Data fusion method and system based on multi-source satellite altimeter data

Abstract

The invention relates to the technical field of data processing, and discloses a data fusion method and system based on multi-source satellite altimeter data. The method comprises the following steps: performing multi-satellite intersection analysis and orbit error data correction on preprocessed multi-source satellite altimeter data to obtain satellite orbit error correction data; carrying out height subtraction processing on the earth water surface height and the average sea surface height on the sea surface height measured by the correction satellite, obtaining along-orbit sea surface height data, carrying out abnormal height data screening, obtaining coastal height abnormal data, carrying out space-time interpolation processing on the coastal height abnormal data, and obtaining a coastal height data set; the method comprises the steps of obtaining interpolation coastal height abnormal data, performing multi-source fusion to obtain gridded sea surface height abnormal fusion data, and performing abnormal reason analysis on the gridded sea surface height abnormal fusion data to obtain a predicted abnormal reason. According to the invention, the accuracy of data fusion based on the multi-source satellite altimeter data is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a data fusion method and system based on multi-source satellite altimeter data. Background Art

[0002] With the continuous development of satellite remote sensing technology, satellite altimeters have become an important tool for monitoring dynamic changes in the ocean. The traditional single satellite altimeter data acquisition method relies on the data of a single satellite altimeter, while the multi-source satellite altimeter data fusion method can provide sea surface height measurement data with higher accuracy and temporal and spatial resolution by integrating data from multiple satellites. These data are widely used in fields such as oceanographic research, ocean dynamics analysis, and climate change monitoring.

[0003] However, the existing multi-source satellite altimeter data fusion methods have some shortcomings. First, the heterogeneity of data sources makes the data analysis, quality control and processing process complicated and time-consuming, which is prone to data inconsistency and quality problems. Second, the traditional methods are not accurate enough in correcting satellite orbit errors, making it difficult to fully utilize the advantages of multi-source data. In addition, in the process of screening and processing abnormal data, the existing methods lack in-depth analysis of the causes of the anomalies, resulting in inaccurate and unreliable prediction results. Summary of the invention

[0004] The present application provides a data fusion method and system based on multi-source satellite altimeter data, which are used to improve the accuracy of data fusion based on multi-source satellite altimeter data.

[0005] In a first aspect, the present application provides a data fusion method based on multi-source satellite altimeter data, the data fusion method based on multi-source satellite altimeter data comprising: performing multi-source satellite altimeter data collection to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set comprises sea surface height measurement values, timestamps and orbit information data; performing parsing and quality control processing on the multi-source satellite altimeter data set to obtain pre-processed multi-source satellite altimeter data; performing multi-satellite intersection analysis and orbit error data correction on the pre-processed multi-source satellite altimeter data to obtain satellite orbit error data. The invention relates to a method for preparing the satellite orbit error correction data; performing a height subtraction process on the sea surface height measured by the correction satellite in the satellite orbit error correction data by the geoid water surface height and the average sea surface height to obtain the along-track sea surface height data; performing abnormal height data screening on the along-track sea surface height data to obtain the coastal height abnormality data; performing spatiotemporal interpolation processing on the coastal height abnormality data to obtain the interpolated coastal height abnormality data; performing multi-source fusion on the interpolated coastal height abnormality data to obtain the gridded sea surface height abnormality fusion data; performing abnormal cause analysis on the gridded sea surface height abnormality fusion data to obtain the predicted abnormal cause.

[0006] In combination with the first aspect, in a first implementation method of the first aspect of the present application, the multi-source satellite altimeter data acquisition is performed to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps and orbital information data, including: obtaining data source identification information, and prioritizing the data source identification information to obtain a data source priority list; according to the data source priority list, the multi-source satellite altimeter data is acquired in batches to obtain batch raw data; a data integrity check is performed on the batch raw data to obtain an integrity check result; according to the integrity check result, the missing data is supplemented for the batch raw data to obtain a completed data set; noise elimination processing is performed on the completed data set to obtain a denoised data set; data format unification processing is performed on the denoised data set to obtain the multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps and orbital information data.

[0007] In combination with the first aspect, in a second implementation method of the first aspect of the present application, the multi-source satellite altimeter data set is parsed and quality controlled to obtain preprocessed multi-source satellite altimeter data, including: performing data format identification on the multi-source satellite altimeter data set to obtain data format type information; decoding the multi-source satellite altimeter data set according to the data format type information to obtain decoded data; performing time synchronization processing on the decoded data to obtain time consistency data; performing spatial coordinate conversion on the time consistency data to obtain unified coordinate system data; performing outlier detection on the unified coordinate system data to obtain outlier marked data; performing data interpolation on the outlier marked data to obtain a completed data set; performing atmospheric delay correction on the completed data set to obtain corrected altitude data; performing data consistency evaluation on the corrected altitude data to obtain the preprocessed multi-source satellite altimeter data.

[0008] In combination with the first aspect, in a third implementation method of the first aspect of the present application, the preprocessed multi-source satellite altimeter data is subjected to multi-satellite intersection analysis and orbit error data correction to obtain satellite orbit error correction data, including: performing orbit parameter extraction on the preprocessed multi-source satellite altimeter data to obtain a satellite orbit parameter set; performing intersection calculation on the satellite orbit parameter set to obtain multi-satellite intersection position data; performing time interpolation on the multi-satellite intersection position data to obtain an intersection time height value; performing difference calculation on the intersection time height value to obtain intersection height difference data; performing orbit error modeling on the intersection height difference data by the least squares method to obtain orbit error model parameters; performing time series analysis on the orbit error model parameters to obtain orbit error time-varying characteristics; performing orbit error correction on the preprocessed multi-source satellite altimeter data according to the orbit error time-varying characteristics to obtain preliminary correction data; performing residual error evaluation on the preliminary correction data to obtain the satellite orbit error correction data.

[0009] In combination with the first aspect, in a fourth implementation method of the first aspect of the present application, the correction satellite measured sea surface height in the satellite orbit error correction data is subjected to height subtraction processing of the geoid water surface height and the average sea surface height to obtain along-track sea surface height data, including: performing coordinate system conversion on the satellite orbit error correction data to obtain the correction satellite measured sea surface height in a unified geocentric coordinate system; performing tidal effect correction on the correction satellite measured sea surface height in the unified geocentric coordinate system to obtain tidal effect-free sea surface height data; performing atmospheric pressure effect correction on the tidal effect-free sea surface height data to obtain sea surface height data under standard atmospheric pressure; performing geoid calculation on the sea surface height data under standard atmospheric pressure to obtain geoid height. The invention relates to a method for obtaining a local optimized geoid model and a dynamic geoid height data; performing a regional harmonic analysis on the geoid height data to obtain a local optimized geoid model; performing a time series analysis on the local optimized geoid model to obtain dynamic geoid height data; performing average sea surface modeling on the acquired historical satellite altimeter data by an empirical orthogonal function decomposition algorithm to obtain an initial average sea surface model; performing an ocean dynamics correction on the initial average sea surface model to obtain optimized average sea surface height data; subtracting the dynamic geoid height data and the optimized average sea surface height data from the sea surface height data under standard atmospheric pressure to obtain preliminary along-track sea surface height anomaly data; performing high-frequency noise filtering on the preliminary along-track sea surface height anomaly data to obtain the along-track sea surface height data.

[0010] In combination with the first aspect, in a fifth implementation of the first aspect of the present application, the along-track sea level height data is screened for abnormal height data to obtain coastal height abnormal data, and the coastal height abnormal data is subjected to spatiotemporal interpolation processing to obtain interpolated coastal height abnormal data, including: performing statistical characteristic analysis on the along-track sea level height data to obtain sea level height distribution characteristic parameters; setting a threshold value for the along-track sea level height data according to the sea level height distribution characteristic parameters to obtain an abnormal height screening threshold value; performing abnormal data screening on the along-track sea level height data based on the abnormal height screening threshold value to obtain a preliminary abnormal height data set; performing spatial clustering analysis on the preliminary abnormal height data set to obtain an abnormal Height spatial distribution characteristics; according to the abnormal height spatial distribution characteristics, the boundary of the preliminary abnormal height data set is extracted to obtain the abnormal height data along the coast; the abnormal height data along the coast is decomposed into time series to obtain time trend items and period items; the time trend items and period items are analyzed in multi-scale by wavelet transform to obtain multi-scale time characteristics; the abnormal height data along the coast is analyzed in space correlation to obtain spatial autocorrelation characteristics; according to the multi-scale time characteristics and the spatial autocorrelation characteristics, a spatiotemporal kriging interpolation model is constructed to obtain optimized interpolation parameters; the abnormal height data along the coast is interpolated in space and time by the optimized interpolation parameters to obtain the interpolated abnormal height data along the coast.

[0011] In combination with the first aspect, in a sixth implementation of the first aspect of the present application, the interpolated coastal height anomaly data is subjected to multi-source fusion to obtain gridded sea surface height anomaly fusion data, and the gridded sea surface height anomaly fusion data is subjected to an anomaly cause analysis to obtain a predicted anomaly cause, including: identifying the data source of the interpolated coastal height anomaly data to obtain a multi-source data identifier; assigning weights to the interpolated coastal height anomaly data according to the multi-source data identifier to obtain an initial fusion weight; performing a spatial consistency check on the interpolated coastal height anomaly data to obtain a spatial consistency score; dynamically adjusting the initial fusion weight according to the spatial consistency score to obtain to optimize the fusion weight; use the optimized fusion weight to perform weighted averaging on the interpolated coastal height anomaly data to obtain preliminary fusion data; perform physical constraint optimization on the preliminary fusion data to obtain gridded sea surface height anomaly fusion data; perform time series decomposition on the gridded sea surface height anomaly fusion data to obtain trend terms, periodic terms and residual terms; perform feature extraction on the trend terms, periodic terms and residual terms through a deep learning algorithm to obtain anomaly feature vectors; perform correlation analysis on the anomaly feature vectors and preset ocean dynamics model parameters to obtain anomaly cause correlation; perform probability sorting on the anomaly causes according to the anomaly cause correlation to obtain the predicted anomaly causes.

[0012] In a second aspect, the present application provides a data fusion system based on multi-source satellite altimeter data, the data fusion system based on multi-source satellite altimeter data comprising:

[0013] An acquisition module is used to acquire multi-source satellite altimeter data to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps and orbit information data;

[0014] A control module, used for parsing and quality controlling the multi-source satellite altimeter data set to obtain pre-processed multi-source satellite altimeter data;

[0015] A correction module, used for performing multi-satellite intersection analysis and orbit error data correction on the preprocessed multi-source satellite altimeter data to obtain satellite orbit error correction data;

[0016] A subtraction module, used for performing height subtraction processing on the sea surface height measured by the correction satellite in the satellite orbit error correction data by subtracting the geoid water surface height and the average sea surface height to obtain along-track sea surface height data;

[0017] A screening module is used to screen the along-track sea level height data for abnormal height data to obtain abnormal coastal height data, and to perform spatiotemporal interpolation processing on the abnormal coastal height data to obtain interpolated abnormal coastal height data;

[0018] The fusion module is used to perform multi-source fusion on the interpolated coastal height anomaly data to obtain gridded sea surface height anomaly fusion data, and perform anomaly cause analysis on the gridded sea surface height anomaly fusion data to obtain predicted anomaly causes.

[0019] In the technical solution provided by the present application, by collecting and analyzing multi-source satellite altimeter data, the priority sorting and batch collection strategy of data source identification information are adopted, which significantly improves the efficiency and integrity of data collection and ensures the diversity and reliability of data sources. The quality and consistency of data are improved through data integrity check, missing data supplementation and noise elimination processing steps. In the process of data analysis and quality control, data format recognition, decoding, time synchronization and spatial coordinate conversion are used to improve the analysis efficiency and time and space consistency of multi-source data. In particular, the accuracy and reliability of data are further improved through measures such as outlier detection and atmospheric delay correction. For the multi-satellite intersection analysis and orbit error data correction steps, the orbit errors between multi-satellite data are effectively eliminated by means of orbit parameter extraction, intersection calculation, time interpolation and difference calculation. Using methods such as least squares and time series analysis, the orbit error model parameters are accurately modeled and corrected, which significantly improves the precision and accuracy of satellite orbit error correction. Furthermore, by subtracting the height of the geoid and the average sea surface height from the sea surface height measured by the correction satellite, combined with the correction of tidal effect, atmospheric pressure effect and geoid calculation, the high accuracy and consistency of the sea surface height data along the track are ensured. In the process of abnormal height data screening and spatiotemporal interpolation processing, the outliers in the sea surface height data are effectively identified and corrected through statistical feature analysis, spatial clustering and spatiotemporal kriging interpolation model. The spatiotemporal interpolation accuracy of abnormal data is further improved by adopting multi-scale time feature and spatial autocorrelation feature analysis. Finally, through the steps of multi-source fusion and abnormal cause analysis, the efficient fusion of multi-source data and the accurate prediction of abnormal data are achieved by using methods such as data source identification, weight allocation, spatial consistency test and physical constraint optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0021] Figure 1 This is a schematic diagram of an embodiment of a data fusion method based on multi-source satellite altimeter data in an embodiment of the present application;

[0022] Figure 2 This is a schematic diagram of an embodiment of a data fusion system based on multi-source satellite altimeter data in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The present application embodiment provides a data fusion method and system based on multi-source satellite altimeter data. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable in appropriate circumstances, so that the embodiments described here can be implemented in an order other than the content illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0024] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the data fusion method based on multi-source satellite altimeter data includes:

[0025] Step S101, collecting multi-source satellite altimeter data to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps and orbit information data;

[0026] Step S102, parsing and quality control processing are performed on the multi-source satellite altimeter data set to obtain pre-processed multi-source satellite altimeter data;

[0027] Step S103, performing multi-satellite intersection analysis and orbit error data correction on the pre-processed multi-source satellite altimeter data to obtain satellite orbit error correction data;

[0028] Step S104, subtracting the sea level measured by the correction satellite in the satellite orbit error correction data from the earth's water level and the average sea level to obtain along-track sea level data;

[0029] Step S105, screening the along-track sea level height data for abnormal height data to obtain abnormal coastal height data, and performing spatiotemporal interpolation processing on the abnormal coastal height data to obtain interpolated abnormal coastal height data;

[0030] Step S106: multi-source fusion is performed on the interpolated coastal height anomaly data to obtain gridded sea surface height anomaly fusion data, and anomaly cause analysis is performed on the gridded sea surface height anomaly fusion data to obtain predicted anomaly causes.

[0031] It is understandable that the execution subject of the present application may be a data fusion system based on multi-source satellite altimeter data, or a terminal or a server, which is not specifically limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0032] Specifically, the data source identification information of different satellites is obtained. The identification information includes satellite name, orbital parameters, sensor type, etc. The priority sorting of data source identification information is to ensure that high-quality and important data sources are processed first. After sorting, satellite altimeter data are collected in batches according to the priority list to form batched original data sets. Each batch of data sets passes the data integrity check to identify the missing data. The missing data is filled by interpolation or historical data to ensure the continuity and integrity of the data. Then, the completed data set is subjected to noise elimination processing. Common methods include median filtering, Kalman filtering, etc. to obtain a denoised data set. Finally, the denoised data set is subjected to data format unification processing to ensure the consistency and compatibility of different satellite data, and a multi-source satellite altimeter data set containing sea surface height measurement values, timestamps and orbital information data is formed. The multi-source satellite altimeter data set is parsed and quality controlled. First, data format identification is performed to determine the format type information of each data file. According to the format type information, the data set is decoded to obtain decoded data. Then, the decoded data is subjected to time synchronization processing to align the data of different satellites on the same time axis to obtain time consistency data. Commonly used techniques for time synchronization processing include timestamp alignment and interpolation. Next, perform spatial coordinate conversion on the time consistency data, unify the satellite data into the same coordinate system, and obtain unified coordinate system data. Unified coordinate system data is used for outlier detection. Common methods include 3σ rule, box plot, etc., to mark outliers. Data interpolation is performed on the marked outliers to ensure data integrity and continuity. Atmospheric delay correction is for the delay effect of the altimeter signal in the transmission of the atmosphere. Common methods include dual-frequency observation, model correction, etc., to obtain more accurate altitude data after correction. Finally, the corrected altitude data is evaluated for data consistency to ensure the quality of the pre-processed data.

[0033] The pre-processed multi-source satellite altimeter data is subjected to multi-satellite intersection analysis and orbit error data correction. First, the satellite orbit parameters, including orbit inclination, ascending node right ascension, perigee parameters, etc., are extracted from the data to form a satellite orbit parameter set. According to the orbit parameter set, the intersection position data of multiple satellites are calculated. These intersection data are obtained through time interpolation methods such as linear interpolation and spline interpolation to obtain the intersection height value. The intersection height values ​​are calculated by difference calculation to obtain the intersection height difference data. The altitude difference data is modeled by the least squares method to obtain the orbit error model parameters. The orbit error model parameters are analyzed in time series to extract the time-varying characteristics of the orbit error. According to the time-varying characteristics, the pre-processed multi-source satellite altimeter data is subjected to orbit error correction to obtain preliminary correction data. The residual error evaluation is the last step. The quality of the corrected data is evaluated by statistical analysis and error propagation methods to ensure the accuracy of the orbit error correction data.

[0034] The height of the sea surface measured by the correction satellite in the satellite orbit error correction data is subtracted from the height of the geoid water surface and the average sea surface height. First, the coordinate system of the orbit error correction data is converted and unified to the geocentric coordinate system. Then, the data is corrected for tidal effects. Commonly used tidal models include global tidal models, local tidal models, etc., to obtain sea surface height data without tidal effects. Atmospheric pressure effect correction is to correct the influence of air pressure on sea surface height through data under standard atmospheric pressure, and obtain sea surface height data under standard atmospheric pressure. Then the geoid height data is calculated, and the methods used include gravity field model calculation, satellite altimetry data analysis, etc. The local optimized geoid model is obtained through regional harmonic analysis, which describes the local sea surface height more finely. The dynamic geoid height data is extracted through time series analysis to extract the time variation characteristics of the geoid. The average sea surface modeling adopts the empirical orthogonal function decomposition algorithm, and the initial average sea surface model is obtained through historical satellite altimeter data. The initial model is optimized by ocean dynamics correction to obtain more accurate average sea surface height data. Finally, the sea surface height data under standard atmospheric pressure is subtracted from the dynamic geoid height data and the optimized average sea surface height data to obtain preliminary along-track sea surface height anomaly data. High-frequency noise filtering further cleans up the high-frequency components in the data and finally obtains the along-track sea surface height data.

[0035] The along-track sea level data is screened for abnormal height data, and abnormal coastal height data is screened out. First, a statistical feature analysis is performed to obtain the characteristic parameters of the sea level distribution. According to these parameters, the abnormal height screening threshold is set to screen out the preliminary abnormal height data set. A spatial clustering analysis is performed on the preliminary abnormal height data set to obtain the spatial distribution characteristics of the abnormal height and extract the boundaries of the abnormal data. The abnormal coastal height data is decomposed into time series to obtain time trend terms and periodic terms. The time trend terms and periodic terms are analyzed at multiple scales by wavelet transform to obtain multi-scale time characteristics. Spatial correlation analysis is used to extract spatial autocorrelation characteristics. Finally, based on the multi-scale time characteristics and spatial autocorrelation characteristics, a spatiotemporal kriging interpolation model is constructed, the interpolation parameters are optimized, and the abnormal coastal height data are interpolated at multiple times to obtain interpolated abnormal coastal height data.

[0036] The interpolated coastal height anomaly data is fused from multiple sources to obtain gridded sea surface height anomaly fusion data. First, the data source of the interpolated coastal height anomaly data is identified to obtain the multi-source data identifier. The initial fusion weight is assigned according to the multi-source data identifier to ensure the contribution ratio of different data sources. The spatial consistency test compares and analyzes the spatial consistency of different data sources to obtain the spatial consistency score. The initial fusion weight is dynamically adjusted according to the score to obtain the optimized fusion weight. The interpolated coastal height anomaly data is weighted averaged using the optimized fusion weight to obtain the preliminary fusion data. The physical constraint optimization modifies the preliminary fusion data through the physical model and actual observation data to obtain the gridded sea surface height anomaly fusion data. The fusion data is decomposed into time series to obtain trend terms, periodic terms and residual terms. The features of these terms are extracted through the deep learning algorithm to form anomaly feature vectors. The anomaly feature vectors are associated with the parameters of the ocean dynamics model to obtain the anomaly cause correlation, and the anomaly causes are ranked by probability to obtain the predicted anomaly causes.

[0037] In the embodiment of the present application, by performing multi-source satellite altimeter data acquisition and analysis processing, the priority sorting and batch acquisition strategy of data source identification information is adopted, which significantly improves the efficiency and integrity of data acquisition and ensures the diversity and reliability of data sources. The quality and consistency of data are improved through data integrity check, missing data supplementation and noise elimination processing steps. In the process of data analysis and quality control, data format recognition, decoding, time synchronization and spatial coordinate conversion are used to improve the analysis efficiency and time-space consistency of multi-source data. In particular, the accuracy and reliability of data are further improved through measures such as outlier detection and atmospheric delay correction. For the multi-satellite intersection analysis and orbit error data correction steps, the orbit errors between multi-satellite data are effectively eliminated by means of orbit parameter extraction, intersection calculation, time interpolation and difference calculation. Using methods such as least squares and time series analysis, the orbit error model parameters are accurately modeled and corrected, which significantly improves the precision and accuracy of satellite orbit error correction. Furthermore, by subtracting the height of the geoid and the average sea surface height from the sea surface height measured by the correction satellite, combined with the correction of tidal effect, atmospheric pressure effect and geoid calculation, the high accuracy and consistency of the sea surface height data along the track are ensured. In the process of abnormal height data screening and spatiotemporal interpolation processing, the outliers in the sea surface height data are effectively identified and corrected through statistical feature analysis, spatial clustering and spatiotemporal kriging interpolation model. The spatiotemporal interpolation accuracy of abnormal data is further improved by adopting multi-scale time feature and spatial autocorrelation feature analysis. Finally, through the steps of multi-source fusion and abnormal cause analysis, the efficient fusion of multi-source data and the accurate prediction of abnormal data are achieved by using methods such as data source identification, weight allocation, spatial consistency test and physical constraint optimization.

[0038] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0039] (1) Obtaining data source identification information, prioritizing the data source identification information, and obtaining a data source priority list;

[0040] (2) According to the data source priority list, multi-source satellite altimeter data are collected in batches to obtain batches of raw data;

[0041] (3) Performing data integrity check on the batches of raw data to obtain integrity check results;

[0042] (4) Based on the integrity check results, missing data is supplemented in batches of original data to obtain a complete data set;

[0043] (5) performing noise elimination processing on the completed data set to obtain a denoised data set;

[0044] (6) The denoised data set is processed into a unified data format to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps, and orbit information data.

[0045] Specifically, relevant information is collected from multiple satellite data sources. This information includes satellite name, sensor type, orbital parameters, data collection frequency, etc. By collecting this identification information, we can have a preliminary understanding of the characteristics and data quality of each data source, providing a basis for subsequent data processing. The purpose of prioritizing the data source identification information is to ensure that data sources with high data quality, wide coverage, and high spatiotemporal resolution are used first in the data processing process. The priority sorting criteria can be determined based on factors such as the stability, historical performance, data update frequency, and accuracy of the data source. Through weight allocation and comprehensive scoring, a data source priority list is obtained. According to the data source priority list, multi-source satellite altimeter data are collected in batches to ensure that high-priority data sources are collected and processed at an early stage. Batch collection can be reasonably arranged according to the satellite's orbital period and data transmission time to avoid data being too concentrated at a certain moment. After the batch raw data is collected, a data integrity check is required to ensure the continuity and integrity of each batch of data in space and time. The data integrity check can be performed by comparing the expected data coverage with the actual collected data range. If data is missing, the missing data is supplemented by interpolation or using historical data to ensure the integrity of the data set. After the missing data is supplemented, a complete data set is formed. The completed data set contains noise data, so noise elimination is required. Common methods for noise elimination include median filtering and Kalman filtering. Median filtering effectively removes mutation noise by taking the middle value in a window in the data set as the representative value of the window. Kalman filtering reduces the impact of random errors and noise by dynamically estimating and correcting the data. After noise elimination, a denoised data set is obtained.

[0046] The data format of the denoised dataset is unified to ensure the consistency of the format and unit of data from different satellites. The format unification process includes converting and aligning the timestamps, coordinate systems, and altitude units of different data sources. The timestamp conversion unifies the data of different time formats into a standard format to ensure temporal comparability. The coordinate system conversion converts the spatial coordinates of each data source into a unified geocentric coordinate system to ensure spatial consistency. The altitude unit conversion unifies the altitude data of different units into a standard unit to ensure the consistency of altitude measurements. After the format unification process, a multi-source satellite altimeter dataset containing sea surface height measurements, timestamps, and orbital information data is formed.

[0047] Through the above processing steps, the high quality and consistency of multi-source satellite altimeter datasets can be guaranteed, providing a reliable data basis for subsequent data analysis, quality control and fusion processing.

[0048] For example, in a multi-source satellite altimeter data fusion project, the data source identification information of multiple satellites was first collected, including the names, orbital parameters, and data collection frequencies of satellites A, B, and C. Through comprehensive evaluation, it was determined that the data source of satellite A had the highest priority, followed by satellite B, and finally satellite C. According to the priority list, the altimeter data of satellite A was collected first, the data of satellite B was collected during the data transmission interval, and the data of satellite C was collected last. Batched raw data sets were formed.

[0049] The integrity check of the batches of raw data revealed that satellite A had missing data at certain times. The missing data was supplemented using the historical data of satellite A and the interpolation method to ensure data continuity. After the supplemented data set was formed, noise elimination was performed. The median filtering method was used to effectively remove the mutation noise in the data to form a denoised data set.

[0050] Subsequently, the denoised dataset was processed to a unified data format. The data timestamps of satellites A, B, and C were unified into UTC format, the coordinate system was converted into the geocentric coordinate system, and the height unit was unified into meters. After the format was unified, a multi-source satellite altimeter dataset containing sea surface height measurements, timestamps, and orbital information data was formed.

[0051] Specifically, assume that the data of satellite A at a certain point in time is 10.5 meters above sea level, the timestamp is "2023-05-01T12:00:00Z", and the orbital information is "orbital parameter A". The data of satellite B at the same point in time is 10.7 meters above sea level, the timestamp is "2023-05-01T12:00:10Z", and the orbital information is "orbital parameter B". Through timestamp conversion, the data timestamp of satellite B is aligned to "2023-05-01T12:00:00Z". After the coordinate system is converted, it is unified into the geocentric coordinate system. After the height unit is converted, ensure that the height data is all meters. Finally, the multi-source satellite altimeter data set is formed as follows:

[0052] Data point 1: sea surface height 10.5 meters, timestamp "2023-05-01T12:00:00Z", orbit information "Orbital parameters A".

[0053] Data point 2: sea surface height 10.7 meters, timestamp "2023-05-01T12:00:00Z", orbit information "orbital parameter B".

[0054] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0055] (1) Identify the data format of multi-source satellite altimeter data sets and obtain data format type information;

[0056] (2) decoding the multi-source satellite altimeter data set according to the data format type information to obtain decoded data;

[0057] (3) Performing time synchronization processing on the decoded data to obtain time-consistent data;

[0058] (4) Performing spatial coordinate transformation on the time-consistent data to obtain unified coordinate system data;

[0059] (5) performing outlier detection on the unified coordinate system data to obtain outlier marked data;

[0060] (6) interpolating the outlier labeled data to obtain a complete data set;

[0061] (7) Perform atmospheric delay correction on the completed data set to obtain corrected altitude data;

[0062] (8) Perform data consistency assessment on the corrected altitude data to obtain preprocessed multi-source satellite altimeter data.

[0063] Specifically, data format identification is performed. The purpose of this step is to determine the format type information of each data file, which includes the data encoding method, field structure, time format and unit, etc. Data format identification is completed by parsing the file header information and the predefined format template. After the identification is completed, the format type information of each data file is recorded to provide a basis for the subsequent decoding process. According to the data format type information, the multi-source satellite altimeter data set is decoded. The decoding process is to convert the encoded data into readable numerical information. The decoding algorithm depends on the format type identified previously, and different decoding algorithms are used for different encoding methods. For example, for binary-encoded data, the corresponding decoding library can be used to convert it into a floating point number or integer value; for text-encoded data, the string parsing method can be used to extract the numerical information. The decoded data includes sea surface height measurements, timestamps, orbit information, etc., forming a decoded data set.

[0064] Time synchronization is performed on the decoded data to ensure that the data from different satellites are aligned on the same time axis. Time synchronization requires converting the timestamps of each data source into a unified time format, such as UTC (Coordinated Universal Time). For data sources with different time resolutions, time alignment can be performed by interpolation. Linear interpolation is a commonly used time interpolation method that achieves time alignment by inserting new data points between two known time points. The result of time synchronization is time-consistent data, ensuring that the time dimension of the data is consistent in subsequent analysis. Spatial coordinate conversion is performed on the time-consistent data to unify the spatial coordinates of each satellite data into the same coordinate system, usually using the geocentric coordinate system. Spatial coordinate conversion requires using the coordinate conversion formula to convert the measured values ​​in the original coordinate system into the geocentric coordinate system based on the satellite's orbital parameters and measurement position. For example, converting geographic coordinates (longitude, latitude, altitude) into geocentric rectangular coordinates (X, Y, Z) ensures the spatial consistency of data from different satellites. The converted data is the unified coordinate system data.

[0065] Outlier detection is performed on unified coordinate system data to identify and mark outliers in the data. Outlier detection methods include statistical analysis and machine learning algorithms. Common statistical analysis methods include the 3σ rule and the box plot method. The former calculates the mean and standard deviation of the data and determines that the data beyond the range of 3 times the standard deviation is an outlier; the latter analyzes the quartiles of the data and determines that the data beyond the upper and lower whiskers is an outlier. Machine learning algorithms such as the isolation forest algorithm identify abnormal patterns in the data by training models. The result of outlier detection is a data set marked with outliers, that is, outlier labeled data.

[0066] Data interpolation is performed on the outlier marked data to fill in the outliers and missing values ​​in the data and ensure the integrity of the data. Interpolation methods include linear interpolation, spline interpolation, and Kriging interpolation. Linear interpolation is simple and efficient by inserting new data points between two known data points; spline interpolation is suitable for smooth data through polynomial function interpolation; Kriging interpolation is suitable for spatial data by considering spatial correlation through geostatistical methods. The interpolated data set is the completed data set. The completed data set needs to be corrected for atmospheric delay in order to eliminate the delay effect of the atmosphere on the altimeter signal and improve the measurement accuracy. Atmospheric delay mainly includes ionospheric delay and tropospheric delay. Ionospheric delay correction usually adopts the dual-frequency observation method to calculate the delay value by the difference of the altimeter signal at two frequencies; tropospheric delay correction usually adopts the model correction method to calculate the correction value by the tropospheric delay model. The corrected data is the corrected altitude data.

[0067] The data consistency of the corrected altitude data is evaluated to ensure the consistency and reliability of the data in time and space. The consistency evaluation methods include statistical analysis and cross-validation. The statistical analysis method evaluates the consistency of the data by calculating the mean, variance and autocorrelation coefficient of the data; the cross-validation method evaluates the reliability of the data by dividing the data into a training set and a validation set, training the model and verifying the prediction accuracy of the model. The data that passes the evaluation is the preprocessed multi-source satellite altimeter data.

[0068] For example, we first identify the formats of data sets from multiple satellites and find that the data from satellite A is encoded in binary and the data from satellite B is encoded in text. According to the identified format type, we use the corresponding decoding algorithms to perform binary decoding on the data from satellite A and text parsing on the data from satellite B to obtain decoded data sets. We perform time synchronization on the decoded data, convert the timestamps of satellite A and satellite B into UTC format, and align the time axis using linear interpolation to form time-consistent data. Next, we perform spatial coordinate conversion on the time-consistent data, converting the geographic coordinates into the geocentric coordinate system to obtain unified coordinate system data.

[0069] In the process of outlier detection, the 3σ rule is used to identify some outliers that are beyond the normal range, and these outliers are interpolated using the linear interpolation method to form a complete data set. Next, the ionospheric and tropospheric delay corrections are performed on the complete data set through the dual-frequency observation method and the tropospheric delay model to obtain the corrected altitude data. Finally, the corrected altitude data is evaluated for data consistency through statistical analysis and cross-validation methods to ensure the reliability and consistency of the data, and finally the preprocessed multi-source satellite altimeter data is formed.

[0070] Through the above processing steps, comprehensive processing and preprocessing of multi-source satellite altimeter data sets are achieved, ensuring the quality and consistency of the data, and laying a solid foundation for subsequent satellite data fusion and analysis. The linear interpolation, dual-frequency observation method and tropospheric delay model used in this process effectively improve the accuracy and reliability of the data. In practical applications, this processing method can effectively cope with the complexity and diversity of multi-source data, provide high-quality satellite altimeter data, and provide important data support for ocean monitoring and environmental research.

[0071] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0072] (1) Extracting orbital parameters from the preprocessed multi-source satellite altimeter data to obtain a satellite orbital parameter set;

[0073] (2) Calculating the intersection points of the satellite orbit parameter set to obtain the multi-satellite intersection point position data;

[0074] (3) Temporally interpolating the multi-satellite intersection position data to obtain the intersection height value at the moment;

[0075] (4) Calculate the difference of the height values ​​of the intersection at each moment to obtain the height difference data of the intersection;

[0076] (5) Using the least squares method to perform orbit error modeling on the intersection height difference data, the orbit error model parameters are obtained;

[0077] (6) Perform time series analysis on orbit error model parameters to obtain the time-varying characteristics of orbit error;

[0078] (7) According to the time-varying characteristics of the orbit error, the orbit error is corrected for the preprocessed multi-source satellite altimeter data to obtain preliminary corrected data;

[0079] (8) Perform residual error evaluation on the preliminary correction data to obtain satellite orbit error correction data.

[0080] Specifically, in order to extract orbital parameters from the preprocessed multi-source satellite altimeter data, it is first necessary to extract the orbital parameters of each satellite from these data. Orbital parameters include orbital inclination, right ascension of ascending node, perigee parameters, etc. These parameters are important information describing the satellite's orbital trajectory and can be obtained by parsing satellite orbital data files or using satellite orbit calculation software. The extracted orbital parameters form a satellite orbital parameter set, which provides a basis for subsequent intersection calculation and orbital error correction. The purpose of performing intersection calculation on the satellite orbital parameter set is to find the intersection position data of different satellite trajectories. The intersection calculation is achieved by solving the orbital equation, which involves the mathematical description and solution method of the satellite orbit. For example, for two satellites A and B, the intersection position of their trajectories can be found by solving the intersection of their orbital equations. The intersection position data is the spatial coordinates and timestamp information of these intersections, which records the intersection points of different satellite trajectories in space.

[0081] Temporal interpolation is performed on the multi-satellite intersection position data to obtain the height value at the intersection time. Temporal interpolation is the process of aligning satellite height data at different times to the same time point. Common methods include linear interpolation and spline interpolation. Linear interpolation inserts new data points between two known data points to achieve a smooth transition of height values ​​in time; spline interpolation interpolates through polynomial functions and is suitable for situations where height values ​​change greatly. The result of temporal interpolation is the height value at the intersection time, ensuring the temporal consistency of different satellite data. The height values ​​at the intersection time are calculated by difference to obtain the intersection height difference data. Difference calculation is to compare the height values ​​of different satellites at the same intersection time and calculate their difference. For example, for satellites A and B, at a certain intersection time, the height values ​​are 10.5 meters and 10.7 meters respectively, with a difference of 0.2 meters. The intersection height difference data records these differences and provides a basis for orbit error modeling.

[0082] The orbit error model is built on the intersection height difference data by the least squares method to obtain the orbit error model parameters. The least squares method is a commonly used data fitting method. The optimal model parameters are obtained by minimizing the square error between the observed data and the model prediction value. In this case, the orbit error model is a mathematical model that describes the change of satellite orbit error over time. The least squares method obtains model parameters such as orbit deviation and error change rate by fitting the intersection height difference data. The purpose of performing time series analysis on the orbit error model parameters is to extract the time-varying characteristics of the orbit error. Time series analysis methods include autoregressive model (AR), moving average model (MA) and autoregressive moving average model (ARMA). Through time series analysis, the periodicity, trend and randomness characteristics of orbit error can be identified. For example, through the ARMA model, the change pattern of orbit error over time can be described, providing a basis for error correction.

[0083] According to the time-varying characteristics of orbital errors, orbital error correction is performed on the preprocessed multi-source satellite altimeter data to obtain preliminary correction data. Error correction is to apply the identified orbital error to the satellite altitude data to correct its measured value. For example, for a satellite altitude value of 10.5 meters at a certain moment, the identified orbital error is 0.2 meters, and the corrected altitude value is 10.3 meters. The preliminary correction data is the altitude value after orbital error correction, which has higher accuracy.

[0084] Perform residual error evaluation on the preliminary correction data to obtain satellite orbit error correction data. Residual error evaluation is to evaluate the quality of data after orbit error correction and judge the effect of error correction. Evaluation methods include residual analysis and error propagation analysis. Residual analysis evaluates the accuracy of error correction by calculating the residual between the corrected data and the true value; error propagation analysis evaluates the stability of error correction by analyzing the propagation path of the error during data processing. The data that passes the evaluation is the final satellite orbit error correction data.

[0085] For example, by preprocessing multiple satellite altimeter data, the orbital parameters of satellites A, B, and C are first extracted to form an orbital parameter set. Then, the intersection position data of these satellites are calculated by solving the orbital equation, and the spatial coordinates and timestamps of the intersection are recorded. Next, the intersection position data are aligned in time using the linear interpolation method to obtain the height value at the intersection time. At a certain intersection time, the height value of satellite A is 10.5 meters, and the height value of satellite B is 10.7 meters. The height difference is calculated to be 0.2 meters. These height difference data are fitted by the least squares method to establish an orbit error model and obtain the model parameters. Time series analysis shows that the orbit error has obvious periodic variation characteristics. According to the results of time series analysis, orbit error correction is performed on the preprocessed satellite altitude data, and the identified errors are applied to the altitude data to obtain preliminary correction data. The residual error evaluation of the preliminary correction data shows that the residual between the corrected data and the true value is significantly reduced, and the error correction effect is significant, and finally high-precision satellite orbit error correction data is obtained.

[0086] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0087] (1) Perform coordinate system conversion on the satellite orbit error correction data to obtain the corrected satellite measured sea surface height in a unified geocentric coordinate system;

[0088] (2) Correcting the sea surface height measured by the correction satellite in the unified geocentric coordinate system for tidal effects to obtain tidal-free sea surface height data;

[0089] (3) Correcting the sea surface height data without tidal effects by atmospheric pressure to obtain the sea surface height data under standard atmospheric pressure;

[0090] (4) performing geoid calculation on the sea level height data under standard atmospheric pressure to obtain geoid height data;

[0091] (5) Perform regional harmonic analysis on geoid height data to obtain a local optimized geoid model;

[0092] (6) Perform time series analysis on the local optimized geoid model to obtain dynamic geoid height data;

[0093] (7) performing mean sea surface modeling on the acquired historical satellite altimeter data by using an empirical orthogonal function decomposition algorithm to obtain an initial mean sea surface model;

[0094] (8) Performing ocean dynamics correction on the initial average sea surface model to obtain optimized average sea surface height data;

[0095] (9) Subtracting the dynamic geoid height data and the optimized average sea surface height data from the sea surface height data under standard atmospheric pressure to obtain preliminary along-track sea surface height anomaly data;

[0096] (10) Perform high-frequency noise filtering on the preliminary along-track sea surface height anomaly data to obtain the along-track sea surface height data.

[0097] Specifically, the data is converted from the original satellite orbit coordinate system to a unified geocentric coordinate system. The geocentric coordinate system is a three-dimensional rectangular coordinate system with the center of mass of the earth as the origin, which is usually used for global geographic information processing. The conversion process requires the use of the satellite's orbital parameters and measurement position, and the original coordinates are converted into geocentric coordinates through the coordinate conversion formula. The core of the formula is to use the satellite's orbital parameters (such as orbital inclination, right ascension of the ascending node, etc.) and the longitude, latitude and altitude of the measurement point to calculate its X, Y, and Z coordinates in the geocentric coordinate system. After conversion, the corrected satellite-measured sea level height in a unified geocentric coordinate system is obtained.

[0098] The purpose of correcting the sea surface height measured by the correction satellite in the unified geocentric coordinate system is to remove the influence of tide on the sea surface height measurement. The tidal effect is mainly caused by the gravitational force of the moon and the sun, and the influence on the sea surface height is periodic. The correction process uses a tidal model, such as a global tidal model or a local tidal model, to calculate the influence of the tidal effect at different times and locations, deduct it from the measured height, and obtain the sea surface height data without tidal effect.

[0099] The sea surface height data without tidal effects also needs to be corrected for atmospheric pressure effects. Changes in atmospheric pressure will cause fluctuations in sea surface height. In order to eliminate this effect, the sea surface height data needs to be corrected to standard atmospheric pressure. The correction method usually uses the atmospheric pressure correction formula, which calculates the difference between the actual atmospheric pressure and the standard atmospheric pressure, and makes corresponding height corrections based on the sensitivity of the sea surface height to atmospheric pressure changes. The sea surface height data obtained in this way eliminates the influence of atmospheric pressure, that is, the sea surface height data under standard atmospheric pressure.

[0100] Next, the geoid is calculated for the sea surface height data under standard atmospheric pressure. The geoid is the average sea surface height when the earth's surface is at rest, and is used as a benchmark for height measurement. The calculation of the geoid depends on the earth's gravity field model and the sea surface height data. The geoid height data can be calculated by combining the sea surface height measured by the satellite altimeter with the earth's gravity field model.

[0101] In order to improve the accuracy of the geoid model, it is necessary to perform regional harmonic analysis on the geoid height data. Regional harmonic analysis extracts the geoid characteristics of the region by performing frequency domain analysis and filtering on the geoid data of the local area, and obtains a locally optimized geoid model. This can better reflect the geoid changes in the local area and improve the accuracy of the model.

[0102] The purpose of time series analysis on the locally optimized geoid model is to extract the dynamic change characteristics of the geoid. Time series analysis methods include autoregressive model (AR), moving average model (MA) and autoregressive moving average model (ARMA). Through time series analysis, the periodicity and trend characteristics of the geoid height over time can be identified, and the dynamic geoid height data can be obtained.

[0103] The empirical orthogonal function decomposition algorithm is used to model the mean sea surface of the acquired historical satellite altimeter data. The empirical orthogonal function decomposition algorithm is a method to decompose complex data into several orthogonal patterns, which is often used for pattern recognition of meteorological and oceanographic data. By decomposing the historical satellite altimeter data, the main spatial and temporal variation patterns can be extracted and the initial mean sea surface model can be established. The initial mean sea surface model reflects the main variation characteristics of the sea surface height in the historical data.

[0104] The purpose of performing ocean dynamic correction on the initial mean sea surface model is to take into account the impact of ocean dynamic processes on sea surface height. Ocean dynamic processes include ocean currents, wind stress, etc., which have a significant impact on sea surface height. By combining the ocean dynamic model with the initial mean sea surface model, the deviation in the initial model can be corrected to obtain more accurate optimized mean sea surface height data.

[0105] Finally, the sea surface height data under standard atmospheric pressure is subtracted from the dynamic geoid height data and the optimized average sea surface height data to obtain the preliminary along-track sea surface height anomaly data. The along-track sea surface height anomaly data reflects the deviation of the sea surface height measured along the satellite trajectory relative to the average sea surface, and is used to analyze the abnormal changes in sea surface height.

[0106] In order to remove the high-frequency noise in the abnormal sea level data along the track, high-frequency noise filtering is required. Common methods for high-frequency noise filtering include Fourier transform and wavelet transform. By decomposing the data into the frequency domain or wavelet domain, high-frequency components are filtered out and low-frequency signals are retained. After high-frequency noise filtering, the final along-track sea level data is obtained.

[0107] For example, in a practical application, the satellite orbit error correction data is first converted from the original orbit coordinate system to the geocentric coordinate system. Assuming that the original coordinates of a certain measurement point are 120 degrees longitude, 30 degrees latitude, and 10 meters in height, the coordinates in the geocentric coordinate system calculated by the conversion formula are X=6378137 meters, Y=0 meters, and Z=0 meters. Next, the global tidal model is used to correct the sea surface height data without tidal effects. Assuming that the tidal effect is 0.5 meters, the height after correction is 9.5 meters. Then, according to the atmospheric pressure correction formula, assuming that the actual atmospheric pressure is 10 hPa higher than the standard atmospheric pressure, the sea surface height under standard atmospheric pressure is corrected to 9.3 meters.

[0108] In the geoid calculation, the geoid height of the point is calculated to be 5 meters using the earth's gravity field model and measurement data. After regional harmonic analysis, the geoid characteristics of the area are extracted to obtain a locally optimized geoid model. Through time series analysis, the dynamic change characteristics of the geoid are identified, and the dynamic geoid height data is 5.2 meters. The historical data is analyzed using the empirical orthogonal function decomposition algorithm to establish an initial average sea level model, and the optimized average sea level height data is 5.1 meters through ocean dynamics correction. Finally, the sea level height data under standard atmospheric pressure is 9.3 meters minus the dynamic geoid height data of 5.2 meters and the optimized average sea level height data of 5.1 meters, and the preliminary along-track sea level height anomaly data is -1 meter. Through high-frequency noise filtering, the high-frequency components in the data are removed, and the final along-track sea level height data is -0.9 meters.

[0109] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0110] (1) Analyze the statistical characteristics of the sea level height data along the track to obtain the characteristic parameters of the sea level height distribution;

[0111] (2) According to the characteristic parameters of sea surface height distribution, the threshold of the sea surface height data along the track is set to obtain the abnormal height screening threshold;

[0112] (3) Screening the along-track sea surface height data for abnormal data based on the abnormal height screening threshold to obtain a preliminary abnormal height data set;

[0113] (4) Perform spatial cluster analysis on the preliminary abnormal height data set to obtain the spatial distribution characteristics of abnormal height;

[0114] (5) According to the spatial distribution characteristics of abnormal height, the boundary of the preliminary abnormal height data set is extracted to obtain the abnormal height data along the coast;

[0115] (6) Decompose the coastal height anomaly data into time series to obtain time trend terms and periodic terms; perform multi-scale analysis on the time trend terms and periodic terms through wavelet transform to obtain multi-scale time characteristics;

[0116] (7) Perform spatial correlation analysis on the coastal height anomaly data to obtain spatial autocorrelation characteristics; construct a spatiotemporal kriging interpolation model based on multi-scale temporal characteristics and spatial autocorrelation characteristics to obtain optimized interpolation parameters;

[0117] (8) By optimizing the interpolation parameters, the coastal height anomaly data is temporally and spatially interpolated to obtain the interpolated coastal height anomaly data.

[0118] Specifically, the statistical characteristics of the sea level data along the track are analyzed, and the statistical quantities such as mean, standard deviation, skewness and kurtosis are calculated to obtain the characteristic parameters of the sea level distribution. These parameters reflect the overall distribution of the sea level. Subsequently, the abnormal height screening threshold is set based on these distribution characteristic parameters. The method of adding or subtracting multiple standard deviations to the mean is usually used, such as the mean ± 3 times the standard deviation, as the preliminary abnormal height screening threshold.

[0119] The set threshold is used to filter the along-track sea surface height data, and the data that exceeds the threshold range is marked as the preliminary abnormal height data set. This step can effectively remove obvious outliers, such as extreme measurements caused by instrument failure or atmospheric interference. Next, the preliminary abnormal height data set is subjected to spatial clustering analysis, and the DBSCAN (density clustering) algorithm is used to divide the abnormal points into different clusters according to their spatial distribution characteristics. This step can identify the spatial distribution pattern of abnormal heights and distinguish between large-scale abnormal areas and isolated abnormal points.

[0120] Based on the clustering results, the boundary of the preliminary abnormal height data set is extracted. The alpha-shape algorithm is used to construct the outline of the abnormal area and obtain the abnormal coastal height data. This step can accurately locate the boundary of the abnormal height area and provide the spatial scope for subsequent analysis. The obtained coastal height anomaly data is decomposed into time series terms, periodic terms and residual terms using the STL (Seasonal and Trend decomposition using Loess) method. This decomposition can reveal the long-term change trend and seasonal fluctuation of the sea surface height anomaly.

[0121] In order to deeply analyze the temporal characteristics, the time trend term and the periodic term are analyzed at multiple scales by wavelet transform. Morlet wavelet is selected to decompose the changes at different time scales and obtain multi-scale temporal characteristics. This step can capture the variation characteristics of sea surface height anomalies at different time scales, such as interannual variation and seasonal variation. At the same time, the spatial correlation analysis of the coastal height anomaly data is carried out, and the Moran's I index is calculated to obtain the spatial autocorrelation characteristics. This analysis can reveal the spatial dependence of the sea surface height anomaly and provide a basis for subsequent interpolation.

[0122] Combining multi-scale temporal characteristics and spatial autocorrelation characteristics, a spatiotemporal kriging interpolation model is constructed. A spatiotemporal variogram model, such as a product model or a sum model, is selected to comprehensively model temporal and spatial correlations. Interpolation parameters, such as the range parameter and sill value of the spatiotemporal variogram, are optimized through cross-validation. Finally, the spatiotemporal kriging interpolation of the coastal height anomaly data is performed using the optimized interpolation parameters to obtain interpolated coastal height anomaly data. This step can fill in the gaps in the observation data and generate a continuous sea surface height anomaly field.

[0123] For example, when processing multi-source satellite altimeter data in a certain sea area, statistical analysis showed that the mean of the sea surface height in this area was 0.5 meters and the standard deviation was 0.2 meters. The threshold for screening abnormal height was set as mean ± 3 times the standard deviation, i.e. -0.1 meters to 1.1 meters. The preliminary abnormal height data set screened out contained about 5% of the data points. Spatial clustering analysis identified three major abnormal areas with areas of 100 square kilometers, 50 square kilometers and 30 square kilometers respectively. Time series decomposition found obvious annual and semi-annual cycle changes. Wavelet analysis further revealed the interannual variation characteristics of 2-7 years. The Moran's I index obtained by spatial correlation analysis was 0.75, indicating a strong spatial positive correlation. Based on these characteristics, the constructed spatiotemporal kriging model achieved a correlation coefficient of 0.85 in cross-validation and a root mean square error of 0.1 meters. The interpolated coastal height anomaly data finally covered the entire study area with a spatial resolution of 0.25 degrees and a temporal resolution of 10 days.

[0124] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0125] (1) Identify the data source of the interpolated coastal height anomaly data and obtain multi-source data identification;

[0126] (2) According to the multi-source data identification, weights are assigned to the interpolated coastal height anomaly data to obtain the initial fusion weights;

[0127] (3) Perform spatial consistency test on the interpolated coastal height anomaly data and obtain spatial consistency score;

[0128] (4) Dynamically adjust the initial fusion weight according to the spatial consistency score to obtain the optimized fusion weight;

[0129] (5) Using the optimized fusion weights, the interpolated coastal height anomaly data are weighted averaged to obtain preliminary fusion data;

[0130] (6) Perform physical constraint optimization on the preliminary fused data to obtain gridded sea surface height anomaly fused data;

[0131] (7) Perform time series decomposition on the gridded sea surface height anomaly fusion data to obtain trend terms, periodic terms, and residual terms;

[0132] (8) Extract features of trend items, cycle items, and residual items through deep learning algorithms to obtain abnormal feature vectors;

[0133] (9) Correlation analysis is performed on the anomaly feature vector and the preset ocean dynamics model parameters to obtain the anomaly cause correlation degree;

[0134] (10) According to the correlation degree of the abnormal causes, the abnormal causes are probabilistically ranked to obtain the predicted abnormal causes.

[0135] Specifically, the data source of the interpolated coastal height anomaly data is identified, and the multi-source data identification is obtained by analyzing the timestamp, orbital parameters and sensor characteristics of the data. This process may involve data from different satellites such as Jason-1 / 2 / 3 and Sentinel-3. According to the multi-source data identification, the interpolated coastal height anomaly data is weighted, and the initial fusion weight is obtained by considering the measurement accuracy, spatiotemporal resolution and data coverage of each satellite. For example, Jason-3 may be assigned a weight of 0.4 due to its higher accuracy, while Sentinel-3 may receive a weight of 0.3.

[0136] Subsequently, the spatial consistency test of the interpolated coastal height anomaly data was performed, and the root mean square error (RMSE) and correlation coefficient of different data sources in the overlapping area were calculated to obtain the spatial consistency score. The spatial consistency score reflects the degree of consistency between different data sources, and high consistency indicates good data quality. The initial fusion weight is dynamically adjusted according to the spatial consistency score, and the weight of the data source with high consistency is increased, and the weight of the data source with low consistency is reduced, so as to obtain the optimized fusion weight.

[0137] The interpolated coastal height anomaly data are weighted averaged using optimized fusion weights, and the weighted average of each grid point is calculated to obtain preliminary fusion data. This step combines the advantages of multi-source data and improves the reliability of the data. Next, the preliminary fusion data is physically constrained and optimized. The variational analysis method is applied to introduce ocean dynamics constraints such as geostrophic balance and mass conservation. The gridded sea surface height anomaly fusion data is obtained by iteratively minimizing the cost function. This process ensures that the fusion results conform to the laws of physics.

[0138] The gridded sea surface height anomaly fusion data is decomposed into time series, and the STL (Seasonal and Trend decomposition using Loess) method is used to decompose the data into trend terms, periodic terms, and residual terms. The trend term reflects long-term changes, the periodic term captures seasonal fluctuations, and the residual term contains short-term fluctuations and noise. Through deep learning algorithms, such as the long short-term memory network (LSTM), the trend term, periodic term, and residual term are extracted to obtain the abnormal feature vector. The LSTM network can capture the long-term dependency of the time series and extract the key features of the sea surface height anomaly.

[0139] The extracted abnormal feature vector is correlated with the preset ocean dynamics model parameters, and the correlation coefficient and mutual information between the two are calculated to obtain the abnormal cause correlation. The preset ocean dynamics model parameters include sea temperature, salinity, wind field and other factors. Finally, the abnormal causes are ranked by probability according to the abnormal cause correlation to obtain the predicted abnormal cause. Factors with high correlation are considered to be more likely abnormal causes.

[0140] For example, in the process of fusion of multi-source satellite altimeter data in a certain sea area, the data of Jason-3 and Sentinel-3 were given initial weights of 0.4 and 0.3 respectively. The spatial consistency test showed that the RMSE of the two in the overlapping area was 3cm and the correlation coefficient was 0.95, showing a high degree of consistency. Based on this result, the weight of Jason-3 was slightly increased to 0.45, while Sentinel-3 remained unchanged. After weighted averaging, the average sea surface height anomaly in the study area of ​​the preliminary fused data was 15cm.

[0141] During the physical constraint optimization process, the geostrophic balance condition was introduced. After five iterations, the cost function value dropped from the initial 100 to 10, and the gridded sea surface height anomaly fusion data that met the dynamic constraints was obtained. Time series decomposition revealed that there was an obvious annual cycle change with an amplitude of about 10 cm. The feature vector extracted by the LSTM network contained 20 elements, reflecting the multiple time scale characteristics of the sea surface height anomaly.

[0142] Correlation analysis of these characteristics with the parameters of the ocean dynamics model showed that the highest correlation coefficient was with sea temperature, reaching 0.8, followed by wind field, with a correlation coefficient of 0.6. Based on these correlations, the predicted causes of the anomaly are ranked as follows: El Nino phenomenon (60% probability), sea level rise caused by strong winds (30% probability) and other factors (10% probability).

[0143] The above describes the data fusion method based on multi-source satellite altimeter data in the embodiment of the present application. The following describes the data fusion system based on multi-source satellite altimeter data in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the data fusion system based on multi-source satellite altimeter data includes:

[0144] The acquisition module 201 is used to acquire multi-source satellite altimeter data to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps and orbit information data;

[0145] The control module 202 is used to parse and perform quality control processing on the multi-source satellite altimeter data set to obtain pre-processed multi-source satellite altimeter data;

[0146] The correction module 203 is used to perform multi-satellite intersection analysis and orbit error data correction on the pre-processed multi-source satellite altimeter data to obtain satellite orbit error correction data;

[0147] A subtraction module 204 is used to perform a height subtraction process on the sea level measured by the correction satellite in the satellite orbit error correction data by the geoid water level and the average sea level to obtain the along-track sea level data;

[0148] The screening module 205 is used to screen the along-track sea level height data for abnormal height data to obtain abnormal coastal height data, and perform spatiotemporal interpolation processing on the abnormal coastal height data to obtain interpolated abnormal coastal height data;

[0149] The fusion module 206 is used to perform multi-source fusion on the interpolated coastal height anomaly data to obtain gridded sea surface height anomaly fusion data, and perform anomaly cause analysis on the gridded sea surface height anomaly fusion data to obtain predicted anomaly causes.

[0150] Through the collaboration of the above components, the data acquisition and analysis of multi-source satellite altimeter data were carried out, and the data source identification information was prioritized and the batch acquisition strategy was adopted to significantly improve the efficiency and integrity of data acquisition, ensuring the diversity and reliability of data sources. The data quality and consistency were improved through data integrity check, missing data supplementation and noise elimination processing steps. In the process of data analysis and quality control, data format recognition, decoding, time synchronization and spatial coordinate conversion were used to improve the analysis efficiency and time and space consistency of multi-source data. In particular, the accuracy and reliability of the data were further improved through measures such as outlier detection and atmospheric delay correction. For the multi-satellite intersection analysis and orbit error data correction steps, the orbit errors between multi-satellite data were effectively eliminated through orbit parameter extraction, intersection calculation, time interpolation and difference calculation. The orbit error model parameters were accurately modeled and corrected using methods such as least squares and time series analysis, significantly improving the accuracy and precision of satellite orbit error correction. Furthermore, by subtracting the height of the geoid and the average sea surface height from the sea surface height measured by the correction satellite, combined with the correction of tidal effect, atmospheric pressure effect and geoid calculation, the high accuracy and consistency of the sea surface height data along the track are ensured. In the process of abnormal height data screening and spatiotemporal interpolation processing, the outliers in the sea surface height data are effectively identified and corrected through statistical feature analysis, spatial clustering and spatiotemporal kriging interpolation model. The spatiotemporal interpolation accuracy of abnormal data is further improved by adopting multi-scale time feature and spatial autocorrelation feature analysis. Finally, through the steps of multi-source fusion and abnormal cause analysis, the efficient fusion of multi-source data and the accurate prediction of abnormal data are achieved by using methods such as data source identification, weight allocation, spatial consistency test and physical constraint optimization.

[0151] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0152] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A data fusion method based on multi-source satellite altimeter data, characterized in that: The data fusion method based on multi-source satellite altimeter data comprises: Perform multi-source satellite altimeter data collection to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps, and orbit information data; Performing analysis and quality control processing on the multi-source satellite altimeter data set to obtain pre-processed multi-source satellite altimeter data; Performing multi-satellite intersection analysis and orbit error data correction on the preprocessed multi-source satellite altimeter data to obtain satellite orbit error correction data; Subtracting the geoid water height and the average sea surface height from the sea surface height measured by the correction satellite in the satellite orbit error correction data to obtain along-track sea surface height data; Performing abnormal height data screening on the along-track sea level height data to obtain abnormal coastal height data, and performing spatiotemporal interpolation processing on the abnormal coastal height data to obtain interpolated abnormal coastal height data; The interpolated coastal height anomaly data is subjected to multi-source fusion to obtain gridded sea surface height anomaly fusion data, and the gridded sea surface height anomaly fusion data is subjected to an anomaly cause analysis to obtain a predicted anomaly cause.

2. The data fusion method based on multi-source satellite altimeter data according to claim 1, characterized in that: The multi-source satellite altimeter data collection is performed to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps and orbit information data, including: Obtaining data source identification information, and prioritizing the data source identification information to obtain a data source priority list; According to the data source priority list, multi-source satellite altimeter data are collected in batches to obtain batches of raw data; Performing a data integrity check on the batched original data to obtain an integrity check result; According to the integrity check result, the missing data of the batched original data are supplemented to obtain a completed data set; Performing noise elimination processing on the completed data set to obtain a noise-reduced data set; The denoised data set is processed into a unified data format to obtain the multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps and orbit information data.

3. The data fusion method based on multi-source satellite altimeter data according to claim 1, characterized in that: The step of parsing and quality controlling the multi-source satellite altimeter data set to obtain pre-processed multi-source satellite altimeter data includes: Performing data format identification on the multi-source satellite altimeter data set to obtain data format type information; According to the data format type information, decoding the multi-source satellite altimeter data set to obtain decoded data; Performing time synchronization processing on the decoded data to obtain time consistency data; Performing spatial coordinate conversion on the time consistency data to obtain unified coordinate system data; Performing outlier detection on the unified coordinate system data to obtain outlier marked data; Performing data interpolation on the outlier labeled data to obtain a completed data set; Performing atmospheric delay correction on the completed data set to obtain corrected altitude data; Performing data consistency assessment on the corrected altitude data to obtain the preprocessed multi-source satellite altimeter data.

4. The data fusion method based on multi-source satellite altimeter data according to claim 1, characterized in that: The performing multi-satellite intersection analysis and orbit error data correction on the pre-processed multi-source satellite altimeter data to obtain satellite orbit error correction data includes: Extracting orbital parameters from the preprocessed multi-source satellite altimeter data to obtain a satellite orbital parameter set; Performing intersection calculation on the satellite orbit parameter set to obtain multi-satellite intersection position data; Performing temporal interpolation on the multi-satellite intersection position data to obtain the intersection height value at the moment; Calculating the difference of the height values ​​of the intersection at each moment to obtain intersection height difference data; Performing track error modeling on the intersection height difference data by the least square method to obtain track error model parameters; Performing time series analysis on the orbit error model parameters to obtain time-varying characteristics of the orbit error; According to the time-varying characteristics of the orbit error, the preprocessed multi-source satellite altimeter data is subjected to orbit error correction to obtain preliminary correction data; Perform residual error evaluation on the preliminary correction data to obtain the satellite orbit error correction data.

5. The data fusion method based on multi-source satellite altimeter data according to claim 1, characterized in that: The step of performing a height subtraction process on the sea level measured by the correction satellite in the satellite orbit error correction data by subtracting the geoid water level and the average sea level to obtain the along-track sea level data comprises: Performing coordinate system conversion on the satellite orbit error correction data to obtain a correction satellite-measured sea surface height in a unified geocentric coordinate system; Performing tidal effect correction on the sea surface height measured by the correction satellite in the unified geocentric coordinate system to obtain tidal effect-free sea surface height data; Performing atmospheric pressure effect correction on the tidal effect-free sea surface height data to obtain sea surface height data under standard atmospheric pressure; Performing geoid calculation on the sea level height data under the standard atmospheric pressure to obtain geoid height data; Performing regional harmonic analysis on the geoid height data to obtain a local optimized geoid model; Performing time series analysis on the local optimized geoid model to obtain dynamic geoid height data; The average sea surface model is constructed by using the empirical orthogonal function decomposition algorithm to obtain the initial average sea surface model based on the historical satellite altimeter data. Performing ocean dynamics correction on the initial average sea surface model to obtain optimized average sea surface height data; Subtracting the dynamic geoid height data and the optimized average sea level height data from the sea level height data under the standard atmospheric pressure to obtain preliminary along-track sea level height anomaly data; The preliminary along-track sea level height anomaly data is subjected to high-frequency noise filtering to obtain the along-track sea level height data.

6. The data fusion method based on multi-source satellite altimeter data according to claim 1, characterized in that: The abnormal height data screening of the along-track sea level height data to obtain abnormal coastal height data, and the spatiotemporal interpolation processing of the abnormal coastal height data to obtain interpolated abnormal coastal height data, include: Performing statistical characteristic analysis on the along-track sea level height data to obtain sea level height distribution characteristic parameters; According to the sea level height distribution characteristic parameters, thresholds are set for the along-track sea level height data to obtain an abnormal height screening threshold; Performing abnormal data screening on the along-track sea level height data based on the abnormal height screening threshold to obtain a preliminary abnormal height data set; Performing spatial cluster analysis on the preliminary abnormal height data set to obtain the spatial distribution characteristics of the abnormal height; According to the spatial distribution characteristics of the abnormal height, the boundary of the preliminary abnormal height data set is extracted to obtain the abnormal height data along the coast; Decomposing the coastal height anomaly data in time series to obtain time trend items and period items; performing multi-scale analysis on the time trend items and period items by wavelet transform to obtain multi-scale time features; Performing spatial correlation analysis on the coastal height anomaly data to obtain spatial autocorrelation characteristics; constructing a spatiotemporal kriging interpolation model based on the multi-scale time characteristics and the spatial autocorrelation characteristics to obtain optimized interpolation parameters; The coastal height anomaly data is temporally and spatially interpolated by using the optimized interpolation parameters to obtain the interpolated coastal height anomaly data.

7. The data fusion method based on multi-source satellite altimeter data according to claim 1, characterized in that: The multi-source fusion of the interpolated coastal height anomaly data is performed to obtain gridded sea surface height anomaly fusion data, and the anomaly cause analysis of the gridded sea surface height anomaly fusion data is performed to obtain the predicted anomaly cause, including: Performing data source identification on the interpolated coastal height anomaly data to obtain a multi-source data identifier; According to the multi-source data identifier, weights are assigned to the interpolated coastal height anomaly data to obtain initial fusion weights; Performing a spatial consistency test on the interpolated coastal height anomaly data to obtain a spatial consistency score; Dynamically adjusting the initial fusion weight according to the spatial consistency score to obtain an optimized fusion weight; Using the optimized fusion weights, weighted averaging the interpolated coastal height anomaly data is performed to obtain preliminary fusion data; Performing physical constraint optimization on the preliminary fused data to obtain gridded sea surface height anomaly fused data; Performing time series decomposition on the gridded sea surface height anomaly fusion data to obtain trend terms, periodic terms and residual terms; Extract features of the trend term, period term and residual term through a deep learning algorithm to obtain an abnormal feature vector; Performing correlation analysis on the abnormal characteristic vector and the preset ocean dynamics model parameters to obtain the abnormal cause correlation degree; The abnormal causes are probabilistically sorted according to the abnormal cause correlation to obtain the predicted abnormal cause.

8. A data fusion system based on multi-source satellite altimeter data, used to execute the data fusion method based on multi-source satellite altimeter data as described in any one of claims 1 to 7, characterized in that: The data fusion system based on multi-source satellite altimeter data includes: An acquisition module is used to acquire multi-source satellite altimeter data to obtain a multi-source satellite altimeter data set, wherein the multi-source satellite altimeter data set includes sea surface height measurement values, timestamps and orbit information data; A control module, used for parsing and quality controlling the multi-source satellite altimeter data set to obtain pre-processed multi-source satellite altimeter data; A correction module, used for performing multi-satellite intersection analysis and orbit error data correction on the preprocessed multi-source satellite altimeter data to obtain satellite orbit error correction data; A subtraction module, used for performing height subtraction processing on the sea surface height measured by the correction satellite in the satellite orbit error correction data by subtracting the geoid water surface height and the average sea surface height to obtain along-track sea surface height data; A screening module is used to screen the along-track sea level height data for abnormal height data to obtain abnormal coastal height data, and to perform spatiotemporal interpolation processing on the abnormal coastal height data to obtain interpolated abnormal coastal height data; The fusion module is used to perform multi-source fusion on the interpolated coastal height anomaly data to obtain gridded sea surface height anomaly fusion data, and perform anomaly cause analysis on the gridded sea surface height anomaly fusion data to obtain predicted anomaly causes.

Images