to surrounding weather observation points, and the ground surface clutter removal unit, in the filtering process of the time domain to which the window function applied to the time series data of the weather radar signal is applied, changes the Doppler velocity width of the ground surface clutter component to be set as a post-processing step to the variable width which is narrower than the fixed width at the discontinuity point extracted by the discontinuity point extraction unit.
[0024] With this configuration, when extracting the rainfall signal component whose Doppler velocity near 0 is the average Doppler velocity, even when the Doppler velocity range of the rainfall signal component is narrow, the possibility of extra rainfall signal components being removed can be reduced by appropriately narrowing the variable Doppler velocity range, thereby reducing the possibility of the rainfall signal component becoming irrecoverable.
[0025] Furthermore, this disclosure provides a weather radar signal processing device characterized in that the discontinuity point extraction unit extracts as discontinuity points weather observation points weather observation points where the Doppler velocity average of the rainfall signal component recovered by the rainfall signal recovery unit has a sign inverted compared to surrounding weather observation points, and where the received power of the rainfall signal component recovered by the rainfall signal recovery unit or the power difference before and after recovery is greater than a predetermined difference compared to surrounding weather observation points, and the ground surface clutter removal unit, in the filtering process of the time domain to which the window function applied to the time series data of the weather radar signal is applied, changes the Doppler velocity width of the ground surface clutter component to be set as a variable width that is wider than the fixed width at the discontinuity point extracted by the discontinuity point extraction unit as a post-processing step.
[0026] With this configuration, when extracting the rainfall signal component whose Doppler velocity is near 0 as the Doppler velocity average, even when the Doppler velocity range of the ground surface clutter component is wide, by appropriately widening the variable Doppler velocity range, the possibility of the remaining ground surface clutter component being extracted as a spurious rainfall signal component can be reduced, and the possibility of excessive recovery of the rainfall signal component can be reduced.
[0027] Furthermore, this disclosure relates to a weather radar signal processing device characterized in that the ground surface clutter removal unit stores various filtering characteristics in the time domain to which the window function applied to the time-series data of the weather radar signal is applied, in accordance with various received intensities and Doppler velocity widths of the ground surface clutter component, prior to weather observation.
[0028] With this configuration, since the filtering matrix is stored before meteorological observation, the computational burden during meteorological observation can be reduced, even when the calculation of filtering characteristics involves the calculation of the inverse matrix, as described later in Equation 14.
[0029] Furthermore, the inventions disclosed above can be combined as much as possible. [Effects of the Invention]
[0030] Thus, when performing time-domain filtering on weather radar signals, this disclosure can (1) achieve both the effect of removing surface clutter components and the effect of improving azimuth beam resolution, (2) reduce the possibility of excessive or incomplete removal of surface clutter, reduce the possibility of irrecoverable or excessive recovery of rainfall signal components, and (3) reduce the computational processing burden during weather observation. [Brief explanation of the drawing]
[0031] [Figure 1] This figure shows the challenges of conventional beam resolution improvement techniques. [Figure 2] This figure shows an overview of conventional beam resolution improvement techniques. [Figure 3] This figure shows the configuration of the weather radar signal processing device described herein. [Figure 4] This figure shows the procedure for removing ground surface clutter according to the present disclosure. [Figure 5] This figure shows an overview of the surface clutter removal process described herein. [Figure 6] This figure shows the challenges of the filtering characteristics calculation process in this disclosure. [Figure 7] This figure shows the procedure for the filtering characteristics storage processing described herein. [Figure 8] This figure shows the effect of the filtering characteristics calculation process in this disclosure. [Figure 9] This figure shows the effect of the filtering characteristics calculation process in this disclosure. [Figure 10] This figure shows an overview of the ground surface clutter removal range setting process in this disclosure. [Figure 11] This figure shows the procedure for setting the ground surface clutter removal range according to the present disclosure. [Modes for carrying out the invention]
[0032] Embodiments of the present disclosure will be described with reference to the attached drawings. The embodiments described below are examples of the implementation of the present disclosure, and the present disclosure is not limited to these embodiments.
[0033] (Conventional beam resolution improvement processing) Figures 1 and 2 show the challenges and overview of conventional beam resolution improvement techniques. In the left column of Figures 1 and 2, the weather radar system is equipped with an antenna A that rotates in the azimuth direction. The time-series data of the weather radar signal at a certain weather observation point is composed of multiple beams B that rotate in the azimuth direction. In the right column of Figures 1 and 2, the signal-to-noise ratio (SNR) of the time-series data of the weather radar signal at the same weather observation point is improved by data averaging (coherent integration).
[0034] In the right column of Figure 1, the time-series data of the weather radar signal at the weather observation point is only subjected to the window function of a rectangular window. Therefore, the weights of beam B near both ends become equal to the weights of beam B other than the ends, and the effective beam width of beam B' deteriorates significantly compared to the original beam width of beam B, thus failing to achieve an improvement in azimuth beam resolution.
[0035] In the right column of Figure 2, a window function such as a Hamming window is applied to the time-series data of the weather radar signal at the weather observation point. As a result, the weight of beam B near both ends becomes smaller than the weight of beam B other than the ends, and the effective beam width of beam B'' does not deteriorate much compared to the beam width of the original beam B, thus improving the azimuth beam resolution.
[0036] (Ground surface clutter removal treatment as disclosed herein) The configuration of the weather radar signal processing device of this disclosure is shown in Figure 3. The weather radar signal processing device R comprises a ground surface clutter removal device 1, a rainfall signal recovery device 2, and a discontinuity point extraction device 3. The weather radar signal processing device R can be realized by installing the weather radar signal processing programs shown in Figures 4, 7, and 11 onto a computer.
[0037] The ground surface clutter removal device 1 (described in Figure 4 and later) comprises an autocorrelation matrix calculation unit 11, a ground surface clutter distribution calculation unit 12, a filtering characteristic calculation / storage unit 13, and a ground surface clutter removal unit 14. The rainfall signal recovery device 2 (described in Figure 4 and later) comprises a rainfall signal distribution calculation unit 21, an autocorrelation matrix calculation unit 22, a filtering processing unit 23, a subtraction unit 24, an addition unit 25, and a rainfall signal convergence determination unit 26. The discontinuity point extraction device 3 (described in Figure 10 and later) comprises a rainfall signal distribution acquisition unit 31, a received power extraction unit 32, a Doppler velocity average extraction unit 33, a Doppler velocity width extraction unit 34, a discontinuity point extraction unit 35, and a filtering characteristic modification unit 36.
[0038] The procedure and overview of the ground surface clutter removal treatment described herein are shown in Figures 4 and 5.
[0039] The ground surface clutter removal device 1 extracts the rainfall signal component by removing the ground surface clutter component from the time-series data of the weather radar signal to which a window function has been applied (see the right column of Figure 2). In this process, it performs time-domain filtering, and sets the Doppler velocity width of the ground surface clutter component to a variable width according to the weather observation point.
[0040] The ground surface clutter removal device 1 receives the time-series radar signal after applying the window function at each meteorological observation point and before removing the ground surface clutter (step S1). As shown in the upper left column of Figure 5, the Doppler velocity spectrum at the input stage of step S1 includes the ground surface clutter component with a Doppler velocity near 0, the rainfall signal component with a finite Doppler velocity, and the noise component.
[0041] The autocorrelation matrix calculation unit 11 calculates an autocorrelation matrix for the time-series radar signal from step S1 (step S2). The ground surface clutter distribution calculation unit 12 calculates the ground surface clutter distribution for the time-series radar signal from step S1 (step S3). Specifically, the ground surface clutter distribution calculation unit 12 performs a Fourier transform on the time-series radar signal from step S1 to obtain a Doppler velocity spectrum and calculates the received intensity and Doppler velocity width of the ground surface clutter distribution.
[0042] The filtering characteristic calculation / storage unit 13 calculates a filtering matrix according to the received intensity and Doppler velocity width of the ground surface clutter distribution in step S3 (step S4). The ground surface clutter removal unit 14 multiplies the autocorrelation matrix in step S2 by the filtering matrix in step S4 (step S5). As shown in the middle left column of Figure 5, the Doppler velocity spectrum at the output stage in step S5 includes some rainfall signal components with finite Doppler velocities and noise components, and some rainfall signal components near a Doppler velocity of 0 are removed.
[0043] The rainfall signal recovery device 2 recovers the rainfall signal component that the ground surface clutter removal device 1 removed along with the ground surface clutter component by calculating the intensity distribution of the received power against the Doppler velocity of the rainfall signal component extracted by the ground surface clutter removal device 1 based on statistical properties.
[0044] The rainfall signal distribution calculation unit 21 calculates the rainfall signal distribution based on the output result of step S5 (step S6). Specifically, the rainfall signal distribution calculation unit 21 calculates the received intensity, average Doppler velocity, and Doppler velocity width of the rainfall signal distribution based on the fact that the output result of step S5 is the autocorrelation matrix of the time-series radar signal after surface clutter removal.
[0045] The rainfall signal distribution calculation unit 21 generates a time-series radar signal based on the rainfall signal distribution in step S6 (step S7). Specifically, the rainfall signal distribution calculation unit 21 generates the time-series radar signal based on the fact that a Gaussian model can be applied to the rainfall signal distribution in step S6. As shown in the lower left column of Figure 5, the Doppler velocity spectrum at the output stage of step S7 includes rainfall signal components with a Doppler velocity near 0, rainfall signal components with a finite Doppler velocity, and noise components. The autocorrelation matrix calculation unit 22 calculates an autocorrelation matrix for the time-series radar signal in step S7 (step S8).
[0046] The filtering processing unit 23 multiplies the autocorrelation matrix from step S8 by the filtering matrix from step S4 (step S9). As shown in the lower right column of Figure 5, the Doppler velocity spectrum at the output stage of step S9 includes some rainfall signal components with finite Doppler velocities and noise components, while some rainfall signal components near a Doppler velocity of 0 are removed. In other words, step S9 is a step that prepares for the recovery of the rainfall signal components.
[0047] The subtraction unit 24 subtracts the output result of step S9 from the autocorrelation matrix of step S8 (step S10). As shown in the middle right column of Figure 5, the Doppler velocity spectrum at the output stage of step S10 includes some rainfall signal components near a Doppler velocity of 0, as well as noise components, and some rainfall signal components with finite Doppler velocities are removed. In other words, step S10 is the step of calculating to what extent the rainfall signal components should be recovered.
[0048] The summing unit 25 adds the output result of step S10 to the output result of step S5 (step S11). As shown in the upper right column of Figure 5, the Doppler velocity spectrum at the output stage of step S11 includes the rainfall signal component with a Doppler velocity near 0, the rainfall signal component with a finite Doppler velocity, and the noise component. In other words, step S11 is the step in which the rainfall signal component is recovered. However, since the recovery of the rainfall signal component is not necessarily fully achieved in step S11, the processing in steps S12 to S14 is executed.
[0049] The rainfall signal distribution calculation unit 21 calculates the rainfall signal distribution based on the output result of step S11 (step S12). Then, the rainfall signal distribution calculation unit 21 generates a time-series radar signal based on the rainfall signal distribution from step S12 (step S13).
[0050] If the rainfall signal convergence determination unit 26 determines that the rainfall signal distribution in step S12 has not converged (NO in step S14), it determines that the recovery of the rainfall signal component has not been sufficiently achieved. The rainfall signal recovery device 2 then repeats the processing in steps S8 to S14.
[0051] The rainfall signal convergence determination unit 26 determines that the rainfall signal distribution in step S12 has converged (YES in step S14), and that the recovery of the rainfall signal component has been sufficiently achieved. The rainfall signal recovery device 2 then outputs the time-series radar signal (time-series radar signal in step S13) after applying the window function and removing ground surface clutter at each meteorological observation point (step S15).
[0052] In this embodiment, the ground clutter component is removed by multiplying the autocorrelation matrix of the time-series radar signal by a filtering matrix. As a modification, the ground clutter component may be removed by performing FIR filtering or IIR filtering on the time-series radar signal. In this embodiment, the rainfall signal component is recovered by applying a Gaussian model to the rainfall signal distribution and then multiplying the autocorrelation matrix of the time-series radar signal after ground clutter removal by a filtering matrix. As a modification, the rainfall signal component may be recovered by simply applying a Gaussian model to the rainfall signal distribution without multiplying the autocorrelation matrix of the time-series radar signal after ground clutter removal by a filtering matrix.
[0053] (Process for calculating filtering characteristics in this disclosure) The following describes how the filtering characteristic calculation / storage unit 13 calculates the filtering matrix according to the received intensity and Doppler velocity width of the ground surface clutter distribution.
[0054] Doppler velocity spectrum S of the surface clutter component C (v) is the number 1. Here, P C σ represents the surface clutter intensity, C This represents the surface clutter velocity width.
Math.
Math.
Math.
[0055] The autocorrelation function R of the time-series radar signal of the ground surface clutter component C (τ) (where τ is the delay time) is expressed by Equation 4 with respect to the wavelength λ.
Math.
Math.
Math.
Math.
[0056] Autocorrelation function R of time-series radar signals of ground surface clutter component C Autocorrelation matrix R of (τ) C This is given by equation 8. Here, m is the autocorrelation function R of the time-series radar signal. C This represents the number of signal points for (τ), and * represents the complex conjugate.
number
number
number
number
[0057] Multiplying the autocorrelation matrix R of the autocorrelation function R(τ) of the time-series radar signal, which combines the ground surface clutter component, rainfall signal component, and noise component, by the filtering matrix A results in equation 12. Here, R' is the result of the multiplication, H represents the complex conjugate transpose, and I m This represents an m x m identity matrix.
number
[0058] When the autocorrelation function R(τ) of a time-series radar signal does not include the rainfall signal component, the following three conditions are required for the surface clutter component to be reduced to a noise level in the multiplication result R'.
number
number
[0059] The problems with the filtering characteristic calculation process of this disclosure are shown in Figure 6. In the upper part of Figure 6, no window function such as a Hamming window is applied to the time-series data of the weather radar signal, and no time-domain filtering is performed. As a result, the ground surface clutter component cannot be properly removed. In the middle part of Figure 6, no window function such as a Hamming window is applied to the time-series data of the weather radar signal, and only a rectangular window function is applied to the time-domain filtering. As a result, the ground surface clutter component can be properly removed. In the lower part of Figure 6, a window function such as a Hamming window is applied to the time-series data of the weather radar signal, and only a rectangular window function is applied to the time-domain filtering. As a result, the ground surface clutter component cannot be properly removed.
[0060] (Filtering characteristic storage processing of this disclosure) In the above method, a window function is applied to the time-series data of the weather radar signal, but the same window function is not applied to the time-domain filtering process. Applying a window function to the time-series data of the weather radar signal changes the statistical properties of the time-series data, such as the Gaussian. Therefore, it is not possible to properly remove the ground surface clutter component, and it is not possible to achieve both the effect of removing the ground surface clutter component and the effect of improving the azimuth beam resolution.
[0061] In the following, a window function is applied to the time-series data of weather radar signals, and the same window function is also applied to the time-domain filtering process. Applying a window function to the time-series data of weather radar signals changes the statistical properties of the time-series data, such as the Gaussian. However, it is possible to appropriately remove the ground surface clutter component, and it is possible to achieve both the effect of removing the ground surface clutter component and the effect of improving the azimuth beam resolution.
[0062] As described above, the filtering characteristic calculation / storage unit 13 calculates various filtering characteristics in the time domain during meteorological observations, depending on the various received intensities and Doppler velocity widths of the ground surface clutter component. However, since the calculation of filtering matrix A involves the calculation of the inverse matrix, as described above in Equation 14, the computational processing burden increases during meteorological observations.
[0063] In the following, the filtering characteristic calculation / storage unit 13 stores various filtering characteristics in the time domain prior to meteorological observation, according to various received intensities and Doppler velocity widths of the ground surface clutter component. As a result, even when the calculation of filtering matrix A includes the calculation of the inverse matrix as described above in Equation 14, the computational processing burden during meteorological observation is reduced.
[0064] The procedure for the filtering characteristic storage process of this disclosure is shown in Figure 7. Steps S21 to S28 are processes performed by the filtering characteristic calculation / storage unit 13.
[0065] Doppler velocity spectra S of various surface clutter components to be removed C Set (v) as shown in equation 1 (step S21). Clutter intensity P of the various surface clutter components to be removed. C and clutter velocity width σ C Set the clutter intensity P (step S22). C and clutter velocity width σ CThe narrower the setting interval, the easier it is to select the optimal filtering matrix A from various filtering matrices A, thus improving the accuracy of surface clutter removal.
[0066] Autocorrelation function R of time-series radar signals of various surface clutter components to be removed C (τ) is calculated as shown in equation 4 (step S23). The autocorrelation function R of the time-series radar signals of various ground surface clutter components to be removed. C For (τ), the autocorrelation matrix R after applying the window function is C This is calculated as shown in step 15 (step S24). Here, w1, ..., w m This is a window function, such as a Hamming window, that is applied not only to time-series radar signals but also to the filtering matrix A.
number
[0067] The autocorrelation function R of the time-series radar signal of the expected noise component. N Set (τ) as shown in equation 6 (step S25). The noise intensity σ after applying the window function to the expected noise component. N 2 Set it as shown in equation 16 (step S26). Here, the noise intensity σ N 2 The higher the setting accuracy, the higher the accuracy of ground surface clutter removal. And w1, ..., w m This is a window function, such as a Hamming window, that is applied not only to time-series radar signals but also to the filtering matrix A.
number
[0068] The filtering matrix A, after applying a window function to remove various surface clutter components, is calculated as shown in Equation 14 (step S27). The filtering matrix A, after applying the window function to remove various surface clutter components, is stored (step S28). The optimal filtering matrix A is selected according to the received intensity and Doppler velocity width of the surface clutter distribution in step S3 (step S4). Here, the method for selecting the optimal filtering matrix A is explained.
[0069] Applying a window function to time-series data of weather radar signals reduces the received intensity and Doppler velocity width of the ground surface clutter component. Therefore, prior to meteorological observation, a window function is applied to the time-series data of weather radar signals to calculate the reduction in the received intensity and Doppler velocity width of the ground surface clutter component. Then, during meteorological observation, the reduction in the received intensity and Doppler velocity width of the ground surface clutter component after applying the window function is added to the received intensity and Doppler velocity width of the ground surface clutter component before applying the window function to calculate the received intensity and Doppler velocity width of the ground surface clutter component before applying the window function. As a result, during meteorological observation, the optimal filtering matrix A can be selected according to the received intensity and Doppler velocity width of the ground surface clutter component before applying the window function.
[0070] Applying a window function to the time-series data of weather radar signals reduces the received intensity of noise components. Therefore, prior to weather observation, a window function is applied to the time-series data of weather radar signals to calculate the reduction in the received intensity of noise components. Then, during weather observation, the reduction in the received intensity of noise components after applying the window function is added to the received intensity of noise components after applying the window function to calculate the received intensity of noise components before applying the window function. This allows for the selection of the optimal filtering matrix A in weather observation, based on the received intensity of noise components before applying the window function.
[0071] Figure 8 shows the effect of the filtering characteristic calculation process of this disclosure. In the upper part of Figure 8, no window function such as a Hamming window is applied to the time-series data of the weather radar signal, and no time-domain filtering is performed. As a result, the ground surface clutter component cannot be properly removed. In the middle part of Figure 8, no window function such as a Hamming window is applied to the time-series data of the weather radar signal, and only a rectangular window function is applied to the time-domain filtering. As a result, the ground surface clutter component can be properly removed. In the lower part of Figure 8, a window function such as a Hamming window is applied to the time-series data of the weather radar signal, and a window function such as a Hamming window is applied to the time-domain filtering. As a result, the ground surface clutter component can be properly removed.
[0072] The effects of the filtering characteristic calculation process of this disclosure are also shown in Figure 9. In the upper part of Figure 9, no window function such as a Hamming window is applied to the time-series data of the weather radar signal, and only a rectangular window function is applied to the time-domain filtering process. In the middle part of Figure 9, a window function such as a Hamming window is applied to the time-series data of the weather radar signal, and a window function such as a Hamming window is applied to the time-domain filtering process. In the lower part of Figure 9, the upper and middle parts of Figure 9 are compared, and in the upper part of Figure 9 (super-resolution OFF), the azimuth beam resolution cannot be improved, but in the middle part of Figure 9 (super-resolution ON), the azimuth beam resolution can be improved.
[0073] (Ground surface clutter removal range setting process of this disclosure) In calculating the filtering matrix A, the Doppler velocity width σ of the ground surface clutter component is considered. C Instead of setting all weather observation points to a fixed width, set them to a variable width.
[0074] Therefore, the Doppler velocity width σ C However, in order to remove the small ground clutter component, the variable Doppler velocity width σ C By appropriately narrowing the Doppler velocity width σ, the possibility of removing components other than ground clutter can be reduced. CHowever, in order to remove the large tree clutter component, the variable Doppler velocity width σ C By appropriately widening the range, the possibility that tree clutter components may not be completely removed can be reduced.
[0075] In the following, when calculating the filtering matrix A, the Doppler velocity width σ of the ground surface clutter component is used. C (1) As an initial process, all meteorological observation points are set to a fixed width only, and (2) as a post-processing step, only a portion of the meteorological observation points where removing the ground surface clutter component or extracting the rainfall signal component is difficult in the initial process are changed to a variable width, while the majority of meteorological observation points where removing the ground surface clutter component or extracting the rainfall signal component is possible in the initial process are kept at a fixed width. Therefore, the variable Doppler velocity width σ C This can reduce the burden of configuration processing.
[0076] Figure 10 shows an overview of the ground surface clutter removal range setting process described herein. Within the rainfall area P, the rainfall intensity and rainfall velocity v P The rainfall parameters, including the rainfall velocity and width, are considered to be spatially continuous. The rainfall point P1 within the rainfall region P is located in a direction offset from the front of the weather radar signal processing device R, and the Doppler velocity v of the rainfall signal component at rainfall point P1 D v is a positive finite value. The rainfall point P2 within the rainfall region P is located almost directly in front of the weather radar signal processing unit R, and the Doppler velocity v of the rainfall signal component at rainfall point P2 is... D It is almost 0.
[0077] First, the Doppler velocity width σ of the rainfall signal component. P The width is not too narrow, and the Doppler velocity width σ of the ground surface clutter component C Let's consider the case where the area is not too wide (second row from the top in Figure 10).
[0078] For rainfall location P1, the Doppler velocity v is a finite value. D When extracting the rainfall signal component, which is the Doppler velocity average, the Doppler velocity width σ of the rainfall signal component is used. PBecause the range is not too narrow, if the fixed removal range for the ground surface clutter component is appropriately set, there is a low possibility that the rainfall signal component will be excessively removed or remain in excess, and there is a high possibility that the rainfall signal component will be recoverable. Therefore, the fixed removal range for the ground surface clutter component will be maintained as is.
[0079] For rainfall location P2, the Doppler velocity v is near 0. D When extracting the rainfall signal component, which is the Doppler velocity average, the Doppler velocity width σ of the rainfall signal component is used. P Because the range is not too narrow, if the fixed removal range for the ground surface clutter component is appropriately set, there is a low possibility that the rainfall signal component will be excessively removed or remain in excess, and there is a high possibility that the rainfall signal component will be recoverable. Therefore, the fixed removal range for the ground surface clutter component will be maintained as is.
[0080] Furthermore, from rainfall point P1 to rainfall point P2, the rainfall parameters, including the received intensity of the rainfall signal component, the average Doppler velocity, and the Doppler velocity width, are spatially continuous.
[0081] Next, the Doppler velocity width σ of the ground surface clutter component. C It is not too wide, but the Doppler velocity width σ of the rainfall signal component P Let's consider the case where it is too narrow (third row from the top in Figure 10).
[0082] For rainfall location P1, the Doppler velocity v is a finite value. D When extracting the rainfall signal component, which is the Doppler velocity average, the Doppler velocity width σ of the rainfall signal component is used. P Although the range is too narrow, if the fixed removal range for the ground surface clutter component is appropriately set, the likelihood of excessive removal or excessive retention of the rainfall signal component is low, and the likelihood of recovery of the rainfall signal component is high. Therefore, the fixed removal range for the ground surface clutter component is maintained as is.
[0083] For rainfall location P2, the Doppler velocity v is near 0. DWhen extracting the rainfall signal component, which is the Doppler velocity average, the Doppler velocity width σ of the rainfall signal component is used. P Because the range is too narrow, even if the fixed removal range for the ground surface clutter component is appropriately set, there is a high possibility that the rainfall signal component will be removed excessively, making it highly likely that the rainfall signal component will become irrecoverable. Therefore, it is desirable to change the removal range for the ground surface clutter component to a variable width that is narrower than the fixed width.
[0084] Furthermore, from rainfall point P1 to rainfall point P2, the rainfall parameters, including the received intensity, average Doppler velocity, and Doppler velocity width of the rainfall signal component, are spatially discontinuous. In particular, because the rainfall signal component becomes irrecoverable, the received intensity of the rainfall signal component is prone to spatial discontinuity. Moreover, if the Gaussian fitting in step S7 is not appropriate, the average Doppler velocity and Doppler velocity width of the rainfall signal component will also become spatially discontinuous.
[0085] Next, the Doppler velocity width σ of the rainfall signal component. P It is not too narrow, but the Doppler velocity width σ of the ground surface clutter component C Let's consider the case where it is too wide (fourth row from the top in Figure 10).
[0086] For rainfall location P1, the Doppler velocity v is a finite value. D When extracting the rainfall signal component, which is the Doppler velocity average, the Doppler velocity width σ of the ground surface clutter component is used. C Although the range is too wide, if the fixed removal range for the ground surface clutter component is set appropriately, the likelihood of excessive removal or excessive retention of the rainfall signal component is low, and the likelihood of recovery of the rainfall signal component is high. Therefore, the fixed removal range for the ground surface clutter component will be maintained as is.
[0087] For rainfall location P2, the Doppler velocity v is near 0. D When extracting the rainfall signal component, which is the Doppler velocity average, the Doppler velocity width σ of the ground surface clutter component is used. CBecause the range is too wide, even if the fixed removal range for surface clutter components is appropriately set, there is a high possibility that the remaining surface clutter components will be extracted as spurious rainfall signal components, and that the rainfall signal components will be excessively restored. Therefore, it is desirable to change the removal range for surface clutter components to a variable width that is wider than the fixed width.
[0088] Furthermore, from rainfall point P1 to rainfall point P2, the rainfall parameters, including the received intensity, average Doppler velocity, and Doppler velocity width of the rainfall signal component, are spatially discontinuous. In particular, because the rainfall signal component undergoes excessive recovery, the received intensity of the rainfall signal component is prone to spatial discontinuity. Moreover, if the Gaussian fitting in step S7 is not appropriate, the average Doppler velocity and Doppler velocity width of the rainfall signal component will also become spatially discontinuous.
[0089] Conversely, a meteorological observation point where the parameters of the rainfall signal component, which should inherently be continuous, are incorrectly discontinuous (discontinuous observation point) is considered a meteorological observation point where the rainfall signal component is irrecoverable or excessively recovered. Therefore, it is possible to distinguish between meteorological observation points where the rainfall signal component is irrecoverable and meteorological observation points where the rainfall signal component is inherently low. Furthermore, it is possible to distinguish between meteorological observation points where the rainfall signal component is excessively recovered and meteorological observation points where the rainfall signal component is inherently high.
[0090] The procedure for setting the ground surface clutter removal range according to this disclosure is shown in Figure 11.
[0091] The filtering characteristic calculation / storage unit 13 calculates the Doppler velocity width σ of the ground surface clutter component that should be set in the time-domain filtering process. C As an initial process, this is set to a fixed width regardless of the weather observation point. Specifically, the filtering characteristic calculation / storage unit 13 calculates the Doppler velocity width σ of the ground surface clutter component based on the antenna beam width, scan speed, and the characteristics of the ground surface clutter component. C The width is set to a fixed width. Then, the ground surface clutter removal device 1 and the rainfall signal recovery device 2 execute the processes of steps S1 to S15 (step S31).
[0092] The rainfall signal distribution acquisition unit 31 acquires the rainfall signal distribution in step S12 at each meteorological observation point (step S32). The received power extraction unit 32, the Doppler velocity average extraction unit 33, and the Doppler velocity width extraction unit 34 extract the received power, Doppler velocity average, and Doppler velocity width from the rainfall signal distribution in step S32 at each meteorological observation point (step S33).
[0093] The discontinuity point extraction unit 35 extracts discontinuity observation points where at least one of the received power, Doppler velocity average, and Doppler velocity width of the rainfall signal component recovered by the rainfall signal recovery device 2 has a discontinuity within the space of the meteorological observation (step S34).
[0094] The filtering characteristic modification unit 36 sets the Doppler velocity width σ of the ground surface clutter component to be set in the time-domain filtering process. C As a post-processing step, the discontinuity point extraction unit 35 changes the fixed width to a variable width at the discontinuity observation points it has extracted (YES in step S34, step S35). Then, the ground surface clutter removal device 1 and the rainfall signal recovery device 2 execute the processes in steps S1 to S15 again (step S35). Here, steps S34 and S35 may be repeated until there are no more discontinuity observation points in step S34.
[0095] Here, we will explain the specific procedures for steps S34 and S35.
[0096] The discontinuity point extraction unit 35 determines at each meteorological observation point whether the Doppler velocity average reverses sign compared to surrounding meteorological observation points (step S41). In other words, the discontinuity point extraction unit 35 searches for meteorological observation points where the Doppler velocity average is near zero.
[0097] For weather observation points where the Doppler velocity average reverses sign compared to surrounding weather observation points (YES in step S42), there is a possibility of excessive removal or retention of the rainfall signal component, so the process proceeds to step S43. For weather observation points where the Doppler velocity average does not reverse sign compared to surrounding weather observation points (NO in step S42), there is a low possibility of excessive removal or retention of the rainfall signal component, so the process is terminated.
[0098] The discontinuity point extraction unit 35 determines at each meteorological observation point whether the received power or the power difference before and after recovery is smaller or larger than a predetermined difference compared to surrounding meteorological observation points (step S43). In other words, the discontinuity point extraction unit 35 searches for meteorological observation points where there is an excess removal or excess retention of rainfall signal components, and extracts such meteorological observation points as discontinuity observation points.
[0099] For weather observation points where the received power or the power difference before and after recovery is smaller than a predetermined difference compared to surrounding weather observation points (in step S44, "small"), the filtering characteristic changing unit 36 sets the Doppler velocity width σ of the ground surface clutter component to be set in the time-domain filtering process. C As a post-processing step, the width is changed to a variable width narrower than the fixed width (step S45). Then, the ground surface clutter removal device 1 and the rainfall signal recovery device 2 repeat the processing in steps S1 to S15 (step S45). Here, steps S34 to S35 and S41 to S46 may be repeated until there are no discontinuous observation points in steps S34 and S44.
[0100] For weather observation points where the received power or the power difference before and after recovery is greater than a predetermined difference compared to surrounding weather observation points (in step S44, "great"), the filtering characteristic changing unit 36 sets the Doppler velocity width σ of the ground surface clutter component to be set in the time-domain filtering process. CAs a post-processing step, the width is changed to a variable width that is wider than the fixed width (step S46). Then, the ground surface clutter removal device 1 and the rainfall signal recovery device 2 perform the processing in steps S1 to S15 again (step S46). Here, steps S34 to S35 and S41 to S46 may be repeated until there are no discontinuous observation points in steps S34 and S44.
[0101] For weather observation points where the received power or the power difference before and after recovery is neither smaller nor larger than a predetermined difference compared to surrounding weather observation points (in step S44, "neither"), the possibility of excessive removal or excessive retention of rainfall signal components is low, and therefore the process is terminated.
[0102] Furthermore, the window function applied to the time-series data of the weather radar signal can be set prior to weather observations according to the rotation speed of antenna A and the size of the target object. During weather observations, even if discontinuous observation points of the rainfall signal component are extracted, the window function may be fixed rather than made variable. [Industrial applicability]
[0103] The weather radar signal processing device and weather radar signal processing program disclosed herein, when performing time-domain filtering on a weather radar signal, (1) achieve both the effect of removing surface clutter components and the effect of improving azimuth beam resolution, (2) reduce the possibility of excessive or incomplete removal of surface clutter, and reduce the possibility of irrecoverable or excessive recovery of rainfall signal components, and (3) reduce the computational processing burden during meteorological observation. [Explanation of Symbols]
[0104] A: Antenna B, B', B": Beam R: Weather radar signal processing unit 1: Ground surface clutter removal device 2: Rain signal recovery device 3: Discontinuity point extraction device 11: Autocorrelation matrix calculation unit 12: Ground surface clutter distribution calculation part 13: Filtering characteristic calculation / storage unit 14: Ground surface clutter removal section 21: Rainfall signal distribution calculation section 22: Autocorrelation Matrix Calculation Unit 23: Filtering Processing Unit 24: Subtraction section 25: Addition section 26: Rainfall signal convergence determination unit 31: Rainfall signal distribution acquisition section 32: Received power extraction unit 33: Doppler velocity-averaged extraction unit 34: Doppler velocity width extraction unit 35: Discontinuity point extraction unit 36: Filtering characteristic modification section P:Rainfall area P1, P2: Rainfall point
Claims
1. A weather radar signal acquisition unit that acquires time-series data of a weather radar signal to which a window function has been applied from a weather radar device equipped with an antenna that rotates in the azimuth direction, In extracting the rainfall signal component by removing the ground surface clutter component from the time-series data of the weather radar signal to which the aforementioned window function has been applied, the ground surface clutter removal unit performs filtering in the time domain to which the aforementioned window function has been applied to the time-series data of the weather radar signal, A rainfall signal recovery unit recovers the rainfall signal component that was removed along with the ground surface clutter component by the ground surface clutter removal unit, by calculating the intensity distribution of the received power with respect to the Doppler velocity of the rainfall signal component extracted by the ground surface clutter removal unit based on statistical properties, A weather radar signal processing device characterized by comprising the following:
2. The ground surface clutter removal unit sets the Doppler velocity width of the ground surface clutter component to a variable width according to the weather observation point during the time-domain filtering process to which the window function applied to the time-series data of the weather radar signal is applied. The weather radar signal processing device according to claim 1, characterized in that
3. The ground surface clutter removal unit, in the time-domain filtering process to which the window function applied to the time-series data of the weather radar signal is applied, sets the Doppler velocity width of the ground surface clutter component to be set as a fixed width regardless of the weather observation point as an initial process. The weather radar signal processing device further includes a discontinuity point extraction unit that extracts discontinuity observation points in which at least one of the received power, Doppler velocity average, and Doppler velocity width of the rainfall signal component recovered by the rainfall signal recovery unit has a discontinuity within the space of the weather observation, The ground surface clutter removal unit, in the time-domain filtering process to which the window function applied to the time-series data of the weather radar signal is applied, changes the Doppler velocity width of the ground surface clutter component to be set as a post-processing step, changing it to a variable width different from the fixed width at the discontinuous observation points extracted by the discontinuous point extraction unit. The weather radar signal processing device according to claim 2, characterized in that
4. The discontinuity point extraction unit extracts as discontinuity points weather observation points weather observation points where the Doppler velocity average of the rainfall signal component recovered by the rainfall signal recovery unit reverses sign compared to surrounding weather observation points, and where the received power of the rainfall signal component recovered by the rainfall signal recovery unit or the power difference before and after recovery is smaller than a predetermined difference compared to surrounding weather observation points. The ground surface clutter removal unit, in the time-domain filtering process to which the window function applied to the time-series data of the weather radar signal is applied, changes the Doppler velocity width of the ground surface clutter component to be set as a post-processing step at the discontinuous observation points extracted by the discontinuous point extraction unit to a variable width that is narrower than the fixed width. The weather radar signal processing device according to claim 3, characterized in that
5. The discontinuity point extraction unit extracts as discontinuity points weather observation points weather observation points where the Doppler velocity average of the rainfall signal component recovered by the rainfall signal recovery unit reverses sign compared to surrounding weather observation points, and where the received power of the rainfall signal component recovered by the rainfall signal recovery unit or the power difference before and after recovery is greater than a predetermined difference compared to surrounding weather observation points. The ground surface clutter removal unit, in the time domain filtering process to which the window function applied to the time series data of the weather radar signal is applied, changes the Doppler velocity width of the ground surface clutter component to be set as a variable width that is wider than the fixed width at the discontinuous observation points extracted by the discontinuous point extraction unit as a post-processing step. The weather radar signal processing device according to claim 3, characterized in that
6. The ground surface clutter removal unit stores, prior to meteorological observation, various filtering characteristics in the time domain to which the window function applied to the time-series data of the weather radar signal is applied, according to various received intensities and Doppler velocity widths of the ground surface clutter component. A weather radar signal processing device according to any one of claims 1 to 5, characterized in that
7. A weather radar signal acquisition step involves acquiring time-series data of a weather radar signal to which a window function has been applied, from a weather radar device equipped with an antenna that rotates in the azimuth direction. In extracting the rainfall signal component by removing the ground surface clutter component from the time-series data of the weather radar signal to which the aforementioned window function has been applied, the process includes a ground surface clutter removal step in which a filtering process is performed in the time domain to which the aforementioned window function has been applied to the time-series data of the weather radar signal, A rainfall signal recovery step is performed to recover the rainfall signal component that was removed along with the ground surface clutter component in the ground surface clutter removal step by calculating the intensity distribution of the received power with respect to the Doppler velocity of the rainfall signal component extracted in the ground surface clutter removal step based on statistical properties, A weather radar signal processing program that causes a computer to execute the following steps in sequence.
Citation Information
Patent Citations
Rainfall signal recovery method and device of polarization-Doppler meteorological radar
CN114415184A
Radar signal processing device and method thereof
JP2002131421A
Meteorological radar signal processor and meteorological radar signal processing program
JP2020106471A
System and method for processing data in weather radar
US20080001808A1
System and method for increasing spectral resolution
US20100231439A1
