A weather radar IQ data quality control method for animal migration monitoring

Abstract

The application provides a weather radar IQ data quality control method for animal migration monitoring, and belongs to the technical field of data processing, and is used for solving the problem that only meteorological targets are controlled in the traditional weather radar data quality control method. The method comprises the following steps: firstly, calculating a power spectrum based on weather radar IQ data; estimating frequency spectrum noise based on the power spectrum, and suppressing the noise background in the frequency spectrum; calculating polarization spectrum features based on the frequency spectrum data, using the slow change characteristics of the polarization spectrum features of ground object interference to perform frequency domain filtering of the ground object interference; combining the clutter phase consistency and the frequency spectrum power discrimination factor to determine the frequency spectrum with the ground object interference, and further performing frequency domain filtering of the ground object interference; and finally, establishing a Bayesian classification model based on the polarization spectrum feature statistical data of the meteorological interference and the animal target, and combining the continuity difference between the two spectrum features to calculate a screening index to suppress the meteorological interference and retain the animal target.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and specifically to a weather radar IQ data quality control method for monitoring animal migration. Background Technology

[0002] Migratory animals are an important component of the aerial ecosystem, and their effective monitoring plays a crucial role in agricultural pest control and ecological environmental protection. Radar, with its all-weather, all-time monitoring capabilities, has become an effective means of monitoring migratory animals. Dedicated radars for insect and bird monitoring are limited by transmission power and antenna aperture, allowing monitoring only within a limited range of a few kilometers. Weather radar, however, can detect the echoes of migratory animal groups near meteorological interference targets such as precipitation clouds within a range of hundreds of kilometers, providing insights into the aerial activities of migratory animals. Because meteorological interference and the movement patterns of migratory animal targets differ significantly, their polarization spectral characteristics on the range-Doppler spectrum show marked differences. Quality control of weather radar IQ data based on polarization spectral characteristics will aid in the monitoring of migratory animals. Since studying migratory animal activity provides crucial support for agricultural pest control and ecological environmental protection, achieving quality control of weather radar IQ data for animal migration monitoring is of significant value.

[0003] Traditional weather radar data quality control methods process meteorological targets, relying on baseline data to perform quality control on a per-resolution-cell basis, suppressing most non-meteorological targets. Furthermore, when suppressing ground object interference, traditional methods require combining baseline data to determine the presence of ground object interference within the resolution cell before performing frequency domain filtering, making ground object interference filtering time-consuming and labor-intensive.

[0004] It is evident that current weather radar data quality control methods primarily focus on meteorological targets, operate on a unit-by-unit basis, and rely on baseline data for processing, resulting in the missed detection of a large number of migratory animal targets in the baseline data images. Therefore, there is an urgent need to develop a weather radar IQ data quality control method that combines polarimetric spectral features for ground object interference filtering and utilizes the continuity difference in polarimetric spectral features between meteorological interference and animal targets for animal migration monitoring. Summary of the Invention

[0005] In view of this, the present invention provides a weather radar IQ data quality control method for monitoring animal migration, which can suppress ground object interference and meteorological interference in the spectrum based on weather radar IQ data without relying on the basic data of the resolution unit, thereby preserving the animal target component in the spectrum and facilitating the monitoring of migratory animals.

[0006] To solve the above-mentioned technical problems, the present invention is implemented as follows.

[0007] A method for quality control of weather radar IQ data for animal migration monitoring, comprising:

[0008] Step 1: Calculate the power spectrum based on weather radar IQ data as the raw spectrum; Step 2: For a single resolution cell, estimate the spectral noise level based on the power distribution of each spectral component in the original spectrum; suppress the spectral components of background noise based on the spectral noise level to generate an original spectral mask; the unsuppressed spectral components in the original spectral mask form multiple preserved spectral bands; the original spectral masks of each resolution cell constitute the first spectral mask of the distance-Doppler spectrum. Step 3: Calculate the polarization spectrum characteristics based on the original spectrum, utilize the gradual variation characteristics of the polarization spectrum characteristics of ground object interference to determine the first spectrum interval of ground object interference, suppress ground object interference on the first spectrum mask, and obtain the second spectrum mask. Step 4: Calculate the clutter phase consistency index based on the original spectrum. Utilize the characteristic that the clutter phase consistency index of ground object interference is higher than the set clutter threshold to determine the second spectral range of ground object interference. Perform ground object interference suppression on the second spectral mask to obtain the third spectral mask. Step 5: Based on the statistical data of the polarization spectral characteristics of meteorological interference and animal targets, establish a Bayesian classification model. Combining the continuity difference of the polarization spectral characteristics of the two, distinguish the categories of each preserved spectral band in the third spectral mask, and suppress meteorological interference and preserve animal targets.

[0009] Preferably, after step 2 generates the original spectral mask for the resolution unit, the method further includes: Based on the original spectral mask, the spectral peak value of each retained spectral band is determined. Combined with the spectral peak value, it is determined whether there is spectral leakage in the current resolution cell. If there is spectral leakage, all spectral components in the current resolution cell are suppressed and the original spectral mask is updated; otherwise, the original spectral mask is not updated.

[0010] Preferably, step 2 specifically includes: Step 201: For a single resolution cell, the original spectrum is reordered to obtain the rearranged spectrum; the rearranged spectrum is plotted with power value on the vertical axis and the proportion of spectral components with power values ​​less than the vertical axis on the horizontal axis; the power values ​​in the rearranged spectrum are made positive. Step 202: Extract data points from the power value flattening region of the rearranged spectrum, perform Poisson fitting, and inversely positiveize the fitted data to obtain the fitted spectrum; the rearranged spectrum is considered to be higher than the fitted spectrum and the difference between the two is greater than a set level value for the first time. The power value at that location represents the spectral noise level. Step 203: Based on the aforementioned spectral noise level, calculate the signal-to-noise ratio (SNR) of each spectral component in the original spectrum of each resolution unit, and compare it with the SNR threshold. Comparison; above the signal-to-noise ratio threshold The spectral components are set to 1, not exceeding the signal-to-noise ratio threshold. The spectral components are set to 0 to obtain the original spectral mask. Original spectral mask The spectral components set to 1 form multiple retained spectral bands; Step 204: Locate the spectral peaks of each retained spectral band in the original spectrum; determine whether the spectral peaks on both sides of the zero Doppler velocity are symmetrical, and whether the standard deviation of the spectral peaks is less than the set power standard deviation threshold. If so, it is determined that there is spectral leakage, and the mask value corresponding to all spectral components in the current resolution unit is set to 0, and the original spectral mask is updated.

[0011] Preferably, in step 204, if it is determined that there is no spectral leakage, then based on the signal-to-noise ratio of each spectral component calculated in step 203, the retained spectral band with a peak signal-to-noise ratio of less than 10dB is suppressed.

[0012] Preferably, in step 204, the determination of whether the spectral peaks on both sides of the zero Doppler velocity are symmetrical is as follows: Symmetry judgment includes positional symmetry and quantitative symmetry; The criterion for determining positional symmetry is as follows: select a spectral peak position on each side of the zero Doppler velocity and find the midpoint position; if the difference between the midpoint position and the zero Doppler velocity is less than a set distance threshold, then the position is determined to be symmetrical. The criterion for determining quantity symmetry is: if the difference in the number of spectral peaks on both sides of zero Doppler velocity is less than a set difference threshold, then it is determined to be quantity symmetric.

[0013] Preferably, step 3 specifically includes: Step 301: For each resolution unit, calculate the polarization spectrum characteristics based on the original spectrum, including spectral co-polarization and spectral differential phase; set the judgment condition for the slow variation characteristics as the standard deviation of spectral co-polarization being less than 0.05 or the standard deviation of spectral differential phase being less than 15°; Step 302: Using the three data points centered on zero Doppler velocity as the current interval, calculate the standard deviation of spectral copolarization and spectral differential phase within the interval. When the slowly varying characteristic judgment condition is met, expand the current interval until the slowly varying characteristic judgment condition is no longer met. Then, the current interval is taken as the first spectral interval of ground object interference. Step 303: Based on the first spectral range, apply the first spectral mask. The corresponding spectral components in the range-Doppler spectrum are set to 0; steps 301-303 are performed on each resolution unit to obtain the second spectral mask of the range-Doppler spectrum. .

[0014] Preferably, the expansion of the current interval in step 302 is as follows: It expands symmetrically to both sides of zero Doppler velocity; Alternatively, the expansion can be initiated towards the zero Doppler velocity side. After each expansion, it is determined whether the current interval satisfies the condition for judging the gradual change characteristic. If it does, the expansion in the current direction continues. If it does not, the boundary of the interval before this expansion is used as the current side boundary, and the expansion is initiated towards the other side of the zero Doppler velocity. After each expansion, it is determined whether the current interval satisfies the condition for judging the gradual change characteristic until the condition is no longer met. In this case, the current interval is a provisional interval. The deviation between the center of symmetry of the provisional interval and the zero Doppler velocity is calculated. If the deviation is less than or equal to twice the spectral resolution, it indicates that the spectrum of the provisional interval is generated by ground object interference, and the provisional interval is taken as the first spectral interval of the ground object interference. Otherwise, the spectral components in the provisional interval are not suppressed.

[0015] Preferably, step 4 further includes: calculating the spectral power ratio index based on the original spectrum; utilizing the characteristic that the spectral power ratio index of ground object interference is higher than the set power ratio, determining the spectral range of ground object interference considering the spectral power ratio index, and combining it with the spectral range of ground object interference determined based on the clutter phase consistency index to jointly form the second spectral range, suppressing ground object interference on the second spectral mask, and obtaining the third spectral mask.

[0016] Preferably, in step 4, the third spectral mask is obtained in the following way: Step 401: Use clutter phase consistency index and spectral power ratio index as ground feature detection indicators; for each resolution cell, according to the second spectral mask... Determine whether there is still an unsuppressed power spectrum within a frequency range of a set length centered on the zero Doppler velocity; if so, calculate the ground object detection index for the current resolution unit, and select the spectrum with the ground object detection index greater than 0.5 as the ground object interference spectrum. Step 402: Take the three data points centered on the zero Doppler velocity as the current region. If the power value at the boundary of the current region is greater than the power value of the adjacent point outside the boundary and the difference is greater than or equal to 5dB, then expand the current region until the condition can no longer be met, and obtain a provisional interval. Step 403: Extract the data in the provisional interval from the ground object interference spectrum determined in step 401, perform Gaussian fitting, and obtain the Gaussian fitted spectrum corresponding to the ground object interference; Step 404: The second spectral interval of ground object interference is formed by taking the position on both sides of the zero Doppler velocity where the original spectral power value is stronger than the Gaussian fitted spectral power value for the first time. Step 405: Apply the second spectral mask according to the second spectral range. The corresponding spectral components are set to 0; steps 401-405 are performed on each resolution unit to obtain the third spectral mask of the range-Doppler spectrum. .

[0017] Preferably, step 5 specifically includes: For each data point d in the range-Doppler spectrum, calculate the polarization spectral features and the mean value of the polarization spectral features within a sliding window centered on the data point d; the polarization spectral features include one or more combinations of spectral copolarization, spectral differential phase, and spectral differential reflectivity; Polarization spectrum characteristics and the corresponding polarization spectrum characteristic mean Inputting each data point into a Bayesian classification model yields polarization spectral features. They belong to meteorological disturbances. and animal targets probability and and the mean value of polarization spectrum characteristics They belong to meteorological disturbances. and animal targets probability and ; Will and Add them together to obtain the meteorological interference. Screening criteria ; Will and Add them together to obtain the animal target. Screening criteria ; Greater than This indicates that data point d is subject to meteorological interference; conversely, it indicates that data point d is a target animal. For each retained spectral band in the third spectral mask, the proportion of data points in the two categories of meteorological interference and animal targets is counted, and the category with the larger proportion is the category of the retained spectral band; the retained spectral band of meteorological interference in the third spectral mask is suppressed to obtain the fourth spectral mask that retains only animal targets; The original spectrum was extracted using a fourth spectral mask to obtain spectral data that retained only the animal target.

[0018] Beneficial effects: (1) This invention uses the power spectrum obtained from weather radar IQ data to estimate the spectral noise level based on the power value distribution of each spectral component, and suppresses the spectral components of background noise based on the spectral noise level. Furthermore, this invention also detects spectral leakage based on spectral peak symmetry and power standard deviation to avoid background noise being misjudged as animal targets due to spectral leakage.

[0019] (2) In a preferred embodiment, when estimating the spectral noise level, Poisson fitting is used to estimate the spectral noise, which takes advantage of the characteristic that spectral noise is randomly distributed as white noise, so as to achieve a reliable estimate of the spectral noise level.

[0020] (3) Based on the gradual change characteristics of the polarization spectrum of ground object interference, the present invention realizes efficient frequency domain filtering of ground object interference, which further improves the accuracy of preserving animal target components.

[0021] (3) This invention uses clutter phase consistency and spectral power discrimination factor to determine the spectrum of ground object interference, establish a ground object spectrum model and cancel it with the original spectrum to achieve frequency domain filtering, further improving the accuracy of preserving animal target components. Among them, clutter phase consistency and spectral power discrimination factor, the former reflects the radial velocity and IQ data phase fluctuation of the target generating echo data within the resolution cell. For ground object interference, since it is stationary in the radial direction relative to the radar, there is almost no fluctuation in its radial velocity and phase, and its clutter phase consistency value is relatively large; the latter reflects the ratio of target power in the clutter interval near zero Doppler velocity in the power spectrum to the power outside the interval. For ground object interference, its energy is concentrated near zero Doppler velocity in the power spectrum, which makes the spectral power discrimination factor maintain a high value. The two indicators detect ground object interference from the perspectives of spatial motion state and spectral distribution, respectively, improving the detection accuracy of missed ground object interference.

[0022] (4) This invention combines the continuity difference of polarization spectral features between meteorological disturbances and animal targets, uses a sliding window to calculate the regional mean features of polarization spectral features, and uses the polarization spectral features and their mean as input to the Bayesian classification model. The spectral feature Bayesian classification model and a preset function are used to calculate screening indicators to suppress meteorological disturbances and retain animal targets. The reason for choosing polarization spectral features and their mean as input to the Bayesian classification model is that although both meteorological disturbances and animal targets are distributed targets, the former is composed of hydrophobic material. Compared to individual animals, the movement state of hydrophobic material differs less and its physical structure is simpler. This makes the polarization spectral features of meteorological disturbances more stable in the distance-Doppler spectrum and their values ​​more concentrated, resulting in smaller variations in the mean obtained through the sliding window processing. This leads to a significant numerical difference in the output results obtained after inputting the mean of the polarization features into the Bayesian classifier between meteorological disturbances and animal targets. Utilizing this difference can effectively improve the accuracy of suppressing meteorological disturbances. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a weather radar IQ data quality control method for monitoring animal migration, as provided in this embodiment. Figure 2 This embodiment provides a schematic diagram of a Poisson fitting estimation of spectral noise. Figure 3 This embodiment provides a schematic diagram of a ground feature interference suppression process based on the slowly varying characteristics of spectral features. Figure 4 This embodiment provides a schematic diagram of a ground object interference suppression process based on clutter modeling cancellation. Figure 5 This is a schematic diagram illustrating the differentiation results between meteorological disturbances and migratory animals provided in this embodiment. Detailed Implementation

[0024] This invention provides a weather radar IQ data quality control method for animal migration monitoring. The core idea is to suppress background noise, ground object interference, and meteorological interference in the spectrum based on weather radar IQ data, without relying on the base data of the resolution unit. Specifically, ground object interference suppression considers the polarization spectral characteristics of background noise and ground object interference, as well as the characteristics of ground object detection indicators such as clutter phase consistency and spectral power ratio, performing multi-round, multi-angle suppression on the spectral data. Simultaneously, during background noise suppression, the influence of spectral leakage is further eliminated. This invention can improve the accuracy of preserving animal target components in the spectrum and improve the quality of weather radar IQ data.

[0025] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0026] Figure 1 The flowchart shown in the figure illustrates a weather radar IQ data quality control method for animal migration monitoring according to an embodiment of the present invention. The method includes the following steps: Step 1: Calculate the power spectrum based on weather radar IQ data as the raw spectrum.

[0027] The vertical axis of the power spectrum represents the power value, and the horizontal axis represents the Doppler velocity v.

[0028] Step 2: Suppress background noise and eliminate the effects of spectral leakage.

[0029] This step estimates the spectral noise level for a single resolution cell based on the power distribution of each spectral component in the original spectrum; then, based on the spectral noise level, it suppresses the spectral components of the background noise to generate the original spectral mask. . In the diagram, 1 indicates retained data, and 0 indicates suppressed data. Unsuppressed spectral components in the original spectral mask form multiple regions, called retained spectral bands. Subsequent steps... After the update, a new mask is formed, which is also composed of 0s and 1s, with a frequency band of consecutive 1s, also known as the reserved frequency band.

[0030] To identify spectral leakage, this step further determines the spectral peak values ​​of each retained spectral band based on the original spectral mask, and uses these peak values ​​to determine whether spectral leakage exists in the current resolution cell. If spectral leakage exists, it indicates that the suppression operation in the previous step was inaccurate. In this case, all spectral components in the current resolution cell are suppressed, i.e., set to 0, and the original spectral mask is updated. The updated spectral masks for each resolution cell constitute the first spectral mask of the distance-Doppler spectrum. If no leakage exists, the original spectral mask is not updated. ; each resolution unit Forming the first spectral mask .

[0031] In a preferred embodiment, step 2 specifically includes the following sub-steps: Step 201: For a single resolution cell, the original spectrum is reordered to obtain the rearranged spectrum; the rearranged spectrum is plotted with power values ​​on the vertical axis and the percentage of spectral components with power values ​​less than the vertical axis on the horizontal axis; see [link to relevant documentation]. Figure 2 .

[0032] Since the input of the Poisson distribution in step 202 is a positive number, the power values ​​in the rearranged spectrum need to be positiveized as follows:

[0033] in, This is the power value after positive conversion. This is the original power value. This is the minimum power value in the spectrum. This indicates the position of the rearranged power spectrum.

[0034] Step 202: Extract data points from the power value smooth regions of the rearranged spectrum, perform Poisson fitting, and then inversely positiveize the fitted data to obtain the fitted spectrum, such as... Figure 2 The orange curve in the image. This occurs when the rearranged spectrum is higher than the fitted spectrum, and the difference between the two exceeds a set level for the first time. The power value at that point is the spectral noise level.

[0035] In a preferred embodiment, a Poisson fit is performed using power values ​​in a smooth region with a proportion between 0.2 and 0.5. After inverse positiveizing the fitted data, the rearranged power spectrum is selected based on the power spectrum being stronger than the fitted power spectrum and being greater than the first power spectrum. The power value at 3dB is the spectral noise level.

[0036] Step 203: Based on the spectral noise level obtained in Step 202, calculate the signal-to-noise ratio (SNR) of each spectral component in the original spectrum of each resolution unit, and compare it with the SNR threshold. Comparison. (The data will be above the signal-to-noise ratio threshold.) The spectral components with a signal-to-noise ratio (SNR) of 1 are set to 1, and those with a SNR of 0 or less are set to 0, thus obtaining the original spectral mask. Original spectral mask The spectral components that are set to 1 form multiple retained spectral bands.

[0037] In a preferred embodiment, the signal-to-noise ratio threshold Choosing 5dB will result in a spectral mask. The calculation formula is as follows:

[0038] in, The position in the original spectral mask is The mask values ​​for the spectral components, For position is The signal-to-noise ratio of the spectral components; The position of the spectral component is indicated by 0, where 0 represents suppressed data and 1 represents retained data.

[0039] Step 204: Detect Spectral Leakage: Locate the spectral peaks of each retained spectral band in the original spectrum; determine whether the spectral peaks on both sides of the zero Doppler velocity are symmetrical, and whether the standard deviation of the spectral peaks is less than the set power standard deviation threshold. If so, it is determined that there is a spectral leak, the spectral data in the current resolution unit is unavailable, the mask value corresponding to all spectral components in the current resolution unit is set to 0, and the original spectral mask is updated.

[0040] In this step, firstly according to Extract the spectral data corresponding to the retained spectral band from the original spectrum, and obtain the spectral peaks within the retained spectral band range; judge from two perspectives: symmetry and spectral peak standard deviation. ① Determine whether the spectral peaks on both sides of zero Doppler velocity are symmetrical: Symmetry judgment includes positional symmetry and numerical symmetry.

[0041] The criteria for determining positional symmetry are as follows: select a spectral peak position on each side of the zero Doppler velocity and find the midpoint position; if the difference between the midpoint position and the zero Doppler velocity is less than a set distance threshold, such as less than 1.5 m / s, then it is determined to be positional symmetric.

[0042] The criterion for determining quantitative symmetry is: if the difference in the number of spectral peaks on both sides of zero Doppler velocity is less than a set difference threshold, for example, a difference threshold of 2, then it is determined to be quantitative symmetric.

[0043] ② Determine whether the standard deviation of the spectral peak is less than the set power standard deviation threshold.

[0044] Calculate the power standard deviation of the peak values ​​extracted from all retained spectral bands. If the power standard deviation is less than the power standard deviation threshold, for example, less than 5 dB, determine the judgment dimension that meets the standard deviation.

[0045] If both conditions ① and ② above are met simultaneously, it indicates that there is spectral leakage in the current spectrum. Therefore, all spectral data of the current resolution unit should be suppressed, meaning that the current resolution unit's... Set all positions that are 1 to 0.

[0046] If no spectrum leakage is determined, no update is needed. Furthermore, based on the signal-to-noise ratio of each spectral component calculated in step 203, if the peak signal-to-noise ratio is less than 10dB, it indicates that the data segment is generated by background noise from an interfering target, and the retained spectral segment of that segment is suppressed.

[0047] Repeat the above operation for each resolving unit to finally obtain the first spectral mask of the range-Doppler spectrum. .

[0048] Step 3: Suppress ground object interference based on polarization spectrum characteristics.

[0049] Step 3 calculates the polarization spectral characteristics based on the original spectrum. Utilizing the gradually varying polarization spectral characteristics of ground object interference, it determines the first spectral interval of the ground object interference and applies a first spectral mask. To suppress ground object interference and obtain a second spectral mask .

[0050] In this embodiment, the polarization spectrum features include spectral co-polarization, spectral differential phase, and spectral differential reflectivity. Two features, spectral co-polarization and spectral differential phase, are used in step 3; in step 5, when suppressing meteorological interference, three features are employed: spectral co-polarization, spectral differential phase, and spectral differential reflectivity.

[0051] The formulas for calculating the three features are as follows:

[0052]

[0053]

[0054] in, For spectral copolarization, For spectral differential phase, Spectral differential reflectance; , These are the spectral values ​​for the horizontal and vertical polarization channels, indicated by superscript. Indicates conjugate computation; m The index of the spectral position during summation calculation. , The power spectra of the horizontal and vertical polarization channels are shown. This represents the spectral position.

[0055] Verification based on extensive field data shows that when ground object interference dominates the spectral components near zero Doppler velocity in a resolving unit, the standard deviation of its spectral co-polarization is less than 0.05 or the standard deviation of its spectral differential phase is less than 15°, indicating that the ground object interference exhibits slowly varying polarization spectral characteristics. Therefore, the spectral range of ground object interference can be dynamically determined by using the standard deviations of spectral co-polarization and spectral differential phase.

[0056] The specific implementation process of step 3 includes the following sub-steps: Step 301: For each resolution unit, calculate the polarization spectral characteristics based on the original spectrum, including spectral co-polarization and spectral differential phase. Set the judgment condition for the gradual variation characteristic as follows: the standard deviation of spectral co-polarization is less than 0.05 or the standard deviation of spectral differential phase is less than 15°. If either condition is met, it is determined that the judgment condition for the gradual variation characteristic is met.

[0057] Step 302: Using the three data points centered at zero Doppler velocity as the current interval, calculate the standard deviation of spectral copolarization and spectral differential phase within the interval. When the above-mentioned slowly varying characteristic judgment condition is met, expand the current interval until the slowly varying characteristic judgment condition is no longer met. Then, the current interval is taken as the first spectral interval of ground object interference.

[0058] There are multiple ways to expand the current interval in this step, and this invention provides two of them.

[0059] Option 1: During expansion, expand symmetrically to both sides of the zero Doppler velocity center. In the first expansion, expand from 3 data points to 5 data points. Each expansion checks the conditions for gradual variation; if the conditions are met, expansion continues; otherwise, it stops, and the current interval is designated as the first spectral interval for ground object interference.

[0060] Option 2: During expansion, first expand towards the side with zero Doppler velocity. After finding the boundary, expand to the other side. Specifically, first expand towards the side with zero Doppler velocity. After each expansion, determine whether the current interval meets the condition for gradual change characteristics. If it does, continue expanding in the current direction. If not, use the boundary of the interval before this expansion as the current side boundary. Expand towards the other side with zero Doppler velocity, and after each expansion, determine whether the current interval meets the condition for gradual change characteristics, until the condition is no longer met. At this point, the center of the current interval may deviate from zero Doppler velocity, not due to ground object interference. Therefore, the current interval is not the final result and can only be considered a provisional interval. Further deviation assessment is needed. When determining deviation, the deviation between the symmetry center of the provisional interval and the zero Doppler velocity is calculated. If the deviation is less than or equal to twice the spectral resolution, it indicates that the spectrum of the first spectral interval is generated by ground object interference, and the provisional interval is taken as the first spectral interval of ground object interference. Otherwise, if the deviation of the provisional interval is too large, it may not be generated by ground object interference, and therefore cannot be suppressed. In this case, the spectral components of the provisional interval are not suppressed, that is, there is no first spectral interval. Then in step 303... It will not be updated.

[0061] Step 303: Based on the first spectral range, apply the first spectral mask. The corresponding spectral components in the image are set to 0; steps 301-303 are performed on each resolution cell to obtain a second spectral mask of the range-Doppler spectrum after filtering based on the slowly varying characteristics of the ground object interference polarization spectrum. .

[0062] Figure 3 This diagram illustrates the ground cover interference suppression process based on the slowly varying spectral characteristics in step 3 of this paper.

[0063] Step 4: Suppress ground feature interference based on ground feature detection indicators.

[0064] In this step, the ground object detection index is calculated based on the original spectrum. The ground object detection index of ground object interference is higher than the set characteristic to determine the second spectrum interval of ground object interference. Ground object interference is suppressed on the second spectrum mask to obtain the third spectrum mask.

[0065] The ground feature detection indicators in this embodiment include: clutter phase consistency and spectral power ratio.

[0066] The formulas for calculating clutter phase consistency and spectral power ratio are as follows:

[0067]

[0068] in, For clutter phase consistency, The raw IQ data, To distinguish the total number of echoes within a cell, m For echo indexing; For the percentage of spectrum power, Power spectrum, For the entire frequency range, v Doppler velocity; For the frequency spectrum of ground object interference:

[0069]

[0070] in, For antenna scanning speed, For wavelength, Antenna elevation angle The beamwidth is 3dB in unidirectional power mode. This refers to the radar dwell time.

[0071] Verified by a large amount of actual measurement data. or A value greater than 0.5 indicates the presence of ground-based interference in the spectrum. Therefore, this step can be selected... or One of them serves as a ground feature detection indicator for subsequent operations. Alternatively, you can choose... and Together they serve as ground feature detection indicators for subsequent operations. If selected... and Then, a spectral interval with ground clutter interference is obtained, and the two intervals are combined to perform mask update.

[0072] When choosing and In this process, the specific steps included are as follows: Step 401: For each resolution unit, according to Determine whether unsuppressed power spectra are still retained in the three data points centered at zero Doppler velocity; if so, calculate clutter phase consistency for the current resolution cell. and spectrum power ratio Filter out or The spectral range greater than 0.5 is considered as the spectrum where ground clutter interference exists—the ground clutter spectrum.

[0073] Step 402: Using the three data points centered on the zero Doppler velocity as the current region, if the power value at the boundary of the current region is greater than the power value of the adjacent point outside the boundary and the difference is greater than or equal to 5dB, then expand the current region until the condition can no longer be met, and obtain a provisional interval; Step 403: Extract the data in the provisional interval from the ground object interference spectrum determined in step 401, perform Gaussian fitting, and obtain the Gaussian fitted spectrum corresponding to the ground object interference.

[0074] Since Gaussian fitting requires positive input data, the power values ​​need to be positiveized as follows before fitting:

[0075] In the formula, This is the power value after positive conversion. This is the original power value. The spectral noise estimated by Poisson fitting in step 2, The positions of the spectral lines used for modeling.

[0076] Step 404: Cancel the original spectrum with the Gaussian fitted spectrum of ground object interference: Take the position on both sides of zero Doppler velocity where the power value of the original spectrum is stronger than the power value of the Gaussian fitted spectrum for the first time as the boundary to form the second spectrum interval of ground object interference.

[0077] Step 405: Apply the second spectral mask according to the second spectral range. The corresponding spectral components are set to 0; steps 401-405 are performed on each resolution unit to obtain the third spectral mask of the range-Doppler spectrum. .

[0078] Figure 4 This diagram illustrates the ground object interference suppression process based on clutter modeling cancellation in step 4 of this paper.

[0079] Step 5: Suppression of meteorological disturbances based on statistical data.

[0080] This step establishes a Bayesian classification model based on statistical data of the polarization spectral characteristics of meteorological disturbances and animal targets. It combines the continuity difference of the polarization spectral characteristics of the two to distinguish the categories of each preserved spectral band in the third spectral mask and suppress meteorological disturbances and preserve animal targets.

[0081] The polarization spectral features used in this embodiment include spectral copolarization, spectral differential phase, and spectral differential reflectance.

[0082] Based on a large amount of statistical data on polarization spectral features, a Bayesian classification model is established to distinguish between meteorological disturbances and animal targets. Since the motion patterns of meteorological disturbances and animal targets differ significantly, and the physical states of particles in meteorological disturbance clouds are significantly similar to those of animal targets, there is a continuous difference in polarization spectral features between the two. A sliding window is used to calculate the mean value of the features within the distance-Doppler spectral window region; changes in the magnitude of the mean value can characterize this continuous difference.

[0083] Taking a 3×3 sliding window as an example, the formula for calculating the feature mean is as follows:

[0084] in, The characteristic mean of the polarization spectrum; For distance position, Doppler spectrum; Representing data points polarization spectrum characteristics, These are the variables used in addition calculations.

[0085] The specific implementation process of this step includes the following sub-steps: Step 501: For each data point d in the range-Doppler spectrum, calculate the mean of three polarization spectral features and the mean of three polarization spectral features within a sliding window centered on data point d. The three polarization spectral features constitute the polarization spectral feature set. The three means form a set of means. .

[0086] Step 502: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full context.] and Input them into the Bayesian classification model respectively, and obtain They belong to meteorological disturbances. and animal targets probability and ,as well as They belong to meteorological disturbances. and animal targets probability and .

[0087] The Bayesian classification model is as follows:

[0088] In the formula, Given a set of features Time belongs to category The general idea, ; This is a dataset of three polarization spectral features; It is a dataset of the means of three polarization characteristics. For set Element. For category; where, Due to meteorological interference, For animal targets.

[0089] Step 503: Calculate the screening criteria.

[0090] The formula for calculating the screening indicators is:

[0091] Will and Add them together to obtain the meteorological interference. Screening criteria ; Will and Add them together to obtain the animal target. Screening criteria .

[0092] Step 504: Determine the type of data points based on the screening criteria.

[0093] Due to the difference in continuity between meteorological disturbances and animal targets, the mean value of the polarization spectrum characteristics of meteorological disturbances varies relatively little. Therefore, when the mean value varies little, Greater than This indicates that data point d is affected by meteorological interference; conversely, when the degree of change is large, Less than This indicates that data point d represents the animal target.

[0094] This step involves determining the point.

[0095] Step 505: For the third spectral mask For each retained spectral band, the proportion of data points in the two categories of meteorological disturbance and animal targets is counted, and the category with the larger proportion is the category of the retained spectral band.

[0096] In practice, if the proportions of the two types of data points are equal, the category with the highest proportion of the three data points near the maximum peak of the retained spectral band is selected as the category of the retained spectral band. Step 506: Apply the third spectral mask By suppressing the spectral bands that retain meteorological interference, a fourth spectral mask is obtained that retains only animal targets. Finally, the fourth spectral mask is used. The original spectrum was subjected to spectrum extraction to obtain spectrum data that retained only the animal target.

[0097] Figure 5This is a schematic diagram illustrating the differentiation between meteorological disturbances and migratory animals.

[0098] This concludes the process.

[0099] The specific embodiments described above only illustrate the design principles of the present invention. The shapes and names of the components in this description may differ and are not limited. Therefore, those skilled in the art can modify or make equivalent substitutions to the technical solutions described in the foregoing embodiments; and these modifications and substitutions do not depart from the inventive spirit and technical solutions of the present invention, and should all fall within the protection scope of the present invention.

Claims

1. A method for controlling the IQ data quality of weather radar for monitoring animal migration, characterized in that, include: Step 1: Calculate the power spectrum based on weather radar IQ data as the raw spectrum; Step 2: For a single resolution cell, estimate the spectral noise level based on the power distribution of each spectral component in the original spectrum; suppress the spectral components of background noise based on the spectral noise level to generate an original spectral mask; the unsuppressed spectral components in the original spectral mask form multiple preserved spectral bands; the original spectral masks of each resolution cell constitute the first spectral mask of the distance-Doppler spectrum. Step 3: Calculate the polarization spectrum characteristics based on the original spectrum, utilize the gradual variation characteristics of the polarization spectrum characteristics of ground object interference to determine the first spectrum interval of ground object interference, suppress ground object interference on the first spectrum mask, and obtain the second spectrum mask. Step 4: Calculate the clutter phase consistency index based on the original spectrum. Utilize the characteristic that the clutter phase consistency index of ground object interference is higher than the set clutter threshold to determine the second spectral range of ground object interference. Perform ground object interference suppression on the second spectral mask to obtain the third spectral mask. Step 5: Based on the statistical data of the polarization spectral characteristics of meteorological interference and animal targets, establish a Bayesian classification model. Combining the continuity difference of the polarization spectral characteristics of the two, distinguish the categories of each preserved spectral band in the third spectral mask, and suppress meteorological interference and preserve animal targets.

2. The weather radar IQ data quality control method for animal migration monitoring as described in claim 1, characterized in that, After generating the original spectral mask in step 2, the method further includes: Based on the original spectral mask, the spectral peak value of each retained spectral band is determined. Combined with the spectral peak value, it is determined whether there is spectral leakage in the current resolution cell. If there is spectral leakage, all spectral components in the current resolution cell are suppressed and the original spectral mask is updated; otherwise, the original spectral mask is not updated.

3. The weather radar IQ data quality control method for animal migration monitoring as described in claim 2, characterized in that, Step 2 specifically includes: Step 201: For a single resolution cell, the original spectrum is reordered to obtain the rearranged spectrum; the rearranged spectrum is plotted with power value on the vertical axis and the proportion of spectral components with power values ​​less than the vertical axis on the horizontal axis; the power values ​​in the rearranged spectrum are made positive. Step 202: Extract data points from the power value flattening region of the rearranged spectrum, perform Poisson fitting, and inversely positiveize the fitted data to obtain the fitted spectrum; the rearranged spectrum is considered to be higher than the fitted spectrum and the difference between the two is greater than a set level value for the first time. The power value at that location represents the spectral noise level. Step 203: Based on the aforementioned spectral noise level, calculate the signal-to-noise ratio (SNR) of each spectral component in the original spectrum of each resolution unit, and compare it with the SNR threshold. Comparison; above the signal-to-noise ratio threshold The spectral components are set to 1, not exceeding the signal-to-noise ratio threshold. The spectral components are set to 0 to obtain the original spectral mask. Original spectral mask The spectral components set to 1 form multiple retained spectral bands; Step 204: Locate the spectral peaks of each retained spectral band in the original spectrum; determine whether the spectral peaks on both sides of the zero Doppler velocity are symmetrical, and whether the standard deviation of the spectral peaks is less than the set power standard deviation threshold. If so, it is determined that there is spectral leakage, and the mask value corresponding to all spectral components in the current resolution unit is set to 0, and the original spectral mask is updated.

4. The weather radar IQ data quality control method for animal migration monitoring as described in claim 3, characterized in that, In step 204, if it is determined that there is no spectral leakage, then based on the signal-to-noise ratio of each spectral component calculated in step 203, the retained spectral bands with a peak signal-to-noise ratio of less than 10dB are suppressed.

5. The weather radar IQ data quality control method for animal migration monitoring as described in claim 3, characterized in that, In step 204, the determination of whether the spectral peaks on both sides of the zero Doppler velocity are symmetrical is as follows: Symmetry judgment includes positional symmetry and quantitative symmetry; The criterion for determining positional symmetry is as follows: select a spectral peak position on each side of the zero Doppler velocity and find the midpoint position; if the difference between the midpoint position and the zero Doppler velocity is less than a set distance threshold, then the position is determined to be symmetrical. The criterion for determining quantity symmetry is: if the difference in the number of spectral peaks on both sides of zero Doppler velocity is less than a set difference threshold, then it is determined to be quantity symmetric.

6. The weather radar IQ data quality control method for animal migration monitoring as described in claim 1, characterized in that, Step 3 specifically includes: Step 301: For each resolution unit, calculate the polarization spectrum characteristics based on the original spectrum, including spectral co-polarization and spectral differential phase; set the judgment condition for the slow variation characteristics as the standard deviation of spectral co-polarization being less than 0.05 or the standard deviation of spectral differential phase being less than 15°; Step 302: Using the three data points centered on zero Doppler velocity as the current interval, calculate the standard deviation of spectral copolarization and spectral differential phase within the interval. When the slowly varying characteristic judgment condition is met, expand the current interval until the slowly varying characteristic judgment condition is no longer met. Then, the current interval is taken as the first spectral interval of ground object interference. Step 303: Based on the first spectral range, apply the first spectral mask. The corresponding spectral component in the value is set to 0; Each resolution cell is processed using steps 301-303 to obtain the second spectral mask of the range-Doppler spectrum. .

7. The weather radar IQ data quality control method for animal migration monitoring as described in claim 6, characterized in that, The expansion of the current interval in step 302 is as follows: It expands symmetrically to both sides of zero Doppler velocity; Alternatively, the range can be expanded towards the zero Doppler velocity side first. After each expansion, it is determined whether the current range meets the condition for judging the gradual change characteristics. If it does, the expansion in the current direction continues. If it does not, the boundary of the range before this expansion is used as the current side boundary, and the range is expanded towards the other side of the zero Doppler velocity side. After each expansion, it is determined whether the current range meets the condition for judging the gradual change characteristics. This process continues until the condition for judging the gradual change characteristics is no longer met, at which point the current range is considered a provisional range. Calculate the deviation between the center of symmetry of the provisional interval and the zero Doppler velocity. If the deviation is less than or equal to twice the spectral resolution, it indicates that the spectrum of the provisional interval is generated by ground object interference, and the provisional interval is taken as the first spectral interval of the ground object interference. Otherwise, the spectral components in the provisional interval are not suppressed.

8. The weather radar IQ data quality control method for animal migration monitoring as described in claim 1, characterized in that, Step 4 further includes: calculating the spectral power ratio index based on the original spectrum; utilizing the characteristic that the spectral power ratio index of ground object interference is higher than the set power ratio, determining the spectral range of ground object interference considering the spectral power ratio index, and combining it with the spectral range of ground object interference determined based on the clutter phase consistency index to form the second spectral range, suppressing ground object interference on the second spectral mask, and obtaining the third spectral mask.

9. The weather radar IQ data quality control method for animal migration monitoring as described in claim 8, characterized in that, In step 4, the third spectral mask is obtained as follows: Step 401: Use clutter phase consistency index and spectral power ratio index as ground feature detection indicators; for each resolution cell, according to the second spectral mask... Determine whether there is still an unsuppressed power spectrum within a frequency range of a set length centered on the zero Doppler velocity; if so, calculate the ground object detection index for the current resolution unit, and select the spectrum with the ground object detection index greater than 0.5 as the ground object interference spectrum. Step 402: Take the three data points centered on the zero Doppler velocity as the current region. If the power value at the boundary of the current region is greater than the power value of the adjacent point outside the boundary and the difference is greater than or equal to 5dB, then expand the current region until the condition can no longer be met, and obtain a provisional interval. Step 403: Extract the data in the provisional interval from the ground object interference spectrum determined in step 401, perform Gaussian fitting, and obtain the Gaussian fitted spectrum corresponding to the ground object interference; Step 404: The second spectral interval of ground object interference is formed by taking the position on both sides of the zero Doppler velocity where the original spectral power value is stronger than the Gaussian fitted spectral power value for the first time. Step 405: Apply the second spectral mask according to the second spectral range. The corresponding spectral components are set to 0; steps 401-405 are performed on each resolution unit to obtain the third spectral mask of the range-Doppler spectrum. .

10. The weather radar IQ data quality control method for animal migration monitoring as described in claim 1, characterized in that, Step 5 specifically includes: For each data point d in the range-Doppler spectrum, calculate the polarization spectral characteristics. and the mean polarization spectrum characteristics within a sliding window centered on data point d The polarization spectrum characteristics This includes one or more combinations of spectral copolarization, spectral differential phase, and spectral differential reflectivity; Polarization spectrum characteristics and the corresponding polarization spectrum characteristic mean Inputting each data point into a Bayesian classification model yields polarization spectral features. They belong to meteorological disturbances. and animal targets probability and and the mean value of polarization spectrum characteristics They belong to meteorological disturbances. and animal targets probability and ; probability and Add them together to obtain the meteorological interference. Screening criteria ; probability and Add them together to obtain the animal target. Screening criteria ; Screening indicators Greater than This indicates that data point d is subject to meteorological interference; conversely, it indicates that data point d is a target animal. For each retained spectral band in the third spectral mask, the proportion of data points in the two categories of meteorological interference and animal targets is counted, and the category with the larger proportion is the category of the retained spectral band; the retained spectral band of meteorological interference in the third spectral mask is suppressed to obtain the fourth spectral mask that retains only animal targets; The original spectrum was extracted using a fourth spectral mask to obtain spectral data that retained only the animal target.

Images