SAR (Synthetic Aperture Radar) sea surface radial flow velocity inversion method based on Bayesian method

Abstract

The invention discloses an SAR (Synthetic Aperture Radar) sea surface radial flow velocity inversion method based on a Bayesian method, and relates to the technical field of marine environment information monitoring. The invention discloses a Bayesian-based SAR sea surface radial flow velocity inversion method, and the method mainly comprises the steps: 1, obtaining a land-containing SAR image in a same-orbit period, extracting land Doppler frequency data, and correcting the range-direction linear inclination caused by an electron pointing error; 2, in combination with a matched wind field, wind wave Doppler shift is removed by means of a geophysical mode function, and preprocessing net frequency shift is obtained; and thirdly, inputting the net frequency shift and the numerical mode ocean flow field into a Bayesian framework, innovatively bringing the electronic pointing error system offset into a state vector, and performing synchronous inversion with the flow velocity to realize electronic pointing error self-adaptive correction. The method improves the inversion precision, stability and reliability, improves the precision of the mode flow field, is high in robustness, does not need land echo assistance, is suitable for wide sea area monitoring, and has engineering values in the fields of marine environment, disaster early warning and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of marine environment information monitoring, and particularly relates to a SAR sea surface radial flow speed inversion method based on a Bayesian method. BACKGROUND

[0002] As a large-scale seawater movement process in the global ocean system, the ocean current is the core driving force of the ocean energy and material circulation. On the one hand, it regulates the global climate pattern by transporting heat and maintains the balance of the marine ecosystem by transporting nutrients. On the other hand, the surface current directly affects the diffusion path of marine pollutants, the emergency response efficiency of marine disasters, and the planning and safety of sea routes, so the high-precision and real-time monitoring of the sea surface flow field is the core demand of marine environment monitoring, ecological protection and sea operation safety management, and has important scientific value and engineering application significance.

[0003] With the iterative development of remote sensing technology, synthetic aperture radar (SAR) has become the core technical means for sea surface radial flow speed inversion and dynamic monitoring due to its all-weather, all-time observation capability, wide coverage and high spatial resolution. Among them, the Doppler centroid anomaly method based on single-beam SAR data gradually becomes the mainstream technical path in this field due to its low dependence on SAR platform attitude information and high inversion accuracy in medium and high resolution scenarios. The core principle is to extract the Doppler centroid frequency shift in the SAR echo signal, that is, the radar wave frequency change caused by the sea surface movement, and inversely deduce the radial flow speed along the radar observation direction combined with the preset physical model.

[0004] However, the Doppler centroid frequency shift obtained by SAR observation is not only contributed by the ocean current, but also the result of superposition of multiple factors: it includes the basic frequency shift caused by the SAR platform motion parameters (such as flight speed, attitude angle), the additional frequency shift caused by sea surface wind wave scattering, and more importantly, the systematic frequency shift caused by electronic pointing error (EPE). Among them, the electronic pointing error is the pointing deviation of the SAR antenna caused by hardware precision or environmental interference, and the Doppler frequency shift caused by the electronic pointing error has the characteristics of "systematic and difficult to quantify". Especially in the open sea (such as the ocean, the far sea) far away from the land, due to the lack of land echoes as a correction reference, this error cannot be effectively eliminated, which becomes the core bottleneck of limiting the inversion accuracy of sea surface radial flow speed.

[0005] Currently, the mainstream correction scheme for EPE in the industry is to use SAR images containing land areas in the same orbit period as the target SAR image. By fitting the Doppler frequency spatial variation law of the land area, an EPE error model is established, and then the model is migrated to the target marine image in the same period to remove the error. However, this scheme has significant limitations: it completely depends on the availability of SAR images of adjacent land areas. Once the target monitoring area is an open sea area without land reference (such as an ocean route or a deep sea fishing ground), it is impossible to obtain effective EPE correction benchmarks, resulting in unstable error removal effects, and even the inversion results deviate from the physical reality, which is difficult to meet the precision requirements of sea surface flow field monitoring in open sea areas.

[0006] In summary, the technical difficulties of the prior art are that the Doppler centroid frequency shift relied on by SAR sea surface radial flow speed inversion is easily disturbed by EPE, and the existing EPE correction method cannot be supported by land area echoes, and is invalid in open sea scenarios.

[0007] Therefore, how to effectively estimate and remove the electronic pointing error in the SAR Doppler frequency without relying on land echoes, and then improve the precision and scene adaptability of sea surface radial flow speed inversion, has become a key technical problem to be solved in the field.

[0008] The information disclosed in this BACKGROUND section is only intended to increase an understanding of the general context in which the present application can be practiced. It is not admitted that any of the information provided in this BACKGROUND section constitutes prior art. SUMMARY

[0009] To solve the above technical problems, the embodiments of the present application provide a SAR sea surface radial flow speed inversion method based on a Bayesian method.

[0010] The present application provides the following technical solutions: a SAR sea surface radial flow speed inversion method based on a Bayesian method, comprising the following steps: Step 1: Obtain a reference SAR image containing land areas in the same orbit period, extract the Doppler frequency data of the land area of the reference SAR image, and extract the systematic variation characteristics of the Doppler frequency along the distance direction caused by the electronic pointing error by a linear fitting method; Step 2: Based on the geophysical model function, combine the target SAR system parameters and the wind field data matched in time and space with the target SAR image to calculate the influence component of the sea surface wind wave on the target SAR observation Doppler shift; Step 3: preprocessing the original observed Doppler frequency of the target SAR, sequentially removing the Doppler frequency tilt effect along the range direction caused by the electronic pointing error and the Doppler frequency shift component caused by the sea wave, to obtain the corrected net Doppler frequency; Step 4: obtaining numerical mode ocean current field data, inputting the numerical mode ocean current field data as a background field and the corrected net Doppler frequency as an observation into a Bayesian inversion framework; Step 5: in the inversion process of the Bayesian inversion framework, the electronic pointing error offset b is included in the state vector and is jointly estimated with the sea surface radial flow velocity parameter, so as to realize the synchronous correction of the electronic pointing error and the accurate inversion of the sea surface radial flow velocity; Step 6: comparing the electronic pointing error offset b obtained by the Bayesian method with the reference value calculated based on the adjacent SAR land area b adj , and the independent reference true value, to evaluate the effectiveness and stability of the method in the estimation of the electronic pointing error; Step 7: introducing high-frequency radar observation data and buoy observation data as independent reference benchmarks to verify the accuracy of the sea surface radial flow velocity results obtained by the Bayesian method, and comparing the sea surface radial flow velocity results with the flow velocity results obtained by the direct inversion method and the numerical mode ocean current field data, to comprehensively evaluate the performance and applicability of the method.

[0011] Preferably, for multiple SAR images under the same orbit period, the systematic variation law of the Doppler frequency along the range direction caused by the electronic pointing error is analyzed in each SAR image respectively, a unified electronic pointing error correction model is established based on the systematic variation law, and the electronic pointing error correction model is used to batch correct the electronic pointing errors of different SAR images under the same orbit period.

[0012] Preferably, the linear fitting step in step 1 is specifically: based on the Doppler frequency data of the land area of the reference SAR image, the tilt change rate of the electronic pointing error along the range direction is calculated by linear fitting; the distance direction tilt error of the Doppler frequency of the target SAR image is linearly de-titled by using the tilt change rate, and the linear trend term of the Doppler frequency along the distance direction caused by the electronic pointing error is eliminated.

[0013] Preferably, the calculation process of the wind wave influence component on the observed Doppler frequency in step 2 is specifically: obtaining the wind field product data at the same time and the same place as the target SAR image, and mapping the spatial resolution of the wind field product data to the resolution of the target SAR Doppler frequency by an interpolation method; combining the incidence angle, polarization mode and band parameters of the target SAR system, the Doppler frequency shift component caused by the sea surface wind wave is calculated based on the geophysical model function.

[0014] Preferably, the Bayesian inversion framework in step 4 or 5 is constructed by a state space model, the state space model takes the sea surface radial velocity and the electronic pointing error offset b as state variables, takes the net Doppler frequency as an observation variable, and takes the numerical mode ocean current field data as a prior constraint of the state variable; the joint solution of the sea surface radial velocity and the electronic pointing error offset b is realized through the state space model.

[0015] Preferably, the Bayesian inversion estimates the state vector by minimizing the cost function, the cost function includes a background constraint term, an observation constraint term and a smoothing constraint term; the balance between the prior knowledge and the observation data is realized through the cost function, and the stability and physical consistency of the inversion result are improved.

[0016] Preferably, different time baselines and space baselines are set, and the electronic pointing error offset b obtained by the Bayesian inversion under different baseline conditions, the reference value obtained by the traditional adjacent scene correction method under different baseline conditions b adj and the deviation from the independent reference true value are calculated; through the deviation analysis, the applicability and stability of the two methods under different space-time baseline conditions are determined, and the robustness of the method in the electronic pointing error estimation is verified.

[0017] Preferably, the correlation coefficient and the standard deviation of the Bayesian inversion velocity result and the HF radar observation data and the buoy observation data are calculated, and the inversion precision is evaluated; the correlation coefficient and the standard deviation of the direct inversion method velocity result and the numerical mode ocean current field data and the HF radar observation data and the buoy observation data are calculated; through multiple comparative analyses, the error characteristics, stability of the Bayesian method under different observation conditions, and the correction ability of the numerical mode ocean current field data are evaluated.

[0018] Preferably, the geophysical model function is a C-band Doppler ocean product physical model, and the Doppler frequency shift component caused by the sea surface wind and wave is calculated by the CDOP model combined with the wind field data and the SAR incidence angle in step 2.

[0019] Preferably, the expression of the Doppler frequency shift component caused by the electronic pointing error obtained by linear fitting in step 1 is: f elec = k x + b ; Wherein, f elec is the Doppler frequency shift component caused by the electronic pointing error, x is the number of pixels in the distance direction, k is the distance direction tilt change coefficient, b is the systematic offset.

[0020] The SAR sea surface radial flow speed inversion method based on the Bayesian method has the following beneficial effects: (1) The electronic pointing error offset and the sea surface radial flow speed are jointly inverted through the Bayesian framework, and the electronic pointing error is no longer dependent on the land echo correction, thereby solving the core problem that the traditional method is invalid in open sea areas and breaking through the traditional technical bottleneck.

[0021] (2) The inversion precision and stability are significantly improved, and specifically, on the one hand, the electronic pointing error estimation precision is high, and the correlation coefficient with the true value reaches 93.92%, which clears the key error interference for flow speed inversion; on the other hand, the flow speed inversion performance is excellent, the standard deviation is reduced by more than 50% compared with the direct method, and is reduced by more than 13.33% compared with the CMEMS background field, the correlation coefficient is significantly improved, and the SAR observation and the ocean model background field are fused, noise is effectively suppressed, the flow field is dynamically corrected, and the physical consistency of the results is ensured.

[0022] (3) Because of the land dependence, it can be stably applied to open sea areas and complex ocean dynamic environments, and finally provides reliable technical support for ocean monitoring and disaster warning, has significant engineering application value and broad popularization prospect. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 It is a flowchart of the SAR sea surface radial flow speed inversion method based on the Bayesian method of the application; Figure 2 It is a land area of the Sentinel-1 SAR image in the same period as the track 91 f elec The variation characteristics along the distance direction; Among them, Figure 2 (a)-(c) in correspond to sub-bands 1-3 of the image on June 21, 2021, respectively; Figure 3 It is an estimated value b The correlation analysis result with the true value; Figure 4 It is two kinds of offsets b The scatter diagram of the result of the estimation method and the true value; Among them, Figure 4 (a)-(e) in are scatter diagrams of the adjacent scene correction method; Figure 4 (f)-(j) in are scatter diagrams of the Bayesian inversion method; the spatial distance range is gradually expanded from 1 scene to 5 scenes; Figure 5 (a) is a scatter density diagram of the Bayesian inversion radial flow speed and the high-frequency radar observation radial flow speed; Figure 5(b) is a scattering density diagram of radial flow velocity observed by CMEMS and high-frequency radar. Figure 6 (a) is a scattering density diagram of Bayesian inversion results and high-frequency radar measurements; Figure 6 (b) is a scattering density map of the direct inversion results and high-frequency radar measurements; Figure 7 (a) shows the comparison between Bayesian inversion radial velocity and buoy radial velocity; Figure 7 (b) shows the comparison results of the radial flow velocity of CMEMS and the radial flow velocity of the buoy. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] To address the problems mentioned in the background section, this invention provides a SAR sea surface radial current velocity inversion method based on Bayesian methods to solve the aforementioned technical problems. The technical solution is as follows: The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0026] like Figure 1 As shown, Figure 1 This paper presents a flowchart illustrating a Bayesian-based method for inverting SAR sea surface radial current velocity. It utilizes RVL (Radial Velocity Observation) Doppler information and OWI (Ocean Wind Field Image) wind field information from the Sentinel-1 (European Space Agency-launched Sentinel-1 SAR satellite) IW model L2-level oceanographic products to correct electronic pointing errors and invert sea surface radial current velocity. The specific steps are as follows: Step 1 uses Sentinel-1 SAR imagery containing land areas to analyze the variation characteristics of Doppler frequency shift with range direction caused by electronic pointing error. For example... Figure 2 As shown, this illustrates the variation of felec (field error) in the range direction for land areas in Sentinel-1 SAR images within the same period of Sentinel-1 SAR orbit 91. (a)–(c) correspond to subbands 1–3 of the image from June 21, 2021, respectively. Analysis results indicate that this error exhibits a significant linear trend in the range direction, which can be expressed as: f elec = k x +b； where, k is the linear slope, b is the systematic offset. Further analysis on different sub-bands shows that the slope of sub-band 1 and sub-band 3 f elec shows a clear linear variation with range, and the fitting slope k is significantly non-zero; the slope of sub-band 2 k is close to zero, showing a weak range dependence. In addition, the intercept of sub-band 2 and sub-band 3 b are close in value.

[0027] Step 2 Based on the analysis results of step 1, the EPE correction is performed on the sub-bands respectively, and the correction process includes slope correction and offset correction. Among them, the slope correction is used to eliminate the linear drift of EPE in the range direction, and the specific method is to use the stable slope parameter k obtained by fitting the SAR image containing land area within the same orbit period to perform deslope processing on the target image.

[0028] Step 3 The sea surface wind wave motion will introduce an additional Doppler shift component, which will affect the extraction accuracy of the radial flow velocity. Therefore, the Doppler component caused by the wind wave is modeled based on the CDOP model in this embodiment, and the OWI wind field data in the OCN product is used as the input parameter, so as to effectively eliminate the interference of wind wave on the observation frequency shift. f ww

[0029] Step 4 In the Bayesian inversion framework, the surface model ocean flow field data in the global ocean analysis and prediction product provided by the Copernicus Marine Environment Monitoring Service is taken as the background flow field, and the Doppler frequency shift after deslope and wind wave effect processing f processed is taken as the observation quantity input. At the same time, the systematic offset caused by EPE b is included in the state vector, and the sea surface radial flow velocity parameter is estimated jointly, so as to realize the synchronous solution of EPE correction and radial flow velocity inversion.

[0030] Step 5 In the Bayesian framework, the state vector is solved by minimizing the cost function: ; In the above formula, J 0 represents the observation term, and J b is the background field term, and its specific expression is as follows: ; ; where, f meas ​This represents the residual Doppler frequency signal from the SAR measurements after preprocessing. During the preprocessing stage, the linear range trend has been removed and the Doppler frequency shift component caused by wind and waves has been corrected. Therefore, it can be considered as the net frequency shift term mainly containing the effects of sea surface radial current velocity and system offset. Within the Bayesian inversion framework, f meas It is used as the input of the observed parameters to retrieve the estimated parameters. Forward model This model reveals the quantitative relationship between SAR Doppler frequency shift, radial flow velocity, and systematic errors. (Background constraints...) u b For the background flow field, offset by the background value b b It originates from SAR images containing land areas within the same orbital period.

[0031] SAR imaging is susceptible to interference from speckle noise, which affects the calculation of Doppler frequencies based on noisy SLC data. f obs It exhibits non-physical abrupt changes in spatial distribution. If directly based on unsmoothed data... f obs Inverting the radial current velocity at the sea surface inevitably inherits these abrupt changes, introducing spurious velocity gradients and disrupting the spatial continuity and physical consistency of the flow field. To alleviate these problems and improve the physical reliability of the inversion results, this invention introduces a gradient constraint smoothing term into the cost function of the Bayesian inversion framework. J s This is to suppress non-stationary disturbances caused by speckle noise during the optimization process.

[0032] ; Step 6: Select SAR images that simultaneously contain both land and ocean areas, and apply the proposed method to the overall EPE offset. b The estimation performance was verified. In this example, Sentinel-1 SAR image data with orbital numbers 33, 34, 70, 76, and 91 were selected, and the estimated values ​​were... b and the true value calculated using land areas b The comparison is as follows: Figure 3 As shown in the figure. Experimental results show that the correlation coefficient between the algorithm's estimated value and the reference value is as high as 93.92%, and the standard deviation between the two is 6.45 Hz, indicating that the proposed method has high accuracy and reliability in electronic pointing error correction.

[0033] Step 7 further validates the advantages of this method in EPE offset estimation by comparing its results with those of traditional neighboring scene correction methods (which utilize reference values ​​obtained by fitting land regions). badj Systematic comparison is performed. The error characteristics of the two methods under different temporal and spatial baseline conditions are analyzed, i.e., the variation of the estimation accuracy with the baseline expansion when the interval between the target SAR image and the reference image gradually increases from 1 adjacent scene to 5 scenes, and the results are shown in Figure 4 The experimental results show that the traditional adjacent scene correction method has higher estimation accuracy when the baseline is short (only 1 scene difference), and its estimation error is lower than that of the Bayesian methods 4(a), (f); when the baseline is expanded to 2 scenes (4(b), 4(g)), the error levels of the two methods tend to be consistent. However, when the scene interval is further expanded to 3 to 5 scenes (3 scenes corresponding to 4(c), 4(h), 4 scenes corresponding to 4(d), 4(i), and 5 scenes corresponding to 4(e), 4(j)), the estimation error of the traditional method increases significantly, showing a clear precision degradation trend.

[0034] Step 8 introduces the HF radar observation data of the US Integrated Ocean Observing System construction, and compares the radial flow velocity results obtained by the present invention with the traditional direct inversion method and the CMEMS background flow field. As shown in Figure 5 , compared with the CMEMS background model, the Bayesian inversion result reduces the error standard deviation by 13.33%, and the correlation coefficient with the HF radar observation value is improved from 64.40% to 72.41%. Compared with the direct method, as shown in Figure 6 , the average standard deviation of the direct method is 0.34 m / s, and the correlation coefficient with the high-frequency radar observation result is only 48.15%. Compared with this, when the Bayesian method is applied to the same data set, the standard deviation can be reduced to 0.11 m / s (more than 50% reduction), and the correlation coefficient can be improved to 83.75%. This result clearly shows that the proposed method has significant advantages in both inversion accuracy and result consistency.

[0035] Step 9 further uses the drift buoy observation data provided by the French Coriolis project to independently verify the method. The comparison results with the buoy observation data are shown in Figure 7 , as shown in Figure 7 , the performance of the Bayesian method is significantly better than that of the CMEMS background field. Compared with CMEMS, the Bayesian method achieves a lower standard deviation (STD=0.24 m / s), while the standard deviation of CMEMS is 0.28 m / s, with a relative improvement rate of 14.29%. In addition, the correlation coefficient (R) is improved from 53.02% to 66.08%, with an absolute increase of 13.06%.

[0036] The above describes the specific embodiments of the present application in combination with the drawings, but is not a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications or variations made by those skilled in the art on the basis of the technical solutions of the present application without creative labor are still within the protection scope of the present application.

Claims

1. A SAR sea surface radial current velocity inversion method based on Bayesian method, characterized in that, Includes the following steps: Step 1: Acquire reference SAR images with the same orbital period and containing land areas, extract Doppler frequency data of land areas from the reference SAR images, and extract the systematic variation characteristics of Doppler frequency along the range direction caused by electronic pointing error through linear fitting method; Step 2: Based on the geophysical model function, combined with the target SAR system parameters and the wind field data that is spatiotemporally matched with the target SAR image, calculate the influence component of sea surface wind and waves on the Doppler frequency shift of the target SAR observation; Step 3: Preprocess the raw Doppler frequencies of the target SAR observations, successively removing the Doppler frequency tilt effect along the range direction caused by electronic pointing error and the Doppler frequency shift component caused by sea waves, to obtain the corrected net Doppler frequency. Step 4: Acquire ocean current field data from numerical models, use the numerical model ocean current field data as the background field, and use the corrected net Doppler frequency as the observation, and input them together into the Bayesian inversion framework. Step 5: In the inversion process of the Bayesian inversion framework, the electron pointing error offset b is incorporated into the state vector and jointly estimated with the sea surface radial current velocity parameter to achieve synchronous correction of the electron pointing error and accurate inversion of the sea surface radial current velocity. Step 6: Compare the electronic pointing error offset b obtained by the Bayesian method with the reference value calculated based on the neighboring SAR land area. b adj The effectiveness and stability of the method in electronic pointing error estimation are evaluated by comparing it with independent reference true values. Step 7: Introduce high-frequency radar observation data and buoy observation data as independent reference benchmarks to verify the accuracy of the sea surface radial velocity results obtained by the Bayesian method inversion. Compare the sea surface radial velocity results with the velocity results obtained by the direct inversion method and the ocean current field data of the numerical model to comprehensively evaluate the performance and applicability of the method.

2. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, For multiple SAR images under the same orbital period, the systematic variation law of Doppler frequency along the range direction caused by electronic pointing error in each SAR image is analyzed. Based on the systematic variation law, a unified electronic pointing error correction model is established, and the electronic pointing error correction model is used to correct the electronic pointing error of different SAR images under the same orbital period in batches.

3. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, The linear fitting step in step 1 is as follows: Based on the Doppler frequency data of the land area of ​​the reference SAR image, the tilt rate of the electronic pointing error with the range direction is calculated by linear fitting; the tilt rate is used to perform linear de-tilting processing on the range tilt error of the Doppler frequency of the target SAR image to eliminate the linear trend term of the Doppler frequency along the range direction caused by the electronic pointing error.

4. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, The calculation process for the influence of wind and waves on the observed Doppler frequency shift in step 2 is as follows: acquire wind field product data at the same time and location as the target SAR image, and map the spatial resolution of the wind field product data to the resolution of the target SAR Doppler frequency through interpolation; combine the incident angle, polarization mode and band parameters of the target SAR system, and calculate the Doppler frequency shift component caused by sea surface wind and waves based on the geophysical model function.

5. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, In step 4 or 5, the Bayesian inversion framework is constructed using a state-space model. The state-space model uses the sea surface radial velocity and the electron pointing error offset b as state variables, the net Doppler frequency as the observation variable, and the ocean current field data from the numerical model as the prior constraints for the state variables. The joint solution of the sea surface radial velocity and the electron pointing error offset b is achieved through the state-space model.

6. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, Bayesian inversion estimates the state vector by minimizing a cost function, which includes background constraints, observation constraints, and smoothing constraints. The cost function achieves a balance between prior knowledge and observation data, thereby improving the stability and physical consistency of the inversion results.

7. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, Different temporal and spatial baselines were set up, and the electronic pointing error offset b obtained by Bayesian inversion and the reference value obtained by the traditional nearby scene correction method were calculated under different baseline conditions. b adj The deviation from the independent reference truth is analyzed; the applicability and stability of the two methods under different spatiotemporal baseline conditions are determined by deviation analysis, and the robustness of the methods in electronic pointing error estimation is verified.

8. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, The correlation coefficients and standard deviations of the Bayesian inversion velocity results with HF radar observation data and buoy observation data were calculated to evaluate the inversion accuracy. At the same time, the correlation coefficients and standard deviations of the velocity results from the direct inversion method and the ocean current field data from the numerical model were calculated with HF radar observation data and buoy observation data, respectively. Through multiple comparative analyses, the error characteristics, stability, and correction capabilities of the Bayesian method under different observation conditions were evaluated.

9. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, The geophysical model function is the C-band Doppler marine product physical model. In step 2, the Doppler frequency shift component caused by sea surface wind and waves is calculated by combining the CDOP model with wind field data and SAR incident angle.

10. The SAR sea surface radial current velocity inversion method based on Bayesian method according to claim 1, characterized in that, In step 1, the expression for the Doppler frequency shift component caused by the electron pointing error, obtained through linear fitting, is as follows: f elec = k⋅x + b ; in, f elec This is the Doppler frequency shift component caused by the electron pointing error. x For distance in pixels, k The distance is the coefficient of inclination. b This is a systematic offset.

Images