Sub-mesoscale signal extraction method based on ku / ka dual-frequency sar altimetry

Abstract

The present application belongs to the technical field of physical ocean and remote sensing application, in particular to a sub-mesoscale signal extraction method based on Ku / Ka dual-frequency SAR height measurement, comprising the following steps: obtaining Ku / Ka dual-frequency sea surface height observation original data set and IMU attitude and trajectory data, registering average sea surface and tidal height to the space-time sampling position of Ku / Ka synchronous observation and deducting from the original height, obtaining two-channel initial sea surface height anomaly sequence, and processing the two-channel initial sea surface height anomaly data to obtain a fused height anomaly sequence; performing attitude coupling mode separation on the fused height anomaly sequence to obtain a height anomaly sequence after removing platform error; simultaneously performing smoothing processing to obtain a smoothed sequence; interpolating the smoothed sequence to a two-dimensional regular grid, performing small-scale secondary smoothing denoising on the grid data, and deducting the mesoscale background field at the same time after interpolating the same grid to obtain a data system of sea surface height anomaly mainly in sub-mesoscale.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of physical oceanography and remote sensing application, in particular to a sub-mesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry. BACKGROUND

[0002] Ocean sub-mesoscale processes generally refer to dynamic phenomena with spatial scales of several kilometers to several tens of kilometers and time scales of several hours to several days, mainly including fine-scale vortex structures, fronts, and frontogenesis zones, etc. Such processes not only have small scales and rapid changes, but also have significant nonlinear and non-stationary characteristics, playing an important role in the vertical transport of ocean heat, salinity, nutrients, and other substances, as well as the exchange of energy between the ocean and the atmosphere. Accurate observation and identification of sub-mesoscale processes not only help to improve ocean-climate coupling models and improve prediction reliability, providing key support for shipping, emergency response, and sea defense warning, but also directly relate to the spatiotemporal migration of fishery resources and the health of nearshore ecosystems, having important practical significance for fishery management, aquaculture, and ecological protection.

[0003] The main means of obtaining sub-mesoscale information currently include in-situ observation, satellite observation, and airborne observation, but all have the contradiction between resolution, coverage, and availability:

[0004] 1) In-situ observation (walk-through section, anchor, glider, drifting buoy, etc.) has advantages in local fine structure analysis, but is limited by sea conditions, ship schedules, and deployment density, with insufficient spatial coverage and timeliness, and high cost and maintenance pressure.

[0005] 2) Traditional satellite altimeters have the advantages of long-term continuity and global coverage, suitable for mesoscale (vortex, planetary wave, etc.) monitoring; however, due to observation geometry, repetition period, and pulse echo bandwidth constraints, the sensitivity to structures below the mesoscale is limited. More importantly, in subsequent quality control and denoising processing, to ensure long-term stability, high wave number components are often weakened or filtered out, making it difficult for public products to stably present sub-mesoscale signals. Although the launch of the SWOT satellite is expected to observe sub-mesoscale processes, at present, due to on-orbit calibration and verification, data release delay, and the maturity of assimilation applications, it is still difficult to continuously analyze rapidly evolving sub-mesoscale processes in a business-oriented manner.

[0006] 3) Airborne observation systems have high temporal and spatial resolution, maneuverability, and repeatable coverage, and are one of the key ways to analyze sub-mesoscale dynamics. However, airborne data is complex: on the one hand, platform attitude changes and height fluctuations introduce system errors that can be comparable to or even exceed the target signal magnitude, easily masking the true dynamic signal; on the other hand, sub-mesoscale signals are often overwhelmed by larger-scale background fields and random noise, making direct identification and quantitative analysis challenging.

[0007] In summary, although airborne observation systems have unique potential in obtaining submesoscale dynamic processes, their effectiveness is limited by data quality control and processing capabilities.

[0008] Ku / Ka dual-frequency SAR altimetry can provide additional physical dimensions for platform attitude error identification, noise modeling, and target feature enhancement due to complementary frequency bands and differences in scattering mechanisms. If a method system for dual-frequency collaborative processing and multi-scale component separation is constructed for airborne scenarios, closed-loop optimization can be achieved in the whole chain of "error suppression-target extraction-product generation", which can be expected to stably recover and present submesoscale signals with high fidelity. In the current public information, the systematic processing framework and reusable products for airborne dual-frequency SAR altimetry are still insufficient; at the same time, under the premise of not damaging the submesoscale structure and statistical characteristics, effectively stripping large-mesoscale background, small-scale perturbations, and random noise is still a key problem that needs to be solved. SUMMARY

[0009] In view of the above problems of the prior art, the present application provides a submesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry, which is based on Ku / Ka dual-frequency SAR airborne observation data, constructs an integrated data processing chain, accurately and stably extracts submesoscale signals and generates high-resolution data products, and is matched with an application system of data products, providing strong data support for the research of submesoscale dynamic processes.

[0010] The technical scheme adopted by the present application to achieve the above-mentioned purpose is: a submesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry, characterized by comprising the following steps:

[0011] S1: obtaining Ku / Ka dual-frequency sea surface height observation original data sets and IMU attitude and trajectory data synchronously collected by an airborne platform;

[0012] S2: according to observation time and latitude and longitude, the average sea surface MSS and the tidal height are matched to the spatio-temporal sampling position of Ku / Ka synchronous observation and are deducted from the original height to obtain the initial sea surface height anomaly sequence of Ku / Ka two channels

[0013] S3: using the BUCG-Alt dual-frequency collaborative processing method, the Ku / Ka initial sea surface height anomaly data are sequentially subjected to adaptive spike anomaly suppression, local system bias correction, uncertainty weighted fusion, and dual-frequency consistency quality control to obtain a fused height anomaly sequence;

[0014] ​S4: Using the attitude-coupled mode extraction ACME method, the fused height anomaly sequence is subjected to empirical mode decomposition to obtain intrinsic mode components, and according to the correlation between the intrinsic mode components and the platform trajectory / attitude change signal, the platform error mode is identified and removed, and the height anomaly sequence after removing the platform error is obtained;

[0015] S5: The height anomaly sequence after removing the platform error is subjected to smoothing processing in a given sub-mesoscale constraint window to obtain a smoothed sequence;

[0016] S6: The smoothed sequence is interpolated to a two-dimensional regular grid, the grid data is subjected to small-scale secondary smoothing and denoising, and the mesoscale background field at the same time is interpolated to the same grid and then subtracted to obtain a data system of sea surface height anomalies mainly in sub-mesoscale.

[0017] The step S2 comprises the following steps:

[0018] S21: Using bilinear interpolation, the MSS data is mapped to the space-time sampling position along the track to obtain a registered MSS sequence ;

[0019] S22: Using the TPXO9 global tidal model, the tidal height at each observation point along the track is calculated in longitude, latitude and time, and the registered tidal height is obtained ;

[0020] S23: Removing the MSS and tidal height components from the original height data of the Ku observation channel:

[0021]

[0022] Removing the MSS and tidal height components from the original height data of the Ka observation channel:

[0023]

[0024] wherein, represents the long-term average state of the sea surface height, is the tidal height;

[0025] Finally, the initial sea surface height anomaly of the Ku / Ka two channels is obtained .

[0026] In step S3, the BUCG-Alt dual-frequency collaborative processing method specifically performs the following steps:

[0027] S31: Adaptive spike anomaly suppression: in the sliding window W(k) centered at sample k with length L, when sample point |x(k) - mean(W(k))| > Tspike, the sample is judged as spike anomaly and replaced by the adjacent mean; wherein, mean is the mean function, Tspike is the spike discrimination threshold;

[0028] S32: Local system bias estimation and correction: in the sliding window W(k) after spike suppression, the local system bias of two channels is estimated by the median, that is:

[0029]

[0030] wherein, sampling point , is the local system bias in the two-channel sliding window W(k), unit m, median is the median function, respectively, Ku channel and Ka channel sea surface height anomaly, unit m;

[0031] Correct any one channel to align to a unified reference to obtain the bias-corrected channel:

[0032]

[0033]

[0034] wherein, is the bias-corrected Ku channel, is the bias-corrected Ku channel, , respectively, the initial sea surface height anomaly of two channels;

[0035] S33: Uncertainty weighted fusion: calculate the standard deviation in the two-channel sliding window W(k) to obtain the internal random error intensity ; and obtain the weighted coefficient in the form of inverse variance, that is:

[0036] ,

[0037] wherein, is the weighted coefficient of Ku channel, dimensionless; is the weighted coefficient of Ka channel, dimensionless;

[0038] S34: respectively weight the two-channel data to obtain the fusion data as:

[0039] ​​

[0040] wherein, is the height anomaly result of two-channel weighted fusion within the sliding window W(k);

[0041] S34: dual-frequency consistency quality control, i.e., quality control according to consistency of dual-frequency data:

[0042] If , the fusion value within W(k) is replaced by the one with smaller random error, or marked as invalid; wherein, is a maximum value function, and the threshold value is adaptively set between 1.5 and 3 according to the local noise level.

[0043] The step S4 is a pose-coupled mode extraction (ACME) method, specifically:

[0044] S41: performing empirical mode decomposition on the fused sea surface height anomaly sequence, i.e., the fused data , to adaptively decompose it into n intrinsic mode function components and a residual trend item , which is expressed as:

[0045]

[0046] wherein, t is the time series along the track, j is the serial number of the IMF, and n is the total number of the decomposed IMFs;

[0047] S42: calculating the correlation between each IMF and the platform track / pose change signal;

[0048] extracting the time series of platform motion parameters provided by the onboard platform inertial measurement unit , wherein the platform motion parameters include pitch angle, roll angle or elevation change;

[0049] calculating the Pearson correlation coefficient between each intrinsic mode function and each type of platform motion parameter sequence , which is:

[0050]

[0051] wherein, is the mean value of the jth IMF component, is the mean value of the platform motion parameter ;

[0052] S43: setting a correlation threshold Tc;

[0053] If there exists , the IMF component is marked as a component to be removed.

[0054] S44: Set all IMF components marked as platform error modes to zero, and reconstruct the sea surface height anomaly sequence after removing platform errors using the remaining IMF components and the residual trend term , and the calculation formula is:

[0055]

[0056] , where E represents the serial number set of all IMF components marked as errors.

[0057] In step S5, the sub-mesoscale constraint window is a spatial scale of 5-50 km selected according to the data resolution requirement.

[0058] The smoothing method is any one of a sliding average or a weighted Gaussian smoothing.

[0059] In step S6, the method comprises the following steps:

[0060] S61: Spatially interpolate the height anomaly sequences corresponding to multiple flight lines after the processing in steps S2-S5 to a regular two-dimensional grid, and for high-resolution airborne observation data, the grid size is selected to be 2 km x 2 km.

[0061] S62: Secondary small-scale noise removal: Perform two-dimensional smoothing processing of the grid data in an m x n window to remove small-scale noise possibly introduced by gridding.

[0062] S63: Mesoscale background field removal: Interpolate the mesoscale data set of the AVISO / CMEMS satellite altimeter grid observation at the same time to the same spatial and temporal resolution as the grid to serve as a mesoscale background field, and subtract the smoothed grid data to obtain a sea surface height anomaly mainly in sub-mesoscale, thereby obtaining a data system of a sea surface height anomaly mainly in sub-mesoscale.

[0063] A system of a Ku / Ka dual-frequency SAR altimetry sub-mesoscale signal extraction method comprises:

[0064] A data acquisition module is configured to acquire Ku / Ka dual-frequency sea surface height observation raw data synchronously collected by an airborne platform and platform attitude and trajectory data provided by an IMU.

[0065] A preprocessing module is configured to perform average sea surface and tidal height registration and subtraction on the raw height data to generate initial sea surface height anomaly sequences in Ku and Ka dual channels.

[0066] ​The dual-frequency cooperative processing module is configured to execute a BUCG-Alt processing method, including sequentially performing adaptive spike suppression, local system bias correction, uncertainty weighted fusion and dual-frequency consistency quality control, and outputting the fused height anomaly sequence;

[0067] The platform error elimination module is configured to perform empirical mode decomposition on the fused sequence, extract an intrinsic mode function, and calculate a correlation in combination with platform attitude / trajectory data, and eliminate an error mode with a correlation higher than a set threshold value;

[0068] The smoothing processing module is configured to perform smoothing processing on the height anomaly sequence after error elimination within a sub-mesoscale constraint window, and suppress high-frequency noise and small-scale disturbances.

[0069] The gridding and product generation module is configured to interpolate the smoothed data to a two-dimensional regular grid, implement secondary small-scale denoising and mesoscale background field subtraction, and finally generate a sub-mesoscale-based sea surface height anomaly data system.

[0070] The output module is configured to store and output the generated data system.

[0071] The generated data system is configured to perform a sub-mesoscale vortex identification and tracking method on the two-dimensional sub-mesoscale sea surface height anomaly data, including the following steps:

[0072] M1: reading the two-dimensional sea surface height anomaly data;

[0073] M2: estimating a strain feature of an anomaly field based on a curvature parameter, locking a sub-mesoscale event high-occurrence area, forming a vortex candidate along a closed contour in the area, and screening according to an amplitude threshold value and a spatial threshold value;

[0074] M3: pairing and time tracking between adjacent time instances according to a center distance and a contour overlap area ratio, and screening according to a duration threshold value; outputting a center position, a radius and a life cycle attribute of the vortex, and generating a visualization and a data file.

[0075] Step M2 specifically includes:

[0076] a. calculating a curvature parameter of the sea surface height anomaly field, including eigenvalues of a strain rate tensor, identifying an area with a large strain rate as a sub-mesoscale event high-occurrence area;

[0077] b. taking a local extreme point as a center, growing outward along a closed contour to form a vortex candidate area in the high-occurrence area;

[0078] c. setting an amplitude threshold value and a spatial scale threshold value to preliminarily screen the candidate vortex;

[0079] d. Record the center position, radius, maximum anomaly amplitude, etc. of each candidate vortex.

[0080] Step M3, specifically:

[0081] a) Spatiotemporal matching of vortex candidates at two consecutive times: if the distance between the centers of the two vortexes is less than a certain set proportion of the sum of their radii, and the contour overlap area ratio is greater than 50%, it is determined that it is the same vortex at different times;

[0082] b) Establishing a vortex trajectory chain and recording its appearance, duration and disappearance time;

[0083] c) Setting a duration threshold, if at least three consecutive times appear, eliminating false vortexes that appear temporarily;

[0084] d) Outputting the life cycle, movement path and intensity change information of each effective vortex, and generating a visual result and a data file.

[0085] The generated data system is also used for performing a sub-mesoscale front identification and tracking method on two-dimensional sea surface height anomaly data, comprising the following steps:

[0086] N1: reading two-dimensional sea surface sub-mesoscale height anomaly data;

[0087] N2: calculating the spatial gradient amplitude and strain rate tensor eigenvalue of the anomaly field, and extracting linear front candidates meeting the gradient amplitude threshold and connectivity constraint;

[0088] N3: screening and tracking the candidates based on direction consistency and time continuity, and outputting front position line, length, intensity and evolution attributes.

[0089] The present application has the following beneficial effects and advantages:

[0090] 1. Robust error correction, improving precision and consistency: The present application relies on the complementary characteristics of Ku / Ka dual frequency and BUCG-Alt method, combined with MSS / tide accurate registration and ACME (attitude coupling modal separation method) error separation mechanism, to suppress non-target disturbances introduced by platform attitude / trajectory, realize the alignment, fusion and strict quality control of two-channel height anomaly, and significantly improve the stability and repeatability of the results.

[0091] 2. High-fidelity sub-mesoscale extraction, significantly enhancing signal distinguishability: The present application uses a combination strategy of multi-scale component analysis, sub-mesoscale constraint window smoothing and "second background field removal", effectively removing large / mesoscale background, random noise and small-scale disturbances, improving signal-to-noise ratio and preserving the boundaries and details of sub-mesoscale structures such as vortices and fronts, without damaging their nonlinear and non-steady-state characteristics, achieving high-fidelity recovery of target signals.

[0092] 3. Standardized product output, supporting scientific research and business application: the application uniformly interpolates the processing results to a two-dimensional regular grid to form a high-resolution sub-mesoscale sea surface height anomaly data product for the observed sea area, and gives a product application demonstration; the product can be directly used for sub-mesoscale vortex and front identification and tracking, model assimilation and verification, and business monitoring and early warning, etc., the processing chain is standardized, reusable and extensible, and is convenient for popularization to other sea areas and platforms, and has significant application value and portability. BRIEF DESCRIPTION OF DRAWINGS

[0093] Figure 1 A flowchart of the sub-mesoscale signal extraction method and data product application of the present patent;

[0094] Figure 2 A schematic diagram of the flight line distribution of the airborne measured data of the embodiment of the present patent;

[0095] Figure 3 A schematic diagram of the partial initial sea surface height anomaly of the two channels of each flight line of the embodiment of the present patent;

[0096] Figure 4 A schematic diagram of the correlation between the platform error mode and the trajectory / attitude change signal of the embodiment of the present patent;

[0097] Figure 5 A schematic diagram of the comparison between the data processing results and the AVISO height data product of the embodiment of the present patent;

[0098] Figure 6 A schematic diagram of the sub-mesoscale vortex identification and tracking results of the data product application of the embodiment of the present patent. DETAILED DESCRIPTION

[0099] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0100] In actual application, the sub-mesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry is as follows:

[0101] S1: Obtain the Ku / Ka dual-frequency sea surface height observation raw data set and the IMU attitude and trajectory data synchronously collected by the airborne platform, Figure 2 A schematic diagram of the flight line distribution of the airborne measured data of the embodiment of the present patent.

[0102] S2: Using bilinear interpolation to map MSS data provided by AVISO to the spatial-temporal sampling positions along the track, to obtain the registered MSS sequence ; Using all the tidal components of TPXO9 global tidal model to calculate the tidal height at each observation point along the track in longitude, latitude and time, to obtain the registered tidal height ;

[0103] Removing MSS and tidal height components from the original height data of Ku observation channel:

[0104]

[0105] Removing MSS and tidal height components from the original height data of Ka observation channel:

[0106]

[0107] wherein, represents the long-term average state of sea surface height, is the tidal height;

[0108] Finally, the initial sea surface height anomaly of Ku / Ka two channels is obtained , such as Figure 3 is the partial initial sea surface height anomaly of each route of the embodiment of the application.

[0109] S3: Using BUCG-Alt dual-frequency collaborative processing method, the Ku / Ka initial sea surface height anomaly data is quality controlled and fused.

[0110] Firstly, adaptive spike anomaly suppression is performed: when the sample point is within the sliding window W(k) centered on sample k and with a length of 180 sampling points, if −mean(W(k))| > , the sample is judged as spike anomaly and replaced by the adjacent mean value, wherein mean is the mean function, and 2 times of the standard deviation in W(k) is taken.

[0111] Then, the local system bias of two channels is estimated by median and the Ku channel is corrected to obtain the reference-aligned two-channel height anomaly , :

[0112]

[0113]

[0114] ​​

[0115] wherein the sampling points , are the bias-corrected Ku channel, are the bias-corrected Ku channel, , are the initial sea level anomaly of the two channels respectively , is the local system bias (m) within the sliding window W(k) of the two channels, median is the median function, is the Ku / Ka channel sea level anomaly (m).

[0116] The inverse variance weight is calculated by the standard deviation within each channel W(k) and the two channels of data are weighted respectively to obtain the fusion data:

[0117] ,

[0118] wherein, is the weighting coefficient of the Ku channel, dimensionless; is the weighting coefficient of the Ka channel, dimensionless;

[0119]

[0120] In the formula, is the weighted fusion height anomaly result of the two channels within the sliding window W(k).

[0121] The dual-frequency consistency quality control is performed, that is, the quality control is performed according to the consistency of the dual-frequency data:

[0122] If , the fusion value within W(k) is replaced by the one with smaller random error, or marked as invalid. is the maximum value function, and the threshold value is adaptively set between 1.5 and 3 according to the local noise level.

[0123] S4: As shown in Figure 4 , the correlation between the platform error mode and the platform trajectory / attitude change signal is 0.84, which corresponds to the 6th to 8th intrinsic mode. The ACME method is used to reconstruct the mode of the fusion data: the EMD is performed on the fused sea level anomaly sequence to obtain the intrinsic mode components, and the correlation between each intrinsic mode and the platform trajectory / attitude change signal is obtained by using the canonical correlation analysis, and the mode with a correlation greater than or equal to 0.8 is regarded as the platform error mode and is removed. The specific process is as follows:

[0124] S41: Empirical mode decomposition is performed on the fused sea surface height anomaly sequence, i.e. the fused data , to adaptively decompose it into n intrinsic mode function components and a residual trend item , which is expressed as:

[0125]

[0126] where t is the time series along the track, j is the serial number of the IMF, and n is the total number of the decomposed IMFs;

[0127] S42: The correlation between each IMF and the platform trajectory / attitude change signal is calculated.

[0128] The time series of platform motion parameters provided by the onboard platform inertial measurement unit are extracted , including the pitch angle, roll angle or elevation change;

[0129] The Pearson correlation coefficient between each intrinsic mode function and each type of platform motion parameter sequence is calculated as:

[0130]

[0131] where is the mean value of the jth IMF component, is the mean value of the platform motion parameter ;

[0132] S43: A correlation threshold Tc is set.

[0133] All IMF components and all platform motion parameters are traversed, and if there exists , the intrinsic mode function component is marked as a component to be removed;

[0134] S44: All IMF components marked as platform error modes are set to zero, and the remaining IMF components and the residual trend item are used to reconstruct the sea surface height anomaly sequence after removing the platform error , and the calculation formula is:

[0135]

[0136] where E represents the serial number set of all IMFs marked as errors.

[0137] ​S5: The height anomaly sequence after removing the platform error is smoothed within a given sub-mesoscale constraint window to obtain a smoothed sequence; the sub-mesoscale constraint window is a spatial scale of 5-50 km selected according to data resolution requirements; and the smoothing method is any one of sliding average or weighted Gaussian smoothing.

[0138] In this embodiment, the data after removing the platform error is subjected to sliding average with a 5km constraint window to suppress high-frequency noise and small-scale disturbance.

[0139] Figure 5 In this patent embodiment, the data after step S5 processing is compared with AVISO height data (mesoscale) interpolated to the same track. It can be seen that the overall trend of the processed data is consistent with the AVISO data, indicating that the changes above the mesoscale are consistent, which shows that our processing method is effective. Further analysis shows that compared with the AVISO data, the observation data shows some fluctuation changes at the scale of several kilometers to tens of kilometers, which are sub-mesoscale signals that cannot be captured by the AVISO mesoscale data.

[0140] S6: The smoothed data is interpolated to a two-dimensional regular grid of 2km*2km, the grid data is subjected to 5*5 two-dimensional smoothing processing to remove small-scale noise; the satellite altimeter mesoscale data provided by CMEMS at the same time is interpolated to the same spatio-temporal resolution as the grid to serve as a mesoscale background field, and is subtracted from the smoothed grid data to obtain a sub-mesoscale-based sea surface height anomaly data system.

[0141] S61: The height anomaly sequence corresponding to the multiple flight lines after steps S2-S5 processing is spatially interpolated to a regular two-dimensional grid, and for high-resolution airborne observation data, the grid size is selected to be 2km*2km;

[0142] S62: Small-scale noise removal: the grid data is subjected to 5*5 window two-dimensional smoothing processing to remove small-scale noise that may be introduced during gridding;

[0143] S63: Mesoscale background field removal: the mesoscale data set of the AVISO / CMEMS satellite altimeter grid observation at the same time is interpolated to the same spatio-temporal resolution as the grid to serve as a mesoscale background field, and is subtracted from the smoothed grid data to obtain a sub-mesoscale-based sea surface height anomaly data system.

[0144] Based on a Ku / Ka dual-frequency SAR height measurement sub-mesoscale signal extraction method, the present application proposes a sub-mesoscale signal extraction system, comprising:

[0145] a data acquisition module, configured to acquire Ku / Ka dual-frequency sea surface height observation raw data synchronously collected by the airborne platform and platform attitude and trajectory data provided by the IMU;

[0146] a preprocessing module, configured to perform average sea surface and tidal height registration and deduction on the raw height data to generate initial sea surface height anomaly sequences of Ku and Ka dual channels;

[0147] a dual-frequency collaborative processing module, configured to perform a BUCG-Alt processing method, including sequentially performing adaptive spike suppression, local system bias correction, uncertainty weighted fusion and dual-frequency consistency quality control, and outputting the fused height anomaly sequence;

[0148] a platform error elimination module, configured to perform empirical mode decomposition on the fused sequence, extract intrinsic mode functions, and calculate correlations in combination with the platform attitude / trajectory data to eliminate error modes with correlations higher than a set threshold;

[0149] a smoothing processing module, configured to perform smoothing processing on the height anomaly sequence after error elimination within a sub-mesoscale constraint window to suppress high-frequency noise and small-scale disturbances;

[0150] a gridding and product generation module, configured to interpolate the smoothed data to a two-dimensional regular grid, perform secondary small-scale denoising and mesoscale background field deduction, and finally generate a sea surface height anomaly data system mainly in sub-mesoscale;

[0151] an output module, configured to store and output the generated data system.

[0152] In addition, the embodiment of the present application also provides a sub-mesoscale vortex identification method based on the sub-mesoscale sea surface height anomaly data system: based on the curvature parameter to estimate the strain characteristics of the anomaly field, lock the high-occurrence area of the sub-mesoscale event; form vortex candidates along the closed contour in the event high-occurrence area, and screen according to a 5cm amplitude threshold and a 30km spatial threshold (equivalent radius range);

[0153] Between adjacent time instances, according to a center distance threshold and a contour overlap area ratio threshold, pairing and time tracking are performed, and screening is performed according to a 3-5 day duration threshold, and the result is output. Figure 6 The sub-mesoscale vortex evolution schematic captured by the embodiment of the present application is shown in the following figure.

[0154] In the embodiment, based on the curvature parameter to estimate the strain characteristics of the anomaly field, lock the high-occurrence area of the sub-mesoscale event; form vortex candidates along the closed contour in the event high-occurrence area, and screen according to a 5cm amplitude threshold and a 30km spatial threshold (equivalent radius range); specifically including the following steps:

[0155] a. Calculate the curvature parameters of the sea surface height anomaly field, including the eigenvalues of the strain rate tensor, and identify the area with large strain rate as the high-risk area of submesoscale events;

[0156] b. In the high-risk area, take the local extreme point as the center and grow outward along the closed contour to form a vortex candidate area;

[0157] c. Set the amplitude threshold and spatial scale threshold to preliminarily screen the candidate vortex;

[0158] d. Record the center position, radius, maximum anomaly amplitude and other attributes of each candidate vortex.

[0159] In this embodiment, the center distance threshold and the contour overlap area ratio threshold are used for pairing and time tracking between adjacent time instances, and the results are output after being screened according to the duration threshold of 3-5 days. Specifically, the results are:

[0160] a) Spatiotemporal matching of vortex candidates in two consecutive time instances: if the distance between the centers of two vortexes is less than a certain set proportion of the sum of their radii, and the contour overlap area ratio is greater than 50%, it is determined that the same vortex is manifested in different time instances;

[0161] b) Establish a vortex trajectory chain to record the appearance, duration and disappearance time;

[0162] c) Set a duration threshold, and if at least three consecutive time instances appear, remove the false vortex that appears temporarily;

[0163] d) Output the life cycle, movement path and intensity change information of each effective vortex, and generate visual results and data files.

[0164] In addition, the embodiment of the present application also provides a submesoscale front identification and tracking method based on submesoscale sea surface height anomaly data system, comprising the following steps:

[0165] N1: Read two-dimensional sea surface submesoscale height anomaly data;

[0166] N2: Calculate the spatial gradient amplitude and strain rate tensor eigenvalue of the anomaly field, and extract linear front candidates that meet the gradient amplitude threshold and connectivity constraints;

[0167] N3: Screen and track the candidates based on direction consistency and time continuity, and output the front position line, length, strength and evolution attributes.

[0168] In conclusion, in combination with the embodiments of the present application, the present application overcomes the difficulties of coupling errors of high platform attitude and mixed aliasing of multi-scale noise in Ku / Ka dual-frequency SAR measurement, provides a new idea for high-fidelity extraction of sub-mesoscale signals and generation of standardized products, and fills the gap of airborne dual-frequency collaborative processing in sub-mesoscale stable productization.

[0169] In this specification, the present application has been described with reference to its specific embodiments. The above embodiments are the preferred embodiments of the present patent, and are not intended to limit the scope of the present application. It should be noted that the present application is not limited to the above specific embodiments, and any improvements, changes, combinations, substitutions and the like made by those skilled in the art without departing from the principles of the present application shall fall within the scope of the claims of the present application.

Claims

1. A submesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry, characterized in that, The method comprises the following steps: S1: obtaining Ku / Ka dual-frequency sea surface height observation raw data sets and IMU attitude and trajectory data synchronously collected by an airborne platform; S2: According to the observation time and the longitude and latitude, the average sea surface MSS and the tidal height are matched to the time and space sampling positions of the Ku / Ka synchronous observation and are subtracted from the original height to obtain the initial sea surface height anomaly sequence of the Ku / Ka two channels ; S3: using a BUCG-Alt dual-frequency collaborative processing method to sequentially implement adaptive spike anomaly suppression, local system bias correction, uncertainty weighted fusion and dual-frequency consistency quality control on the Ku / Ka initial sea surface height anomaly data, and obtaining a fused height anomaly sequence; S4: using an attitude-coupled mode extraction ACME method to perform empirical mode decomposition on the fused height anomaly sequence, obtaining intrinsic mode components, and identifying and removing platform error modes according to the correlation between the intrinsic mode components and the platform trajectory / attitude change signals, and obtaining a height anomaly sequence after removing platform errors; S5: performing smoothing processing on the height anomaly sequence after removing platform errors within a given sub-mesoscale constraint window, and obtaining a smoothed sequence; S6: interpolating the smoothed sequence to a two-dimensional regular grid, performing small-scale secondary smoothing and denoising on the grid data, and subtracting the mesoscale background field at the same time to obtain a data system of sub-mesoscale sea surface height anomalies.

2. The submesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry according to claim 1, characterized in that, The step S2 comprises the following steps: S21: mapping the MSS data to the spatio-temporal sampling locations along the track using bilinear interpolation to obtain a registered MSS sequence ; S22: using the tidal model, the tidal height at each observation point along the track is calculated using all the tidal components of the TPXO9 global tidal model at the longitude and latitude of each observation point along the track and at the time, to obtain the registered tidal height ; S23: Remove MSS and tidal height components from raw height data from Ku observation channel ;​ From raw height data of the Ka observation pass MSS and tidal height components are removed: ; wherein, represents a long-term average state of the sea surface height, is the tidal height; Ku / Ka two-channel initial sea surface height anomaly is obtained finally .

3. The submesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry according to claim 1, characterized in that, In step S3, the BUCG-Alt dual-frequency collaborative processing method specifically performs the following steps: S31: Adaptive spike anomaly suppression: In a sliding window W(k) centered at sample k with length L, when sample point |xk−mean(W(k))|>δs, is replaced by the adjacent mean value; where mean is the mean function, is the spike discrimination threshold.​ S32: local system bias estimation and correction: in the sliding window W(k) after spike suppression, the local system bias of the two channels is estimated by the median, that is: ； where the sampling points , is the local system bias in the two-channel sliding window W(k), in meters, and median is the median function, are the Ku and Ka channel sea surface height anomalies, respectively, in meters. Correcting any one channel to align to a unified reference to obtain the bias-corrected channel: ， ； wherein, is the bias corrected Ku channel, is the bias corrected Ku channel, , are the initial sea surface height anomalies of the two channels respectively ; S33: uncertainty weighted fusion: calculate the standard deviation within the two-channel sliding window W(k) to obtain the internal random error strength ; and the inverse variance form is used to obtain the weighted coefficient, that is: ， ； wherein, Ku is a weighting factor for the Ku channel, dimensionless; Ka is a weighting factor for the Ka channel, dimensionless; S34: weighting the two channels of data respectively to obtain the fused data: ； wherein, is the highly abnormal result of the two-channel weighted fusion within the sliding window W(k); S34: dual-frequency consistency quality control, that is, quality control according to the consistency of dual-frequency data: If , replace the fusion value in W(k) with the one with smaller random error, or mark it as invalid; where is the max function, and the threshold is adaptively set between 1.5 and 3 according to the local noise level.

4. The submesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry according to claim 1, characterized in that, In step S4, the attitude-coupled mode extraction ACME method specifically comprises: S41: Empirical mode decomposition is performed on the fused sea surface height anomaly sequence, i.e. the fused data , to adaptively decompose it into n intrinsic mode function components and a residual trend item , and is expressed as: ； Wherein, t is the time series along the track, j is the serial number of IMF, and n is the total number of IMFs obtained by decomposition; S42: calculating the correlation between each IMF and the platform trajectory / attitude change signal; Extracting the time series of platform motion parameters provided by the airborne platform inertial measurement unit The platform motion parameters include pitch angle, roll angle, or elevation change; Calculate each intrinsic mode function separately. With each type of platform motion parameter sequence Pearson correlation coefficient between for: ； wherein, is the mean value of the jth IMF component, is the mean value of the platform motion parameter . S43: setting a correlation threshold Tc; traversing all IMF components and all platform motion parameters, if there is then marking this component of the proper mode function as a component to be culled; S44: set all IMF components labeled as platform error mode to zero, use the remaining IMF components and the residual trend term The reconstructed sea level anomaly sequence after removing the platform error The calculation formula is: ； Wherein, E represents the serial number set of all IMF components marked as errors.

5. The submesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry according to claim 1, characterized in that, In step S5, the sub-mesoscale constraint window is: selecting a spatial scale of 5-50 km according to the data resolution requirement; The smoothing method is: any one of sliding average or weighted Gaussian smoothing.

6. The submesoscale signal extraction method based on Ku / Ka dual-frequency SAR altimetry according to claim 1, characterized in that, The step S6 specifically comprises: S61: spatially interpolating the height anomaly sequences corresponding to multiple flight lines processed through steps S2-S5 to a regular two-dimensional grid, and for high-resolution airborne observation data, the grid size is selected to be 2km×2km; S62: secondary small-scale noise removal: performing two-dimensional smoothing processing of the grid data in an m×n window to remove small-scale noise that may be introduced by gridding; S63: mesoscale background removal: interpolate the mesoscale dataset of the same time AVISO / CMEMS satellite altimeter grid observation to the spatiotemporal resolution consistent with the grid as the mesoscale background field, and deduct from the smoothed grid data to obtain the sub-mesoscale dominant sea surface height anomaly, obtain the data system of sub-mesoscale dominant sea surface height anomaly.

7. The system of sub-mesoscale signal extraction method for Ku / Ka dual-frequency SAR altimetry according to claim 1, characterized in that, It comprises: a data acquisition module for acquiring Ku / Ka dual-frequency sea surface height observation raw data synchronously collected by an airborne platform and platform attitude and trajectory data provided by an IMU; a preprocessing module for implementing average sea surface and tidal height registration and deduction on the original height data to generate initial sea surface height anomaly sequences of Ku and Ka dual channels; a dual-frequency collaborative processing module for executing a BUCG-Alt processing method, including sequentially performing adaptive spike suppression, local system bias correction, uncertainty weighted fusion and dual-frequency consistency quality control, and outputting the fused height anomaly sequence; a platform error elimination module for performing empirical mode decomposition on the fused sequence, extracting intrinsic mode functions, and calculating correlations in combination with platform attitude / trajectory data to eliminate error modes with correlations higher than a set threshold; a smoothing processing module for performing smoothing processing on the height anomaly sequence after error elimination within a sub-mesoscale constraint window to suppress high-frequency noise and small-scale disturbances; a gridding and product generation module for interpolating the smoothed data to a two-dimensional regular grid, implementing secondary small-scale denoising and mesoscale background field deduction, and finally generating a sub-mesoscale dominant sea surface height anomaly data system; an output module for storing and outputting the generated data system.

8. The system of claim 7, wherein the Ku / Ka dual-frequency SAR altimetry submesoscale signal extraction method comprises: The generated data system is used to perform a sub-mesoscale vortex identification and tracking method on the two-dimensional sub-mesoscale sea surface height anomaly data, comprising the following steps: M1: read the two-dimensional sub-mesoscale sea surface height anomaly data; M2: estimate the strain characteristics of the anomaly field based on the curvature parameter, lock the sub-mesoscale event high-occurrence area; form vortex candidates along the closed contour in the area and screen them according to the amplitude threshold and spatial threshold; a. Calculate the curvature parameter of the sea surface height anomaly field, including the eigenvalues of the strain rate tensor, identify the area with large strain rate as the sub-mesoscale event high-occurrence area; b. In the high-occurrence area, take the local extreme point as the center, grow along the closed contour outward to form the vortex candidate area; c. Set the amplitude threshold and spatial scale threshold to preliminarily screen the candidate vortexes; d. Record the center position, radius, maximum anomaly amplitude and other attributes of each candidate vortex; M3: between adjacent time instances, perform pairing and time tracking according to the center distance and contour overlap area ratio, and screen according to the duration threshold; output the center position, radius and life cycle attributes of the vortex, and generate visualization and data files; a) spatiotemporal matching of vortex candidates in two consecutive time instances: if the center distance of two vortexes is less than a certain set proportion of the sum of their radii, and the contour overlap area ratio is greater than 50%, it is determined that they are the same vortex at different times; b) establish a vortex trajectory chain to record the time of its appearance, duration and disappearance; c) Set a duration threshold, if at least 3 consecutive time, eliminate false vortex appearing briefly; d) Output each valid vortex life cycle, moving path, intensity change information, and generate visualization results and data files.

9. The system of claim 7, wherein the Ku / Ka dual-frequency SAR altimetry submesoscale signal extraction method further comprises: The generated data system is also used for performing a sub-mesoscale front identification and tracking method on two-dimensional sub-mesoscale sea surface height anomaly data, including the following steps: N1: reading two-dimensional sub-mesoscale sea surface height anomaly data; N2: calculating the spatial gradient amplitude and strain rate tensor eigenvalue of the anomaly field, and extracting linear front candidates satisfying the gradient amplitude threshold and connectivity constraint; N3: screening and tracking the candidates based on direction consistency and time continuity, outputting front position line, length, strength and evolution attributes.

Images