Method and system for observing submesoscale structure of yellow sea cold water mass

Abstract

The present application relates to the technical field of marine observation and data processing, in particular to a kind of Yellow Sea cold water mass submesoscale structure observation method and its observation system, comprising the following steps: step one: data acquisition;Obtain multiple profile data by the temperature-salinity-depth instrument carried by wave energy profile float, import profile data into observation system, form profile data txt file, as raw data;Step two: feature recognition and three-dimensional reconstruction;Raw data is processed using python program, to obtain temperature-depth-time array and salinity-depth-time array;Feature recognition, analysis and reconstruction of three-dimensional structure of submesoscale phenomenon are carried out jointly, and the observation of the submesoscale structure of the Yellow Sea cold water mass is realized through the reconstructed three-dimensional structure.The present application realizes the dynamic monitoring of the boundary submesoscale process of the Yellow Sea cold water mass by integrating wave energy profile float and satellite remote sensing technology.

Description

A method and system for observing the sub-mesoscale structure of the cold water mass in the Yellow Sea Technical Field

[0001] This invention relates to the field of marine observation and data processing technology, and in particular to an observation method and system for the sub-mesoscale structure of the cold water mass in the Yellow Sea. Background Technology

[0002] Submesoscale processes are important phenomena in the ocean, typically exhibiting spatial scales of 10 to 100 kilometers and temporal scales of several days to tens of days. They play a crucial role in ocean dynamics, mass exchange, and ecosystems, especially submesoscale phenomena such as eddies and frontal structures, which have profound impacts on ocean energy distribution, climate change, and fisheries resources. However, due to the small scale and rapid changes of submesoscale processes, traditional observation techniques and numerical models struggle to effectively capture them, particularly in complex waters such as the Yellow Sea cold water mass, where existing methods still have limitations in terms of spatial resolution, temporal resolution, and data coverage.

[0003] While existing satellite remote sensing technologies can provide data over a wide area, their low spatial and temporal resolution makes it difficult to accurately capture the details of sub-mesoscale phenomena. Buoy observations, although providing high temporal resolution in-situ data, have limited coverage, offering only point-like observations and failing to effectively reflect regional sub-mesoscale processes. Numerical simulation methods, while providing detailed information, are sensitive to initial conditions and often struggle to accurately simulate the generation and dissipation of eddies in complex ocean areas. Multi-source data fusion methods can compensate for the shortcomings of single data sources, but they still face challenges in handling accuracy differences and spatiotemporal matching between different data sources.

[0004] Therefore, a new method is urgently needed to combine high-resolution field data with satellite remote sensing data, overcome the limitations of existing technologies, reconstruct the three-dimensional structure of submesoscale processes within the Yellow Sea cold water mass, and effectively visualize their spatial and temporal evolution. This method will provide more accurate and comprehensive technical support for the study of submesoscale processes. Summary of the Invention

[0005] To address the aforementioned shortcomings, the present invention aims to provide a method and system for observing the sub-mesoscale structure of the Yellow Sea cold water mass, thereby solving the problem that existing technologies cannot simultaneously acquire the multi-parameter three-dimensional structure of sub-mesoscale processes and their ecological effects.

[0006] To achieve the above objectives, this invention provides a method for observing the sub-mesoscale structure of the Yellow Sea cold water mass, comprising the following steps:

[0007] Step 1: Data Acquisition;

[0008] Multiple profile data were acquired using a CTD (Conductivity, Temperature, Depth) meter mounted on a wave energy profiling buoy. The profile data was then imported into the observation system to form a txt file of profile data, which served as the raw data.

[0009] Step 2: Feature recognition and 3D reconstruction;

[0010] The raw data was processed using a Python program to obtain temperature-depth-time arrays and salinity-depth-time arrays. The temperature-depth-time arrays and salinity-depth-time arrays were then used to identify features, analyze and reconstruct the three-dimensional structure of the sub-mesoscale phenomenon. The reconstructed three-dimensional structure enabled the observation of the sub-mesoscale structure of the Yellow Sea cold water mass.

[0011] As a preferred technical solution, in step one, the wave energy profiling buoy equipped with a temperature, salinity, and depth (TDT) meter descends from the sea surface to the bottom to obtain a descent profile; it then rises from the bottom to the sea surface to obtain an ascent profile. The descent and ascent profiles are repeated sequentially to acquire data from multiple descent and ascent profiles. The data from each descent and ascent profile are then imported into the observation system as profile data.

[0012] As a preferred technical solution, the sampling frequency of the temperature, salinity, and depth meter is ≥30 minutes / time, and the vertical resolution is ≤5 cm.

[0013] As a preferred technical solution, step two, feature recognition and 3D reconstruction, includes the following steps:

[0014] Step 21: Process the data from the ascending profile;

[0015] Step 22, multi-scale signal extraction based on frequency band separation;

[0016] Step 23: Dynamic feature identification, multi-source data fusion, and reproduction of the three-dimensional morphology, spatial location, and intensity evolution of the sub-mesoscale structure to achieve reconstruction from single-point observation to three-dimensional field.

[0017] As a preferred technical solution, step 21 involves processing the data from the ascending profile, including the following steps:

[0018] Step 211: Import the raw observation data into the Python program;

[0019] Step 212: Perform data analysis and profile direction filtering;

[0020] Step 213: Remove abnormal data and identify and clean the noise to obtain cleaned data;

[0021] Step 214: Calculate the density profile using the temperature profile and salinity profile;

[0022] Based on the data after cleaning, a profile diagram of temperature-depth-time and salinity-depth-time was drawn.

[0023] As a preferred technical solution, in step 212, the Python program reads the raw data in tabular form row by row, parses out the time, depth, temperature and salinity data, and stores them into the corresponding temperature-depth-time array and salinity-depth-time array respectively, and constructs a three-dimensional index structure of the raw data through the temperature-depth-time array and salinity-depth-time array;

[0024] The depth sequences in the temperature-depth-time array and the salinity-depth-time array are extracted and scanned. The motion state of the buoy is determined by analyzing the depth change gradient between adjacent time steps: when the depth decreases with time, it is identified as an uplift profile; uplift profiles are selected for data processing.

[0025] As a preferred technical solution, in step 213, after selecting the uplift profile, abnormal data exceeding the physical reasonable threshold is first removed from the data in the temperature-depth-time array and the salinity-depth-time array; then, the median absolute deviation method is used to identify and clean the isolated spike noise in the data to obtain the cleaned data.

[0026] Among them, the median absolute deviation method is used to identify and clean isolated spike noise in the data. The process is as follows: first, calculate the median of all data points, then calculate the absolute deviation of each data point from the median, and finally take the median of the obtained absolute deviations as the standard for measuring the degree of data dispersion.

[0027] After removing outliers, the original profile data is reconstructed using the Lagrange quadratic interpolation algorithm. The mathematical model for the interpolation calculation is as follows:

[0028] First, in the original observation sequence, select the three known measured points that are closest to the depth x of the target to be interpolated, and denote them as follows: , and Construct three Lagrange basis functions. , and The expanded form is as follows:

[0029]

[0030] Subsequently, a quadratic interpolation polynomial is constructed based on the aforementioned basis functions. :

[0031]

[0032] In the above formula:

[0033] x represents the target depth: the depth value on the preset standard isobath grid. In this embodiment, it is a specific value in the standard depth sequence generated starting from the sea surface and increasing at fixed intervals of 5cm.

[0034] x0, x1, and x2 represent the observation depths: these refer to the three actual sampling depth points in the original dataset that are closest to the target depth x in terms of vertical distance.

[0035] f(x0), f(x1), and f(x2) are measured values: corresponding to the measured temperature or salinity data at the observation depths x0, x1, and x2, respectively.

[0036] L2(x) is an estimated value: it refers to the standardized temperature or salinity data at the target depth x obtained through interpolation.

[0037] As a preferred technical solution, in step 22, a fourth-order Butterworth bandpass filter is used to process the time series data in the cleaned data: first, the time series is forward filtered, then the filtered time series result is reversed and reverse filtered, and finally, the sub-mesoscale temperature and salinity components with a period range of 2 to 15 days are output.

[0038] As a preferred technical solution, step 23, dynamic feature recognition and three-dimensional field reconstruction, includes the following steps:

[0039] First, the horizontal density gradient of seawater is calculated, and the OW parameters are calculated in combination with the flow field inverted from satellite SLA; at the same time, the frontogenesis function is calculated.

[0040] Then, the spatial intersection of the high-density gradient region, the negative OW parameter region, and the positive frontogenetic function region obtained from the calculation of the horizontal density gradient of seawater was identified as the key region for the interaction between the sub-mesoscale front and the vortex.

[0041] Finally, the optimal interpolation method was used to fuse multi-source data and reconstruct the three-dimensional structure of the sub-mesoscale phenomenon.

[0042] This invention also provides an observation system for the sub-mesoscale structure of the Yellow Sea cold water mass, which applies the observation method described above and includes a data acquisition module, a data processing module, a three-dimensional reconstruction module, and an impact assessment module.

[0043] The data acquisition module is used to deploy wave energy profile buoy arrays in the core area of ​​the Yellow Sea cold water mass;

[0044] The data processing module processes the raw data, including a fourth-order Butterworth filter algorithm module and a peak event detection module.

[0045] The three-dimensional reconstruction module uses data assimilation technology to fuse buoy profiles and satellite SLA data to construct a three-dimensional spatiotemporal distribution model of sub-mesoscale events;

[0046] The impact assessment module quantifies the contribution rate of sub-mesoscale processes to static height variation and analyzes their regulatory role on nutrient transport by combining vertical velocity profiles and chlorophyll distribution.

[0047] Also includes:

[0048] Data storage unit supports PB-level multi-source data parallel storage.

[0049] The visualization unit provides an interactive interface for 3D dynamic models.

[0050] This invention provides a method and system for observing the sub-mesoscale structure of the Yellow Sea Cold Water Mass (YSCWM). By integrating wave energy profiling buoys and satellite remote sensing technologies, it enables dynamic monitoring of sub-mesoscale processes at the boundary of the YSCWM. The system acquires temperature, salinity, and depth profile data at a 30-minute temporal resolution and a 5-centimeter vertical resolution. Combined with parameters such as sea surface height anomaly (SLA) and chlorophyll concentration (CHL) retrieved from satellites, it uses the gradient threshold method and frontogenesis function criterion to quantitatively analyze the generation and dissipation processes of sub-mesoscale eddies and fronts and their impact on static height and vertical chlorophyll distribution. Attached Figure Description

[0051] Figure 1 is a flowchart of the observation method for the sub-mesoscale structure of the Yellow Sea cold water mass in Example 1;

[0052] Figure 2 is a flowchart of feature recognition and 3D reconstruction in step two;

[0053] Figure 3 is a temperature and salinity profile of the cleaned data obtained in step 213;

[0054] Figure 4 shows the temperature-salinity-density profile obtained after filtering with a Butterworth filter.

[0055] Figure 5 is a block diagram of the observation system for the sub-mesoscale structure of the Yellow Sea cold water mass in Example 2. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0057] Example 1:

[0058] Referring to Figures 1 and 2, this invention provides a method for observing the sub-mesoscale structure of the Yellow Sea cold water mass, comprising the following steps:

[0059] Step 1: Data Acquisition.

[0060] Multiple profile data are acquired using a CTD (Conductivity, Temperature, and Depth) meter mounted on a wave energy profiling buoy. Specifically, the wave energy profiling buoy with the CTD meter descends from the sea surface to the bottom to obtain the descent profile; it then rises from the bottom to the sea surface to obtain the ascent profile. These descent and ascent profiles are repeated sequentially to acquire data from multiple descent and ascent profiles. The data from each descent and ascent profile is imported into the observation system to form a txt file containing the profile data, which serves as the raw data.

[0061] Data from each descending and ascending profile includes temperature, salinity, depth, and time. For example, temperature data at a specific time and depth can be represented as temperature-depth-time data; salinity data at a specific time and depth can be represented as salinity-depth-time data.

[0062] This embodiment was conducted in a field experiment at 35.18°N, 122.26°E. A wave energy profiling buoy was placed, equipped with a temperature, salinity, and depth (CTD) meter. The CTD meter had a sampling frequency of ≥30 minutes / time and a vertical resolution of ≤5 cm, supporting long-term storage and breakpoint resume. Simultaneously, sea surface height anomaly (SLA) data and chlorophyll (CHL) data from the Copernicus (CMEMS) satellite were acquired for comparison and verification.

[0063] The wave energy profiling buoy and the temperature-depth (CTD) meter used in this invention are both existing technologies. The wave energy profiling buoy can be the one described in application publication number CN111891295A, entitled "A Cold Water Mass Observation Device and Method for the Yellow Sea". The temperature-depth (CTD) meter can be obtained commercially.

[0064] Step 2: Feature recognition and 3D reconstruction.

[0065] The raw data (in tabular form) was imported into a Python program to obtain temperature-depth-time arrays and salinity-depth-time arrays. Feature recognition, analysis and reconstruction of the three-dimensional structure of the sub-mesoscale phenomenon were performed on the temperature-depth-time arrays and salinity-depth-time arrays. The sub-mesoscale structure of the Yellow Sea cold water mass was observed through the reconstructed three-dimensional structure.

[0066] Includes the following steps:

[0067] Step 21: Process the data from the ascending profile.

[0068] Includes the following steps:

[0069] Step 211: Import the raw observation data into the Python program.

[0070] Step 212: Perform data analysis and profile direction filtering.

[0071] Specifically, the Python program reads the raw data in tabular form row by row, parses out the time, depth, temperature, and salinity data, and stores them into the corresponding temperature-depth-time array and salinity-depth-time array, respectively; it then extracts the depth sequence (Depth array) from the temperature-depth-time array and salinity-depth-time array, scans it, and filters out the uplift profile.

[0072] The process of selecting rising profiles is as follows: the buoy's motion state is determined by analyzing the depth change gradient between adjacent time steps. When the depth decreases over time, it is identified as a rising profile. The rising profiles are then selected for data processing. This invention uses only rising profiles as the data basis for subsequent analysis. The physical basis for this selection is that during the descent of the wave energy profiling buoy, the mechanical core tightens the cable, a process directly dragged by the random undulations of the sea waves, resulting in jerky and shaky descent speeds, unstable flow fields around the sensor, and high data noise. However, during the ascent after bottoming out, the mechanical core releases, and the wave energy profiling buoy rises at a constant speed solely based on the constant net buoyancy provided by its own float. This dynamic characteristic ensures that the speed of water flowing past the sensor is constant and stable, thus obtaining observational data that is continuous in time and has the lowest noise level.

[0073] Step 213: Remove abnormal data and identify and clean the noise to obtain cleaned data.

[0074] For the temperature-depth-time array and salinity-depth-time array data in the selected uplift profile, outlier data exceeding the physically reasonable threshold are first removed; then, the median absolute deviation method (MAD) is used to identify and clean isolated spike noise in the data to obtain cleaned data.

[0075] The Median Absolute Deviation (MAD) method is used to identify and clean isolated spike noise in the data. The process involves first calculating the median of all data points, then calculating the absolute deviation of each data point from the median, and finally taking the median of the obtained absolute deviations as the standard for measuring the dispersion of the data. Compared with the standard deviation, the Median Absolute Deviation (MAD) method is less sensitive to extreme outliers, can more accurately preserve the true physical signal, and avoids misclassifying strong peak signals as noise.

[0076] After removing outliers, this embodiment addresses the non-uniform sampling problem caused by the vertical movement of wave energy profiling buoys. It employs a Lagrange quadratic interpolation algorithm to reconstruct the original profile data. The core logic of this algorithm lies in constructing a quadratic polynomial using three measured data points in the neighborhood of the target depth point. This polynomial maps the discrete, non-uniform data onto a standard uniform-depth grid with a vertical resolution of 5 cm, thus providing high-density data support for subsequent fine-grained thermocline calculations and 3D visualization.

[0077] Specifically, the mathematical model for interpolation calculation is constructed as follows:

[0078] First, in the original observation sequence, select the three known measured points that are closest to the depth x of the target to be interpolated, and denote them as follows: , and Construct three Lagrange basis functions. , and The expanded form is as follows:

[0079]

[0080] Subsequently, a quadratic interpolation polynomial is constructed based on the aforementioned basis functions. :

[0081]

[0082] The physical meanings of the variables in the above formula are as follows:

[0083] x (target depth): refers to the depth value on the preset standard isobath grid. In this embodiment, it is a specific value in the standard depth sequence generated with the sea surface as the starting point and increments at fixed intervals of 5 cm.

[0084] x0, x1, x2 (observation depth): refers to the three actual sampling depth points in the original dataset that are closest to the target depth x in terms of vertical distance.

[0085] f(x0), f(x1), f(x2) (measured values): These correspond to the measured temperature or salinity data at observation depths x0, x1, and x2, respectively.

[0086] L2(x) (estimated value): refers to the temperature or salinity standardized data at the target depth x obtained by interpolation.

[0087] Step 214: Calculate the density profile using the temperature profile and salinity profile.

[0088] Based on the data after cleaning, a profile diagram of temperature-depth-time and salinity-depth-time was drawn, as shown in Figure 3.

[0089] Step 214: First, obtain the temperature-depth-time and salinity-depth-time three-dimensional arrays of the target sea area obtained in the previous step, and ensure that the grids of the two sets of data correspond one-to-one at the time sampling points and depth levels through the spatiotemporal registration module. For each registered spatiotemporal data point, the program first converts the depth value into the corresponding hydrostatic pressure value based on the principle of hydrostatics and the latitudinal geographic information of the target sea area, thereby constructing a pressure data matrix that is completely consistent with the original temperature and salinity data dimensions.

[0090] Subsequently, based on the internationally accepted standard thermodynamic seawater equation of state, the aligned salinity, temperature, and calculated pressure values ​​are used as input variables, and nonlinear coupling calculations are performed point-by-point. This calculation step utilizes the Gibbs energy function or polynomial fitting model from the standard algorithm library to solve for the in-situ density or potential density at the specific spatiotemporal location. The program traverses all spatial depth layers and temporal sampling points, ultimately generating a three-dimensional array of density-depth-time.

[0091] Step 22: Multi-scale signal extraction based on frequency band separation.

[0092] Multi-scale frequency band separation is performed on the cleaned data to extract specific sub-mesoscale temperature, salinity, and density components.

[0093] A fourth-order Butterworth bandpass filter was used to process the time series data in the cleaned data. Specifically, a bidirectional zero-phase filtering technique was employed: the time series was first forward-filtered, and then the result was reversed for inverse filtering. Figure 4 shows the temperature, salinity, and density profiles after Butterworth filtering. The fourth-order Butterworth bandpass filter completely eliminates the phase shift introduced by the filter, ensuring that the separated submesoscale events (such as the time of frontal passage) are strictly aligned with their actual occurrence times on the time axis. Ultimately, the submesoscale temperature and salinity components with a period ranging from 2 to 15 days are output.

[0094] In this invention, the spatiotemporal scales of different ocean dynamic processes are defined as follows: mesoscale processes refer to dynamic processes with a spatial scale of tens to hundreds of kilometers, a time scale of 15 to 150 days, and mainly dominated by the balance of the Coriolis force (such as mesoscale eddies); while sub-mesoscale processes refer to dynamic processes with a spatial scale of 1 to 10 kilometers, a time scale of 2 to 15 days, significant non-geostrophic characteristics, and large vertical velocities.

[0095] Step 23: Dynamic feature identification, multi-source data fusion, and reproduction of the three-dimensional morphology, spatial location, and intensity evolution of the sub-mesoscale structure to achieve reconstruction from single-point observation to three-dimensional field.

[0096] Dynamic characteristics were identified by using the separated sub-mesoscale temperature and salinity components and joint physical criteria.

[0097] First, the horizontal density gradient of the seawater is calculated, and the Okubo-Weiss parameter (OW parameter) is calculated in combination with the flow field inverted from the satellite SLA; at the same time, the frontogenic function is calculated.

[0098] The OW parameter is a physical quantity used to quantitatively describe the motion state of fluid micro-elements. It is defined as the square of the fluid deformation rate tensor minus the square of the vorticity tensor. When the OW parameter is negative, it indicates that the rotational effect of the fluid exceeds the deformation effect, and this region is physically determined to be the core region of the vortex.

[0099] The frontogenetic function is a diagnostic measure used to describe how the kinematic effects of the flow field cause the gradient of fluid properties (such as temperature or density) to increase or decrease over time. A positive frontogenetic function value means that the flow field is sharpening the front through compression.

[0100] Then, the spatial intersection of the high-density gradient region, the negative OW parameter region, and the positive frontogenetic function region obtained from the calculation of the horizontal density gradient of seawater was identified as the key region for the interaction between the sub-mesoscale front and the vortex.

[0101] Finally, optimal interpolation (OI) was used to fuse multi-source data and reconstruct the three-dimensional structure of the sub-mesoscale phenomenon. OI, based on the error covariance matrix between the observed data and the background field, uses weighted calculations to fuse discrete buoy observation points and satellite observation surfaces onto a regular three-dimensional grid. This invention uses a climatological mean field as the background and buoy temperature-salinity data and satellite SLA data as observation inputs to ultimately generate a high-resolution three-dimensional voxel dataset. This dataset fully reproduces the three-dimensional morphology, spatial location, and intensity evolution of the sub-mesoscale structure within the Yellow Sea cold water mass region, completing the reconstruction from single-point observations to a three-dimensional field.

[0102] This invention uses wave energy profile buoys as the core data source, combined with satellite products and subsequent data processing and reconstruction processes, to achieve the identification and three-dimensional characterization of sub-mesoscale processes in cold water mass regions, as shown in Figures 3 and 4.

[0103] Example 2:

[0104] Referring to Figure 5, the present invention provides an observation system for the sub-mesoscale structure of the Yellow Sea cold water mass, including a data acquisition module, a data processing module, a three-dimensional reconstruction module, and an impact assessment module.

[0105] The data acquisition module is used to deploy a wave energy profiling buoy array in the core area of ​​the Yellow Sea cold water mass. Specifically, it deploys wave energy profiling buoys (Blue Elf buoys) equipped with CTD sensors to achieve temperature, salinity, and depth (TDM) profiling observations every 30 minutes, with a vertical resolution of ≤5 cm. The buoys integrate CTD sensors, have a data storage period of ≥30 days, and support breakpoint resume.

[0106] The data processing module processes the raw data, including a fourth-order Butterworth filtering algorithm and a frontogenesis event detection module. Specifically, a fourth-order Butterworth filter is used to eliminate noise and extract the 15-150 day mesoscale and 2-15 day sub-mesoscale energy components; the gradient threshold method is used to extract the vortex core region; and the frontogenesis function criterion is used to detect frontogenesis events, setting a judgment threshold of 1-100 km for horizontal scale and ≥0.01 s⁻¹ for vertical shear.

[0107] The 3D reconstruction module uses data assimilation technology to fuse buoy profiles and satellite SLA data to construct a 3D spatiotemporal distribution model of submesoscale events.

[0108] The impact assessment module quantifies the contribution of sub-mesoscale processes to static height variation and analyzes their regulatory role on nutrient transport by combining vertical velocity profiles and chlorophyll distribution.

[0109] Data storage unit supports PB-level multi-source data parallel storage.

[0110] The visualization unit provides an interactive interface for 3D dynamic models.

[0111] experiment:

[0112] From June 3 to July 4, 2022, a buoy was deployed at 35.18°N, 122.26°E to simultaneously acquire SLA and CHL data from the CMEMS satellite. Submesoscale signals were separated by band filtering, and the changes in temperature, salinity, and density profiles during the frontogenesis event from June 16 to 20, 2022, were compared to verify the chlorophyll response characteristics.

[0113] The system acquires temperature, salinity, and depth profile data with a time resolution of 30 minutes and a vertical resolution of 5 centimeters. Combined with parameters such as sea surface height anomaly (SLA) and chlorophyll concentration (CHL) retrieved from satellites, it uses the gradient threshold method and frontogenesis function criterion to quantitatively analyze the generation-dissipation process of sub-mesoscale eddies and fronts and their impact on static height and vertical chlorophyll distribution.

[0114] Experiments show that sub-mesoscale processes contribute 19.46% to the hydrostatic height variation of the upper water layer (2-30 meters) and significantly alter the surface chlorophyll enrichment pattern through vertical mixing.

[0115] This invention provides a method and system for observing the sub-mesoscale structure of the Yellow Sea Cold Water Mass (YSCWM), offering a high spatiotemporal resolution data framework for nearshore dynamics research and ecosystem simulation. By integrating wave energy profiling buoys and satellite remote sensing technologies, dynamic monitoring of sub-mesoscale processes at the boundary of the YSCWM is achieved. The system acquires temperature, salinity, and depth profile data at a 30-minute temporal resolution and a 5-centimeter vertical resolution. Combined with parameters such as sea surface height anomaly (SLA) and chlorophyll concentration (CHL) retrieved from satellites, the system uses the gradient threshold method and frontogenesis function criterion to quantitatively analyze the generation and dissipation processes of sub-mesoscale eddies and fronts and their impact on static height and vertical chlorophyll distribution.

[0116] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A method for observing the sub-mesoscale structure of the cold water mass in the Yellow Sea, characterized in that, Includes the following steps: Step 1: Data Acquisition; Multiple profile data were acquired using a wave energy profiling buoy equipped with a temperature, salinity, and depth (TDM) instrument. This data was then imported into the observation system to create a txt file, which served as the raw data. Step 2: Feature Recognition and 3D Reconstruction; The raw data was processed using a Python program to obtain temperature-depth-time and salinity-depth-time arrays. Feature recognition, analysis, and reconstruction of the 3D structure of the sub-mesoscale phenomenon were performed on the temperature-depth-time and salinity-depth-time arrays. The reconstructed 3D structure enabled observation of the sub-mesoscale structure of the Yellow Sea cold water mass. Step 2, feature recognition and 3D reconstruction, includes the following steps: Step 21, processing the data from the rising profile; Step 22, multi-scale signal extraction based on frequency band separation; Step 23, dynamic feature recognition, multi-source data fusion, reproducing the 3D morphology, spatial location, and intensity evolution of the sub-mesoscale structure, realizing the reconstruction from single-point observation to a 3D field; Step 21, processing the data from the rising profile, includes the following steps: Step 211, importing the raw observation data into a Python program; Step 212, parsing the data and filtering the profile direction; Step 213: Remove outlier data and identify and clean noise to obtain cleaned data; Step 214: Calculate density profile using temperature and salinity profiles: Based on the cleaned data, draw temperature-depth-time and salinity-depth-time profile diagrams; In step 212, the Python program reads the raw data in tabular form row by row, parses out the time, depth, temperature, and salinity data, and stores them into the corresponding temperature-depth-time and salinity-depth-time arrays respectively, constructing the density profile using the temperature-depth-time and salinity-depth-time arrays. A three-dimensional index structure for the original data is constructed; depth sequences are extracted from the temperature-depth-time array and the salinity-depth-time array and scanned. The buoy's motion state is determined by analyzing the depth change gradient between adjacent time steps: when the depth decreases with time, it is identified as an uplift profile; uplift profiles are selected for data processing; in step 213, after selecting the uplift profile, abnormal data exceeding the physically reasonable threshold is first removed from the data in the temperature-depth-time array and the salinity-depth-time array; then, the median absolute deviation method is used to identify isolated spike noise in the data. The data is cleaned to obtain cleaned data. The median absolute deviation method is used to identify and clean isolated spike noise in the data. The process is as follows: first, the median of all data points is calculated; then, the absolute deviation of each data point from the median is calculated; finally, the median of the obtained absolute deviations is used as a standard to measure the degree of data dispersion. After removing outliers, the original profile data is reconstructed using the Lagrange quadratic interpolation algorithm. The mathematical model for interpolation calculation is constructed as follows: First, in the original observation sequence, the three known measured points closest to the depth x of the target to be interpolated are selected, denoted as... 、 and Construct three Lagrange basis functions 、 and The expanded form is as follows: Subsequently, a quadratic interpolation polynomial is constructed based on the aforementioned basis functions. ： In the above formula: x is the target depth, referring to the depth value on the preset standard isobath grid, specifically the value in the standard depth sequence generated with the sea surface as the starting point and increasing at fixed intervals of 5 cm; x0, x1, and x2 are the observation depths, referring to the three actual sampling depth points in the original dataset that are closest to the target depth x in terms of vertical distance; f(x0), f(x1), and f(x2) are the measured values, corresponding to the measured temperature or salinity data at the observation depths x0, x1, and x2, respectively; L2(x) is the estimated value, referring to the standardized temperature or salinity data at the target depth x obtained through interpolation calculation; in step 22, a fourth-order Butterworth bandpass filter is used to filter the time series in the cleaned data. Data processing: First, the time series is forward filtered, then the filtered time series result is reversed and reverse filtered. Finally, the sub-mesoscale temperature and salinity components with a period range of 2 to 15 days are output. In step 23, dynamic feature identification and three-dimensional field reconstruction include the following steps: First, the horizontal density gradient of seawater is calculated, and the OW parameters are calculated in combination with the flow field inverted by satellite SLA; at the same time, the frontogenetic function is calculated; then, the spatial intersection of the high density gradient region, the negative OW parameter region, and the positive frontogenetic function region obtained from the calculation of the horizontal density gradient of seawater is identified as the key region for the interaction between the sub-mesoscale front and the vortex; finally, the optimal interpolation method is used to fuse multi-source data and reconstruct the three-dimensional structure of the sub-mesoscale phenomenon.

2. The method for observing the sub-mesoscale structure of the Yellow Sea cold water mass according to claim 1, characterized in that, In step one, the wave energy profiling buoy equipped with a temperature, salinity, and depth (TDT) meter descends from the sea surface to the bottom to form a descent profile; it then rises from the bottom to the sea surface to form an ascent profile. The descent and ascent profiles are cycled back and forth to acquire data from multiple descent and ascent profiles. The data from each descent and ascent profile are then imported into the observation system as profile data.

3. The method for observing the sub-mesoscale structure of the Yellow Sea cold water mass according to claim 2, characterized in that, The sampling frequency of the temperature, salinity, and depth meter is ≥30 minutes / time, and the vertical resolution is ≤5 cm.

4. An observation system for the sub-mesoscale structure of the Yellow Sea cold water mass, using the observation method described in any one of claims 1-3, characterized in that, The system includes a data acquisition module, a data processing module, a 3D reconstruction module, and an impact assessment module. The data acquisition module is used to deploy a wave energy profiling buoy array in the core area of ​​the Yellow Sea cold water mass. The data processing module processes the raw data, including a fourth-order Butterworth filtering algorithm module and a frontogenesis event detection module. The 3D reconstruction module fuses buoy profiling and satellite SLA data using data assimilation technology to construct a 3D spatiotemporal distribution model of submesoscale events. The impact assessment module quantifies the contribution rate of submesoscale processes to hydrostatic height variation and analyzes its regulatory effect on nutrient transport by combining vertical velocity profiles and chlorophyll distribution. The system also includes a data storage unit supporting PB-level multi-source parallel storage and a visualization unit providing an interactive interface for the 3D dynamic model.

Images