Polar multi-source remote sensing sea wave significant wave height optimized interpolation fusion method and device

Abstract

This invention provides a method and apparatus for optimizing the interpolation and fusion of significant wave heights from multi-source remote sensing data in polar regions. Belonging to the field of remote sensing, the method includes: preprocessing collected data from sea ice edge regions and open sea surfaces; performing Arctic geometric adaptation correction to eliminate systematic biases from multiple satellite sensors, ensuring data consistency and accuracy; using real-time wave height forecasts as the background field, constructing a weighting function based on the satellite observation error covariance matrix, and performing dynamic background field fusion to improve the accuracy and reliability of sparse data areas; setting different weights according to the different characteristics of sea ice edge regions and open sea surfaces, and introducing a transition weighting function for partitioned fusion to avoid boundary discontinuities, thereby obtaining the final fused wave heights; using K-fold cross-validation to verify the robustness of the fusion method, and simultaneously using multi-dimensional data for accuracy verification. This invention improves the problem of insufficient spatial representativeness in accuracy verification caused by the scarcity and concentrated distribution of Arctic in-situ observations.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing, specifically relating to a method and apparatus for optimizing interpolation and fusion of effective wave height of polar multi-source remote sensing ocean waves. Background Technology

[0002] Significant wave height (SWH) is a key indicator for measuring ocean wave energy. It is defined as the average value of the first third of the largest waves in the wave height. This parameter directly reflects the state of the marine dynamic environment and is indispensable for navigation safety in Arctic shipping routes, structural design of offshore platforms, research on the interaction between sea ice and waves, and validation of climate models. Especially against the backdrop of rapid warming in the Arctic region, the melting of sea ice leads to a continuous increase in wave energy. Therefore, accurately obtaining significant wave height data has become the scientific basis for assessing polar marine disaster risks and ecological evolution.

[0003] The advent of satellite altimeters has brought a new breakthrough to the measurement of significant wave height. It measures sea surface height by emitting radar pulses, thereby retrieving the significant wave height. This technology effectively solves the problem of insufficient spatial coverage of traditional buoys in the Arctic ice region, enabling global observation. Currently, multiple satellites, including CryoSat-2 and SARAL / AltiKa, are capable of observing the Arctic region, and data fusion technologies, such as the European CMEMS global wave product, have become important means to improve the spatiotemporal resolution of observations. However, the unique environmental conditions of the Arctic region present many challenges to existing technologies, such as the susceptibility of altimeter signals to reflections from ice sheets and model grid adaptation issues, all of which urgently require targeted optimization.

[0004] Existing effective wave height (SWH) data fusion techniques suffer from several key problems. First, they simply remove data from all sea ice regions without distinguishing between the sea ice margins and thick ice areas. However, field observations show that the sea ice margins still exhibit strong wave activity and significant undulation characteristics. Ignoring this data not only fails to reflect the true sea conditions in the ice margins but also limits the coverage of Arctic SWH products in high-latitude regions, disrupts data continuity, and ultimately affects their spatial integrity and scientific validity. Second, the fusion process often relies on simple weighted fusion, which cannot accurately reflect spatial distribution characteristics. For example, the inverse distance weighting method only considers distance weighting and does not account for spatial correlation. In the face of the spatial discontinuity of the wave field caused by sea ice in the Arctic, it cannot capture complex spatial variability and fails to incorporate background field data, resulting in limited interpolation accuracy in the sparsely observed Arctic. The Kriging method assumes stable spatial correlation, but Arctic sea ice makes the wave field non-stationary, causing the variogram model to fail. Furthermore, the lack of integration with external model data leads to insufficient reliability of interpolation results in sparsely dataned Arctic regions. The successive correction method does not explicitly consider the characteristics of background field errors and differences in the quality of observational data, and lacks the ability to assimilate multiple data sources. Finally, the verification reliability is insufficient; due to the scarcity and concentrated distribution of Arctic buoys, the spatial representativeness of accuracy verification is severely inadequate. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method and apparatus for optimizing interpolation and fusion of significant wave height from multi-source remote sensing of ocean waves in polar regions.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for optimizing the interpolation and fusion of significant wave height from multi-source remote sensing of ocean waves in polar regions includes the following steps:

[0008] Step 1: Preprocess the collected data on the sea ice edge area and open sea surface;

[0009] Step 2: Perform Arctic geometry adaptation correction to eliminate system biases from multiple satellite sensors and ensure data consistency and accuracy;

[0010] Step 3: Using real-time wave height forecast as the background field, construct a weighting function by combining the satellite observation error covariance matrix, and perform dynamic background field fusion to improve the accuracy and reliability of sparse data regions.

[0011] Step 4: Based on the different characteristics of the sea ice edge area and the open sea surface, set different weights and introduce a transition weight function to perform partition fusion, thereby obtaining the final fused wave height;

[0012] Step 5: Use K-fold cross-validation to verify the robustness of the fusion, and at the same time verify the accuracy using multi-dimensional data.

[0013] This invention also provides a polar multi-source remote sensing ocean wave significant wave height optimization interpolation and fusion device, comprising the following modules:

[0014] The preprocessing module preprocesses the collected data from the sea ice edge area and open sea surface.

[0015] The calibration module performs Arctic geometry adaptation calibration to eliminate system biases from multiple satellite sensors and ensure data consistency and accuracy.

[0016] The fusion module uses real-time wave height forecast as the background field and combines it with the satellite observation error covariance matrix to construct a weighting function to perform dynamic background field fusion, thereby improving the accuracy and reliability of data in sparse areas.

[0017] The weight setting module sets different weights based on the different characteristics of the sea ice edge area and the open sea surface, and introduces a transition weight function to perform partition fusion, thereby obtaining the final fused wave height;

[0018] The verification module uses the K-fold cross-validation method to verify the robustness of the fusion, and also uses multi-dimensional data to verify accuracy.

[0019] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described polar multi-source remote sensing ocean wave effective wave height optimization interpolation fusion method.

[0020] The present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described polar multi-source remote sensing ocean wave effective wave height optimization interpolation fusion method.

[0021] Beneficial effects:

[0022] 1. Data Preprocessing and Partitioning: A rigorous data preprocessing workflow was established to partition the data from the sea ice edge region and the open sea surface, improving data quality and thus enhancing the overall accuracy of the fusion results. Simultaneously, partitioned fusion was performed for the sea ice edge region and the open sea surface, with different weights assigned and a cosine-smooth transition weight function introduced to avoid boundary discontinuities, conform to physical laws, reduce errors, and fill the gaps in effective wave height in the sea ice edge region.

[0023] 2. Improve the verification system: The robustness of the fusion method is verified through K-fold cross-validation (successively removing single-satellite data to generate the fused field), quantitatively revealing the differences in the contribution of each satellite and providing a basis for multi-source data selection. At the same time, multi-dimensional accuracy verification is carried out by combining Arctic in-situ observation data, fusion results, WaveWatch III background field data, and similar international data (such as CMEMS), which improves the problem of insufficient spatial representativeness of accuracy verification caused by the scarcity and concentrated distribution of Arctic in-situ observations.

[0024] This invention is applicable not only to areas where sea ice exists, such as the Arctic region, but also to open sea areas, and is suitable for precision verification systems in polar regions. Attached Figure Description

[0025] Figure 1 This is a flowchart of a polar multi-source remote sensing ocean wave effective wave height optimization interpolation fusion method according to the present invention;

[0026] Figure 2 This is a schematic diagram of a polar multi-source remote sensing ocean wave effective wave height optimization interpolation and fusion device according to the present invention;

[0027] Figure 3a , Figure 3b This is a verification image for the accuracy of the fusion results; among them, Figure 3a This is the optimal interpolation fusion result proposed in this invention; Figure 3b This is a near-real-time product of Global Ocean L4 significant wave height (multi-source satellite fusion grid data provided by Copernicus Ocean Service, belonging to Level 4). Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0029] like Figure 1 As shown, this invention proposes a method for optimizing the interpolation and fusion of significant wave heights of ocean waves from multiple sources in polar regions, comprising the following steps:

[0030] Step 1: Data Preprocessing: Develop a data preprocessing workflow with clearly defined partitions, reasonable thresholds, and noise removal capabilities. Process the observation data from the sea ice edge area and the open sea area separately to improve the quality of the observation data. Develop a data preprocessing workflow to process the data from the sea ice edge area and the open sea area separately to improve data quality and thus improve the overall accuracy of the fusion results.

[0031] Step 2, Arctic geometric adaptation correction: Under the orthogonal projection coordinate system, the spatiotemporal matching algorithm is used to obtain the regression relationship of different sensor data, which is used to correct the system bias of multi-sensor data; Arctic geometric adaptation correction is performed, under the orthogonal projection coordinate system, the spatiotemporal matching algorithm is used to obtain the regression relationship of different sensor data, which is used to eliminate the system bias of multi-satellite sensors;

[0032] Step 3: Dynamic Background Field Fusion to Obtain Dynamic Background Fusion Results. The WaveWatch III (WW3, public website: https: / / polar.ncep.noaa.gov / waves / ), a third-generation global wave model developed and publicly released by the National Oceanic and Atmospheric Administration (NOAA), is used as the background field. Its real-time wave height forecast data is used as the background field, and a fusion weight function is constructed by combining the satellite observation error covariance matrix to achieve high-precision fusion in sparse data regions. Dynamic background field fusion, using the real-time wave height forecast from the WaveWatch III model as the background field and combining it with the satellite observation error covariance matrix to construct a weight function, demonstrates higher accuracy and reliability compared to other methods in sparse data regions.

[0033] Step 4: Perform partitioned fusion based on the dynamic background fusion results. Set different weights for observations in the sea ice edge area and the open sea surface, and introduce a transition weight function based on cosine smooth transition to avoid boundary discontinuities.

[0034] Step 5: Improve the verification system. Verify the robustness of the fusion method through K-fold cross-validation (successively removing single-star data to generate the fused field). At the same time, conduct multi-dimensional accuracy verification by combining Arctic in-situ observation data, fusion results, background field data and similar international data.

[0035] Specifically, step 1 includes:

[0036] (1) Data Collection: Collect satellite data with Arctic observation capabilities, including CFOSAT, CryoSat-2, HY-2B, Jason-3, Sentinel-3A, Sentinel-3B, Sentinel-6MF, ALTIKA / SARAL, and SWOT nadir radar altimeter observations. Each observation data includes longitude, latitude, time, significant wave height, and root mean square error of significant wave height (RMSE). The study also acquires real-time wave height forecast data from the WaveWatch III model, near-real-time Copernicus Global Ocean L4 significant wave height products, and CMEMS wave height analysis data. This data includes longitude, latitude, time, and significant wave height information.

[0037] (2) Data preprocessing: Outliers exist in the collected nadir radar altimeter data. First, thresholding and labeling methods are used to filter out the outliers. Among them, the root mean square error of the significant wave height is used. The threshold, The threshold is calculated as follows:

[0038] The effective wave height (SWH) values ​​(0m to 12m) measured by the nadir radar altimeter were divided into bins with a width of 0.5m and a step size of 0.05m. The values ​​within each bin were calculated. The mean and standard deviation are used to retain only bins with more than 100 data points. An upper threshold is set for each SWH bin. The calculation formula is:

[0039] ;

[0040] in, for mean Standard deviation It is an exponential function.

[0041] right of The threshold is fitted to a second-order polynomial and extrapolated to 12m. Construct... Threshold lookup table :

[0042] ;

[0043] in, The fitted second-order polynomial function, This is the value of the polynomial at SWH=12m. Use this lookup table for different SWH values. Threshold control is applied to remove data exceeding a certain threshold. This method can be used to eliminate erroneous valid wave height measurements caused by land interference, a sudden increase in the radar backscattering coefficient sigma0, and attenuation due to heavy rainfall. To protect observational data in the sea ice marginal zone, this processing is only applied to areas with sea ice concentration equal to 0; it is not used for areas with sea ice concentration greater than 0.

[0044] Then, Empirical Mode Decomposition (EMD) filtering is applied to the data. EMD filtering is an adaptive time-frequency analysis method mainly used to process nonlinear and non-stationary signals. EMD decomposes complex signals into several IMF components and a residual component by progressively extracting the signal's Intrinsic Mode Function (IMF), thereby achieving signal denoising, filtering, and trend separation. Unlike traditional Fourier transform or wavelet transform, EMD does not require preset basis functions and decomposes entirely based on the signal's own characteristics. The specific steps are as follows:

[0045] Step (1) Obtain the original effective wave height sequence obtained by satellite radar altimeter inversion ,in The time variable corresponding to the significant wave height data can be obtained simultaneously when acquiring the significant wave height data;

[0046] Step (2) in the original effective wave height sequence Search for all local maxima and local minima in the time interval and record their corresponding time positions. and and the corresponding significant wave height value and ;

[0047] Step (3) uses cubic spline interpolation to connect all local maxima and local minima to construct the upper envelope of the sequence. and lower envelope :

[0048] ;

[0049] Step (4) Calculate the mean of the upper and lower envelopes. : ;

[0050] Step (5) Calculate the difference sequence: ;

[0051] judge Does it satisfy the condition of the intrinsic mode function, i.e. The absolute value of the difference between the number of zero intersections and the number of extreme points does not exceed 1, and the integral value over the entire sequence interval is close to zero, that is: ;

[0052] If the conditions are met, then It was identified as the first IMF, denoted as If not satisfied, then... Repeat steps (2) to (5) above as a new sequence until the IMF condition is met;

[0053] Step (6) Extract the first IMF ( After that, update the remaining sequence. : .when hour, ;

[0054] Repeat steps (2) to (6) above until the remaining sequence is obtained. If it cannot be further decomposed or the energy is below a predetermined threshold, it will eventually become... IMF components , … and residual trend sequence .

[0055] Subsequently, based on the frequency characteristics of each IMF component, the preceding components were removed. High-frequency IMFs (recommended based on experience) (Taking the value to 2), low-frequency information and trend terms are retained, resulting in the filtered effective wave height sequence. :

[0056] ;

[0057] in, For the total number of IMF funds, The initial IMF order for filtering. For the first One IMF component, This represents the residual trend sequence.

[0058] Specifically, step 2 includes:

[0059] Due to systematic errors inherent in multiple satellite sensors, improving data consistency is crucial. Typically, cross-calibration involves comparing buoy data with altimeter data from various satellites. However, in the Arctic region, in-situ observation data is scarce, concentrated, and lacks spatial representativeness. Therefore, it is more suitable to use data from a rigorously calibrated satellite as a benchmark to calibrate data from other satellites. Currently, the latest benchmark satellite used by CMEMS is Sentinel-6MF.

[0060] However, the effective wave height measured by the nadir point of a satellite radar altimeter is discrete data along the orbit, and the grid density along the longitude direction increases with latitude. To facilitate cross-orbit matching and correction, an orthogonal projection method is used to transform the data to a plane coordinate system, thereby improving matching accuracy.

[0061] During the matching process, a one-hour time window and a 50 km radius area were defined. Data within this window was considered matched data, but the time window typically contained multiple measurements. To ensure data quality, the standard deviation of each satellite's data within the window was first calculated, and outliers exceeding three times the standard deviation were removed. Then, the average of the remaining data was calculated as the effective wave height of that satellite at the matching point. By acquiring multiple matching points, the least squares method was used to fit the data, establishing a correction relationship. Based on this, a systematic correction was performed on the satellite data to be corrected, improving the consistency between different satellite data.

[0062] Specifically, step 3 includes:

[0063] Optimal interpolation optimizes the estimation results by combining the background field (model computation) and observational data. Its core idea is to use observational data to correct the model's background field, thereby obtaining the analysis field.

[0064] (1)

[0065] in, Points to be interpolated Effective wave height at the location (analysis field); The background field value corresponding to the point to be interpolated; The first point around the interpolation point Values ​​for each observation point; The first point around the interpolation point Background field values ​​corresponding to each observation point; Points to be interpolated The surrounding The weight of each observation point Points to be interpolated The total number of observation points selected in the surrounding area. The weights are calculated by minimizing the variance of the analysis error:

[0066] (2)

[0067] in, It is the covariance matrix of the background error; It is the covariance matrix of the observation error; , and These are the variances of the observation error and the background error, respectively. The point to be interpolated The weights of the surrounding observation points, The point to be interpolated The background error covariance between each observation point.

[0068] If the observation errors are uncorrelated unit array Equation (2) can be written as:

[0069] (3)

[0070] in, It is a distance-related function, as shown in equation (4):

[0071] (4)

[0072] in, To analyze the distance between the point of analysis and the observation point, The scale is related to the background field error.

[0073] Solving equation (3) yields the interpolation points. The weights of the surrounding observation points can be substituted into equation (1) to obtain the effective wave height of the interpolated point.

[0074] Specifically, step 4 includes:

[0075] For optimizing the effective wave height in the sea ice marginal region, a partitioned optimization approach was used, and the following strategy was constructed:

[0076] (1) In open sea areas, based on experience, it is believed that the observation error is relatively small, so a smaller observation error variance is set to make the weight of the observation data relatively larger; considering that the waves travel a long distance on the open sea surface, a larger correlation scale is set based on experience. ;

[0077] (2) In areas with low sea ice concentration, based on experience, it is believed that the observation error is relatively large, so a larger observation error variance is set to make the background field data have a greater weight; considering that the propagation of sea waves in the sea ice edge area is affected by sea ice and the propagation distance is short, a smaller correlation scale is set based on experience.

[0078] (3) When sea ice concentration exceeds a certain critical value (empirical threshold), the observation data is considered unreliable and only model data is used;

[0079] (4) When the sea ice concentration exceeds another empirical threshold, the model data is considered unreliable and the data is set to null.

[0080] This partitioning weighting achieves fusion partitioning optimization. To ensure the continuity and rationality of the wave height field in the sea ice marginal region across different physical regions, it is based on sea ice concentration. (Unit: %) The fusion region is divided into four segments, and the final fusion wave height is defined using a piecewise function. :

[0081] ;

[0082] in, , , These represent the effective wave height estimation results in open water, periglacial region, and background model field, respectively. The results of the first-stage fusion are defined as follows:

[0083] ;

[0084] Cosine weighting function and The definition is as follows:

[0085] ;

[0086] ;

[0087] When sea ice concentration is 0%, the interpolation results from open water are used directly; when sea ice concentration is between 0% and 15%, the interpolation results from open water are used. and Cosine smoothing weighting between the values; when sea ice concentration is between 15–30%, the results of the previous stage fusion are... Background Field of the Pattern A rapid transition fusion is performed; when the sea ice concentration is between 30% and 50%, the background model field results are used directly; when the sea ice concentration exceeds 50%, it is set to empty. This method balances physical plausibility and spatial continuity, and can effectively improve the wave height estimation quality in ice-covered areas.

[0088] Specifically, step 5 includes:

[0089] In the Arctic region, although in-situ observation data is relatively scarce and spatially concentrated, it can still be used for accuracy assessment. To comprehensively verify the accuracy of the fused data, a multi-dimensional verification method was adopted, combining Arctic in-situ observation data, fused results, background field data, and similar international data for comprehensive evaluation. Specifically, the fused results, background field data, and similar international data were compared and analyzed with the in-situ observation data to quantify the error characteristics of different datasets.

[0090] The following statistical indicators were used to assess accuracy:

[0091] Bias: Measures the systematic error between the fused result and the reference data (in-situ observation data or satellite data not involved in the fusion). The calculation formula is:

[0092] ;

[0093] in, To integrate data or background field data, This includes in-situ observation data or satellite data that was not fused. To match the number of data points, j represents the index value.

[0094] Root Mean Square Error (RMSE): Measures the overall error level between the fused data and the reference data. The formula is as follows:

[0095] ;

[0096] Scatter Index (SI): A measure of the proportion of error to the observed value, defined as:

[0097] ;

[0098] in, The mean of the reference data (in-situ observations or satellite data not involved in fusion).

[0099] The Pearson Correlation Coefficient (R) measures the correlation between the fused data and the reference data. The formula for calculating it is:

[0100] ;

[0101] in, and These are the mean values ​​of the reference data and the fused data, respectively.

[0102] By comparing the accuracy assessment results of the fusion results with those of the background field data, the role of fused satellite observation data in improving the ability to express the effective wave height of the background field can be verified; by comparing the accuracy assessment results of the fusion results with those of similar international data, the overall accuracy level of the fused data can be further evaluated.

[0103] However, due to the scarcity and limited spatial representativeness of in-situ observation data in the Arctic region, a K-fold cross-validation method is employed to evaluate the robustness of the fusion algorithm and analyze the contribution of individual satellite data to the fusion results in order to enhance the reliability of accuracy verification. Specifically, in each round of cross-validation, individual satellite data are successively removed, a new fusion field is generated based on the remaining data, and the removed satellite data is used for independent verification. This method can assess the impact of individual satellite data on the fusion results and ensure the stability of the fusion method.

[0104] In addition, since satellite data has a wide coverage and relatively uniform spatial distribution, drawing on the idea of ​​K-fold cross-validation, data from a satellite that was not involved in the fusion was used to replace the original observation data for accuracy assessment. This was done to compensate for the lack of spatial representativeness caused by the scarcity of original observation data, thereby providing a more comprehensive assessment of the accuracy of the fused data.

[0105] Preferably, it is not necessary to acquire all altimeter data during data collection. If there is little satellite data, the amount of observation data can be increased by reducing the spatiotemporal resolution of the fusion.

[0106] Preferably, in the multi-task cross-correction during data preprocessing, other satellites can be used instead of the Sentinel-6MF satellite as the reference satellite.

[0107] Preferably, other projections suitable for polar regions can be used instead of orthographic projections during data preprocessing.

[0108] Preferably, optimal interpolation does not necessarily have to use WaveWatch III as the background field; it can also use wave numerical model data such as WAM.

[0109] like Figure 2 As shown, the present invention also provides a polar multi-source remote sensing ocean wave significant wave height optimization interpolation and fusion device, comprising the following modules:

[0110] The preprocessing module preprocesses the collected data from the sea ice edge area and open sea surface.

[0111] The calibration module performs Arctic geometry adaptation calibration to eliminate system biases from multiple satellite sensors and ensure data consistency and accuracy.

[0112] The fusion module uses real-time wave height forecast as the background field and combines it with the satellite observation error covariance matrix to construct a weighting function to perform dynamic background field fusion, thereby improving the accuracy and reliability of data in sparse areas.

[0113] The weight setting module sets different weights based on the different characteristics of the sea ice edge area and the open sea surface, and introduces a transition weight function to perform partition fusion, avoid boundary discontinuities, and thus obtain the final fused wave height.

[0114] The validation module uses K-fold cross-validation to verify the robustness of the fusion method, and also uses multi-dimensional data to verify accuracy.

[0115] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described polar multi-source remote sensing ocean wave effective wave height optimization interpolation fusion method.

[0116] The present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described polar multi-source remote sensing ocean wave effective wave height optimization interpolation fusion method.

[0117] like Figure 3a , Figure 3b The image shown is a verification chart of the fusion results. The results are compared and verified with ERA5 (a global atmospheric and oceanic reanalysis dataset provided by the European Centre for Medium-Range Weather Forecasts). Figure 3a This is the optimal interpolation fusion result proposed in this invention. Figure 3b This is a near-real-time product of Global Ocean L4 significant wave height (multi-source satellite fused grid data provided by the Copernicus Ocean Service, belonging to Level 4). The inversion results obtained by the method of this invention show good consistency with the observed data. A total of N=40164 sample points were evaluated, and the root mean square error (RMSE) was 0.336 m, the systematic bias (Bias) was 0.111 m, and the correlation coefficient (R) reached 0.979, indicating that the method has high accuracy and stability in significant wave height fusion.

[0118] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0119] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0120] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0121] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0122] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0123] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for optimizing interpolation and fusion of significant wave height from multi-source remote sensing of ocean waves in polar regions, characterized in that, Includes the following steps: Step 1: Preprocess the collected data on the sea ice edge area and open sea surface; Step 2: Perform Arctic geometry adaptation correction to eliminate system biases from multiple satellite sensors and ensure data consistency and accuracy; Step 3: Using real-time wave height forecast as the background field, construct a weighting function by combining the satellite observation error covariance matrix, and perform dynamic background field fusion to improve the accuracy and reliability of sparse data regions. Step 4: Based on the different characteristics of the sea ice edge area and the open sea surface, set different weights and introduce a transition weight function to perform partition fusion, thereby obtaining the final fused wave height; According to sea ice concentration The fusion region is divided into four segments, and the final fusion wave height is defined using a piecewise function. : ； The unit for sea ice concentration C is %. , , These represent the effective wave height estimation results in open water, periglacial region, and background model field, respectively. The results of the first-stage fusion are defined as follows: ； Cosine weighting function and The definition is as follows: ； ； Step 5: Use K-fold cross-validation to verify the robustness of the fusion, and at the same time verify the accuracy using multi-dimensional data.

2. The polar multi-source remote sensing ocean wave significant height optimization interpolation and fusion method according to claim 1, characterized in that, In step 1, the data preprocessing includes: Outliers are filtered using thresholding and flag methods to construct the root mean square error of effective wave height. A threshold lookup table was used to calculate the root mean square error of the effective wave height at SWH, measured by radar altimeters at different nadir points. Implement threshold control to remove data that exceeds the threshold. Empirical mode decomposition (EMD) filtering is applied to the data. By progressively extracting the intrinsic mode functions (IMFs) of the signal, the complex signal is decomposed into multiple IMFs and a residual component, thereby achieving signal denoising, filtering, and trend separation.

3. The polar multi-source remote sensing ocean wave significant height optimization interpolation and fusion method according to claim 1, characterized in that, In step 2, the Arctic geometry adaptation correction includes: The data is transformed to a plane coordinate system using orthogonal projection. Set the radius of the time window and spatial window, and perform data matching within the range of the time window and spatial window; By fitting the data using the least squares method, a correction relationship is established, and the satellite data to be corrected is systematically corrected.

4. The polar multi-source remote sensing ocean wave significant height optimization interpolation and fusion method according to claim 1, characterized in that, In step 3, dynamic background field fusion includes: The weights are obtained by minimizing the variance of the analysis error, and then substituted into the interpolation formula to obtain the effective wave height value of the point to be interpolated.

5. The polar multi-source remote sensing ocean wave significant height optimization interpolation and fusion method according to claim 1, characterized in that, In step 5, the verification method includes: K-fold cross-validation is used to successively remove single-star data to generate a fused field; Multi-dimensional accuracy verification was conducted by combining Arctic in-situ observation data, fusion results, background field data, and similar international data, and quantitative analysis was performed using statistical indicators.

6. A polar multi-source remote sensing ocean wave significant wave height optimization interpolation and fusion device, characterized in that, Includes the following modules: The preprocessing module preprocesses the collected data from the sea ice edge area and open sea surface. The calibration module performs Arctic geometry adaptation calibration to eliminate system biases from multiple satellite sensors and ensure data consistency and accuracy. The fusion module uses real-time wave height forecast as the background field and combines it with the satellite observation error covariance matrix to construct a weighting function to perform dynamic background field fusion, thereby improving the accuracy and reliability of data in sparse areas. The weight setting module sets different weights based on the different characteristics of the sea ice edge area and the open sea surface, and introduces a transition weight function to perform partition fusion, thereby obtaining the final fused wave height; According to sea ice concentration The fusion region is divided into four segments, and the final fusion wave height is defined using a piecewise function. : ； The unit for sea ice concentration C is %. , , These represent the effective wave height estimation results in open water, periglacial region, and background model field, respectively. The results of the first-stage fusion are defined as follows: ； Cosine weighting function and The definition is as follows: ； ； The verification module uses the K-fold cross-validation method to verify the robustness of the fusion, and also uses multi-dimensional data to verify accuracy.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the polar multi-source remote sensing effective wave height optimization interpolation fusion method as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the polar multi-source remote sensing effective wave height optimization interpolation fusion method as described in any one of claims 1 to 5.

Images