A mesoscale vortex three-dimensional temperature field inversion method, device, equipment and medium in the sea

Abstract

The application discloses a mesoscale vortex three-dimensional temperature field inversion method, device, equipment and medium, and relates to the field of ocean observation and data inversion. The method comprises the following steps: acquiring multi-source ocean observation data; identifying a mesoscale vortex based on the multi-source ocean observation data and determining a three-dimensional structure of the mesoscale vortex; constructing a mesoscale vortex three-dimensional temperature field inversion model by using a regression analysis method according to the three-dimensional structure of the mesoscale vortex based on the multi-source ocean observation data; solving regression coefficients in the mesoscale vortex three-dimensional temperature field inversion model by using a hierarchical least square method; and inversing a three-dimensional temperature field of the mesoscale vortex by using the mesoscale vortex three-dimensional temperature field inversion model according to the regression coefficients, while embedding a dynamic constraint in the inversion process to obtain the temperature of each standard layer of the mesoscale vortex. The application improves the precision and reliability of the inversion result of the temperature of each standard layer of the mesoscale vortex.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ocean observation and data inversion, and in particular to a method, device, equipment and medium for three-dimensional temperature field inversion of mesoscale eddies in the ocean. BACKGROUND

[0002] Mesoscale eddies in the ocean are the key carriers of ocean energy transfer and heat transport, and their three-dimensional temperature field characteristics are crucial for ocean heat balance, climate prediction and ocean resource development. Current means for obtaining their temperature field have limitations: in-situ observation has high precision, but lacks spatial coverage and time sampling, and is prone to observation blind spots; satellite remote sensing can observe a large area, but can only obtain the ocean surface temperature and cannot obtain the three-dimensional structure. Current inversion of the three-dimensional structure of mesoscale eddies is affected by the temporal and spatial resolution and data accuracy of satellite remote sensing, and cannot accurately depict the three-dimensional structure of mesoscale eddies.

[0003] In summary, the prior art cannot achieve high-precision and high-reliability inversion, and there is an urgent need for a method that can fully integrate multi-source data, adapt to the stratification characteristics of the ocean and comply with the laws of dynamics. SUMMARY

[0004] The purpose of the present application is to provide a method, device, equipment and medium for three-dimensional temperature field inversion of mesoscale eddies in the ocean, which can improve the precision and reliability of the temperature inversion results of each standard layer of mesoscale eddies.

[0005] To achieve the above-mentioned purpose, the present application provides the following solutions:

[0006] In a first aspect, the present application provides a method for three-dimensional temperature field inversion of mesoscale eddies in the ocean, comprising:

[0007] obtaining multi-source ocean observation data;

[0008] based on the multi-source ocean observation data, identifying mesoscale eddies and determining the three-dimensional structure of mesoscale eddies;

[0009] based on the multi-source ocean observation data, constructing a three-dimensional temperature field inversion model of mesoscale eddies using a regression analysis method according to the three-dimensional structure of mesoscale eddies;

[0010] solving the regression coefficients in the three-dimensional temperature field inversion model of mesoscale eddies using a stratified least squares method;

[0011] using the three-dimensional temperature field inversion model of mesoscale eddies to invert the three-dimensional temperature field of mesoscale eddies according to the regression coefficients, while embedding a dynamic constraint in the inversion process to obtain the temperature of each standard layer of mesoscale eddies.

[0012] In a second aspect, the present application provides a device for three-dimensional temperature field inversion of mesoscale eddies in the ocean, comprising:

[0013] The data acquisition module is configured to acquire multi-source marine observation data.

[0014] The mesoscale vortex identification module is configured to identify a mesoscale vortex based on the multi-source marine observation data and determine a three-dimensional structure of the mesoscale vortex.

[0015] The model construction module is configured to construct a three-dimensional temperature field inversion model of the mesoscale vortex based on the multi-source marine observation data and the three-dimensional structure of the mesoscale vortex by using a regression analysis method.

[0016] The coefficient solving module is configured to solve regression coefficients in the three-dimensional temperature field inversion model of the mesoscale vortex by using a hierarchical least square method.

[0017] The temperature inversion module is configured to invert a three-dimensional temperature field of the mesoscale vortex by using the three-dimensional temperature field inversion model of the mesoscale vortex according to the regression coefficients, while embedding a dynamic constraint in the inversion process to obtain temperatures of standard layers of the mesoscale vortex.

[0018] In a third aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the above-mentioned marine mesoscale vortex three-dimensional temperature field inversion method.

[0019] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the above-mentioned marine mesoscale vortex three-dimensional temperature field inversion method.

[0020] According to the embodiments provided in the present application, the present application has the following technical effects: the present application provides a marine mesoscale vortex three-dimensional temperature field inversion method, device, equipment and medium, by acquiring multi-source marine observation data, the comprehensiveness and reliability of the basic data required for inversion are ensured, high-quality data support is provided for subsequent mesoscale vortex three-dimensional temperature field inversion, information deviation caused by single data source limitation is avoided, the mesoscale vortex is first identified based on multi-source data and its three-dimensional structure is determined, then a regression analysis method is used for modeling, so that the mesoscale vortex three-dimensional temperature field inversion model can fit the actual shape of the vortex, the model specificity and rationality are improved, the blind modeling error caused by deviating from the actual structure is reduced, the hierarchical least square method is used to solve the regression coefficients, the characteristics of different layers of the ocean can be adapted, the coefficient calculation accuracy is improved, the fitting ability of the model to different layers of the ocean is enhanced, the dynamic constraint is embedded in the inversion, the inversion result is ensured to comply with the law of ocean dynamics, unreasonable temperature values are effectively avoided, and finally the reliability of the temperature inversion result of each standard layer of the mesoscale vortex is improved. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings described below only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0022] Figure 1 An application environment diagram of a three-dimensional temperature field inversion method of a mesoscale eddy in the sea according to an embodiment of the present application.

[0023] Figure 2 A flowchart of a three-dimensional temperature field inversion method of a mesoscale eddy in the sea according to an embodiment of the present application.

[0024] Figure 3 A functional module diagram of a three-dimensional temperature field inversion device of a mesoscale eddy in the sea according to an embodiment of the present application.

[0025] Figure 4 A structural diagram of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0027] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0028] The three-dimensional temperature field inversion method of a mesoscale eddy in the sea provided by the embodiments of the present application can be applied in an application environment as shown in the figure. Figure 1 In the application environment, the terminal 101 communicates with the server 102 through a network. The data storage system can store data required to be processed by the server 102. The data storage system can be separately arranged, or integrated on the server 102, or placed on the cloud or other servers. The terminal 101 can send multi-source marine observation data to the server 102. After receiving the multi-source marine observation data, the server 102 performs mesoscale eddy identification, and constructs a three-dimensional temperature field inversion model of a mesoscale eddy to invert the three-dimensional temperature field of the mesoscale eddy. The server 102 can feed back the temperature of each standard layer of the mesoscale eddy to the terminal 101. In addition, in some embodiments, the three-dimensional temperature field inversion method of a mesoscale eddy in the sea can also be realized by the server 102 or the terminal 101 alone.

[0029] The terminal 101 can be, but is not limited to, various desktop computers, notebook computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things device can be a smart speaker, a smart television, a smart air conditioner, a smart vehicle-mounted device, etc. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The server 102 can be implemented by a stand-alone server or a server cluster composed of multiple servers, and can also be a cloud server.

[0030] In an exemplary embodiment, as shown in Figure 2 A mesoscale vortex three-dimensional temperature field inversion method is provided, which is executed by a computer device, specifically, can be executed by a terminal or a server, or both. In the embodiments of the present application, the method is applied to the server 102 in Figure 1 , including the following steps 201 to 205.

[0031] Step 201, obtaining multi-source ocean observation data.

[0032] The multi-source ocean observation data includes a buoy temperature and salinity dataset, satellite altimeter data, sea surface temperature reanalysis data, a marine climatic dataset, and drift-type sea-air interface buoy observation data. Each data is described in detail as follows.

[0033] (1) The buoy temperature and salinity dataset adopts the Array for Real-time Geostrophic Oceanography (Argo) dataset, which has 58 vertical layers. The Argo dataset is subjected to real-time quality inspection and post-quality control. The original dataset includes about 600,000 temperature and salinity profiles covering the globe, as well as time and position information of the profile observation, and the profile extends from a depth of 2000m to the sea surface.

[0034] The screening conditions of the Argo dataset in the present application are as follows.

[0035] 1) The shallowest data in the Argo profile data is located between the sea surface and 10dbar, and the deepest data point corresponds to a depth greater than 1000dbar.

[0036] 2) The depth interval between two consecutive data points meets the following screening conditions: the data interval between 0-100dbar does not exceed 25dbar; the data interval between 100-300dbar does not exceed 50dbar; and the data interval between 300-1000dbar does not exceed 100dbar.

[0037] (2) In AVISO project, there are two kinds of global gridded sea surface height products: Sea Level Anomaly (SLA) and Absolute Dynamic Topography (ADT). SLA is approximately obtained by removing the Mean Sea Surface Height (MSSH) from the Sea Surface Height (SSH) in a certain period. The MSSH is usually the average of 7 years (1993-1999) of SSH. ADT is equal to SLA plus Mean Dynamic Topography (MDT). Here, MDT is calculated from 4.5 years of Gravity Recovery and Climate Experiment (GRACE) data, 15 years of altimeter data and in situ data. The purpose of removing an average value from the SSH or ADT is to weaken the influence of the mean flow to highlight the mesoscale signal. Therefore, SLA is more suitable for studying mesoscale eddies, while ADT is often used for general ocean circulation studies.

[0038] There are two kinds of SLA data: daily and weekly, which are fused from data of multiple orbit satellites such as T / P, Jason-1, Jason-2 and Geosat Follow-On. The data used in this application is daily SLA data.

[0039] In addition, it needs to be explained that although AVISO has corrected the altimeter product and eliminated the error caused by the tidal channel sea surface pressure, the influence still exists in the shallow water area. Previous studies will choose to remove the data in the area shallower than a certain depth contour (such as 200m), but given the particularity of this application (the research depth involves 200m or shallower), completely discarding the data in the shallow water area will inevitably affect the identification of eddies near the depth contour. Therefore, in this application, the data at land and large islands is removed, but the data at small islands is retained.

[0040] (3) The sea surface temperature reanalysis data is the Optimum Interpolation Sea Surface Temperature (OISST) dataset. The OISST is improved on the basis of the Comprehensive Ocean-Atmosphere Data Set (COADS) by using data from different observation platforms, such as satellites, buoys, and ship measurements. It strictly controls the quality of satellite observation data, including adjusting the bias of satellite observation and ship measurement (relative to the reference buoy), compensating for the differences between platforms and the bias of the sensor, and using the Optimum Interpolation (OI) algorithm to interpolate the buoy and ship data to the blank area of the remotely sensed sea surface temperature, thereby improving the spatial coverage and temporal and spatial resolution of the data. The starting time of the OISST dataset is 1981, and the horizontal resolution is 0.25°x 0.25°. In this application, the OISST data in the global range since 2016 is used.

[0041] (4) The ocean climatological data set.

[0042] The three-dimensional temperature field structure inversion of mesoscale eddies generally requires the use of outliers. In order to obtain these outliers, an average value needs to be removed from the Argo profile data. In actual operation, this average value can be replaced by climatological data. The climatological data used in this application is the WOA18 dataset. The WOA18 dataset provides objectively analyzed standard layer temperature and salinity field data with a spatial horizontal resolution of 1°x 1°. The WOA18 annual average data is used as the climatological temperature and salinity field data to calculate the temperature and salinity anomalies caused by eddies.

[0043] (5) The Drifting Air-Sea Interface Buoy (DrIB) observation data can measure 11 different physical parameters above the sea surface 3m, including air temperature, air pressure, wind direction, wind speed, relative humidity, and sea surface temperature (20cm underwater) and wave parameters. The observation SST data used in this application has a time interval of 1 hour and an observation accuracy of ±0.01℃.

[0044] Further, the above-mentioned multi-source ocean observation data is standardized. First, the sea surface height anomaly (SSH, SLA) data provided by the satellite altimeter, the in-situ observation (Argo) temperature and salinity vertical profile data, and the objective reanalysis data are integrated. The original data is strictly controlled for quality, and outliers are removed, and the in-situ observation is interpolated to the standard depth layer in the vertical direction to ensure the consistency and reliability of the data. The specific steps are as follows.

[0045] (1) Data source integration.

[0046] Satellite data: OISST, SLA, SSH, ADT.

[0047] In-situ observation data: Argo profile temperature data (0-200dbar), DrIB water temperature data.

[0048] Background field data: Climatological temperature profile Tclim(z), from WOA18.

[0049] (2) Data quality control.

[0050] In order to improve the accuracy of the results, the above data is quality controlled and screened. The quality control mainly includes two steps: observation layer quality control and standard layer quality control.

[0051] The observation layer quality control includes the following steps.

[0052] 1) Area test: select the range of the study area (0-180°E, 0-60°N) and the range of the study time (2018).

[0053] 2) Repeated depth data test: select the data with the same depth in each temperature-salinity observation profile and delete it.

[0054] 3) Delete data with large root mean square error, and calculate the average of the remaining data as the final observation value at that depth.

[0055] 4) Stability test: remove observation profiles that change too much in the vertical direction.

[0056] 5) Range control: select the data range that meets the actual situation and remove NaN values.

[0057] 6) Repeated profile test: Since the historical observation data collected comes from different databases, there may be multiple highly coincident or even repeated temperature-salinity observation profiles at the same location and same time, so select the profile data with better quality.

[0058] 7) Outlier removal: remove data with an absolute deviation greater than 3x quartile range.

[0059] The standard layer quality control includes the following steps.

[0060] 1) Constructing the standard layer dataset, the temperature observation profile after the above quality control is interpolated to construct the standard layer dataset, which is divided into 17 layers in the vertical direction: 0m, 5m, 10m, 15m, 20m, 25m, 30m, 35m, 40m, 45m, 50m, 75m, 100m, 125m, 150m, 175m, 200m. The stability test is performed on the standard layer data to obtain the in-situ observation temperature and salinity profile data set suitable for the application.

[0061] 2) The depth layer to be interpolated is calculated one by one. The vortex water layer data of the upper and lower layers of the interpolation layer is extracted. For example, for the first 40m water layer to be interpolated, the vortex water layer data of the 35m and 50m layers is extracted.

[0062] 3) The center of gravity of the upper and lower vortex zones is calculated, and the center of gravity position of the interpolation layer is calculated according to the distance between the interpolation layer and the upper and lower layers.

[0063] Step 202, based on the multi-source marine observation data, the mesoscale vortex is identified, and the three-dimensional structure of the mesoscale vortex is determined.

[0064] In a specific application example, when the mesoscale vortex is inverted, it is necessary to accurately identify the vortex core and vortex edge position of the mesoscale vortex.

[0065] The application is aimed at the sea surface area, based on the satellite altimeter data, and the parameter method (Okubo Weiss, OW) is used to identify the mesoscale vortex to determine the three-dimensional structure of the mesoscale vortex in the sea surface area.

[0066] The OW method starts from the physical properties of the mesoscale vortex, and identifies it through the sea surface height SSH or the sea surface temperature, and parameterizes the information of the physical field. When the parameter exceeds the set threshold, it is a mesoscale vortex. The application uses sea surface height anomaly SLA and sea surface height SSH to identify the vortex.

[0067] The OW number is defined as: ; wherein, W is the OW number, Sn is the normal strain, , Ss is the tangential strain, , w is the relative vorticity, . is the differential of the zonal component of the flow direction in the x-axis direction, is the differential of the zonal component of the flow direction in the y-axis direction, is the differential of the zonal component of the flow direction in the x-axis direction, is the differential of the zonal component of the flow direction in the y-axis direction.

[0068] Assuming the geostrophic balance, i.e. =0, the OW number can be simplified as: .

[0069] According to the SLA, the geostrophic relationship is calculated , where, u is the zonal component of the flow direction, v is the meridional component of the flow direction, g is the acceleration of gravity, taking 9.8 m / s 2 , f is the Coriolis parameter, , Ω is the angular velocity of the earth rotation, is the latitude, representing the effect of the earth rotation on the flow, with the unit of per second (1 / s). Then the OW number can be deformed as: ; where, is the mixed second-order partial derivative of the sea surface height; is the second-order derivative of the sea surface height in the east direction (curvature), reflecting the bending degree of the SLA contour in the east direction, and the positive value indicates the upward convex (local high pressure), and the negative value indicates the downward concave (local low pressure); is the second-order derivative of the sea surface height in the north direction, reflecting the bending degree of the SLA contour in the north direction.

[0070] If |W|<0.2σ w in a certain area, it is determined as a mesoscale vortex, σ w is the standard deviation of the OW number in the region. Specifically, the flow field is divided into different types: W>0.2σ w is mainly stretched, W<-0.2σ w is mainly vorticity, and the absolute value of W≤0.2σ w is the background flow field. The area of W<-0.2σ w is considered as the center of the vortex, and the positive and negative of the SLA value is used to determine whether it is a cyclonic vortex or an anticyclonic vortex.

[0071] For the area below the sea surface (5m layer-200m layer), based on the satellite altimeter data, the winding angle (Winding-Angle, WA) method is used to identify the mesoscale vortex, so as to determine the three-dimensional structure of the mesoscale vortex below the sea surface.

[0072] The core logic of the WA method is that vortexes are closed circulation structures with "rotationality" in the flow field, whose velocity vector field satisfies certain geometric constraints (such as linear variation of radial velocity component, consistent rotation direction), and the stream function contour is in a closed form. The WA method determines the vortex boundary and type by first screening vortex centers that meet the constraint conditions, then tracking the outermost closed streamlines centered on the center, and finally determining the vortex boundary and type. The whole process does not rely on the calculation of vorticity, divergence and other physical parameters, but only on the identification of the geometric characteristics of the flow field. In simple terms, first calculate the streamlines from the velocity field, then retain the closed streamlines according to the Winding-Angle principle, and then the streamlines gathered together are classified as a vortex. The implementation of the WA method is usually divided into two steps of vortex center detection and vortex boundary delineation. The input data of the WA method are SLA, SSH and their corresponding longitude and latitude and time data, and the output data are vortex edge longitude and latitude, vortex center longitude and latitude, vortex radius, etc.

[0073] (1) Vortex center detection: First, we need to find the points that may become the "seed" of the vortex in the flow field. The WA method is implemented by analyzing the behavior of the velocity vector around each grid point. The specific steps are as follows.

[0074] 1) Calculate the reference angle. For each point in the flow field (candidate point), check the velocity vectors of several adjacent points around it. Calculate the vector from the candidate point to the adjacent point : , is the radial vector. Calculate the angle between the velocity vector and the radial vector .

[0075] 2) Apply geometric constraints. Velocity direction constraint: the velocity vector should be roughly perpendicular to the radial vector . This means that is close to 90° or 270° (i.e. tangential flow), rather than towards or away from the center (radial flow). Rotation consistency constraint: the velocity directions of all surrounding points should exhibit a systematic clockwise or counterclockwise rotation pattern.

[0076] 3) Select candidate centers. Calculate the proportion of velocity vector points that meet the above constraints within a small area. If the proportion exceeds a certain threshold, the candidate point is marked as a candidate vortex center.

[0077] (2) Vortex boundary delineation: After finding the candidate vortex center, determine the boundary of the vortex. The specific steps are as follows.

[0078] 1) Calculate the stream function integral and the wrapping angle. Starting from the identified vortex center, the flow field is integrated in the opposite direction to generate a stream line. The wrapping angle is defined as the angle of the cumulative change in the direction of the velocity vector of a particle on the stream line when it moves around the vortex center. Specifically, the change in the direction of the velocity vector is calculated point by point along the stream line and accumulated to obtain the wrapping angle.

[0079] 2) Determine the closed stream line. The direction of the velocity vector of a stream line must rotate a full circle to form a closed loop. Therefore, the absolute value of the wrapping angle corresponding to a closed stream line must be close to 2π (360°). Specifically, the stream line integrated from the center is checked, and when it is found that the absolute value of the wrapping angle of a certain stream line first reaches or exceeds 2π, it is determined that the outermost closed stream line is found.

[0080] 3) Determine the final boundary. The area enclosed by the outermost closed stream line is the boundary of the vortex. At the same time, the type of vortex (cyclonic or anticyclonic) is determined according to the direction of stream line integration (clockwise or counterclockwise).

[0081] In addition, the SSH local extreme value in the grid range can also be used to determine the vortex center, and the stream lines near the vortex center are calculated, and the outermost closed stream line is used as the vortex boundary, and the SSH contour is used instead of the stream line, thereby saving the time of calculating the stream line.

[0082] Step 203, based on the multi-source ocean observation data, according to the three-dimensional structure of the mesoscale vortex, a three-dimensional temperature field inversion model of the mesoscale vortex is constructed by using regression analysis method.

[0083] In one specific application example, first, integrate historical observation and reanalysis data, establish a predictive mathematical function by quantifying the statistical correlation between dependent variables and independent variables, and then apply it to temperature field prediction. The present application designs 17 standard layers (0m, 5m, 10m, 15m, 20m, 25m, 30m, 35m, 40m, 45m, 50m, 75m, 100m, 125m, 150m, 175m, 200m). Step 203 includes steps 31 to 36.

[0084] Step 31, according to the sea surface temperature reanalysis data to determine the sea surface temperature data (SST), reflecting the intensity of the vortex thermal forcing (cyclonic vortex corresponds to cold anomaly, anticyclonic vortex corresponds to warm anomaly).

[0085] Step 32, according to the satellite altimeter data to determine the sea level anomaly data (SLA), which represents the potential energy change of the vortex dynamic core and is directly related to the rotational speed and vertical scale of the vortex.

[0086] Step 33, determine sea surface thermal anomaly data (SST') according to the buoy temperature-salinity dataset (Argo temperature), the ocean climatological dataset (WOA18) and the drifting sea-air interface buoy observation data (DrIB observed water temperature), capture the asymmetric effect of temperature anomaly (such as the difference in nonlinear response of cold / warm eddies).

[0087] Step 34, determine thermal-dynamic coupling data according to the sea surface thermal anomaly data and the sea surface height anomaly data, quantify the thermal-dynamic coupling effect (such as the SST' and SLA cooperative influence on the thermocline deformation). The thermal-dynamic coupling data is the product of the sea surface thermal anomaly data and the sea surface height anomaly data, i.e. SST' x SLA.

[0088] Step 35, determine eddy intensity vertical structure quadratic modulation data according to the sea surface height anomaly data, represent the quadratic modulation of eddy intensity to vertical structure. The eddy intensity vertical structure quadratic modulation data is the square of the sea surface height anomaly data.

[0089] Step 36, use the sea surface height anomaly data, the sea surface thermal anomaly data and the thermal-dynamic coupling data as independent variables, and the temperature of each standard layer of the mesoscale eddy as dependent variable, and use regression analysis method to construct the mesoscale eddy three-dimensional temperature field inversion model. The SST in step 31 and the in step 35 are data used in the calculation process.

[0090] For each standard depth layer, the present application establishes an independent mesoscale eddy three-dimensional temperature field inversion model: .

[0091] Wherein, z is the depth coordinate, which is positive from the sea surface ( z =0) downward, with the unit of meter, is the temperature of the standard layer at the depth of , is the sea surface thermal anomaly data, is the sea surface height anomaly data, is the thermal-dynamic coupling data, is a random error term, is the vertical transmission efficiency of the sea surface thermal anomaly to the deep water, which is regulated by the mixed layer convection (surface layer) and the turbulent diffusion (deep layer), , , , is the regression coefficient, represents the linear response coefficient of the sea surface thermal anomaly, which represents the temperature change caused by the displacement of the thermocline by SLA, linear response coefficient representing the sea surface height anomaly, characterizing the vertical displacement of thermocline corresponding to a unit change of SLA (exponentially decaying with depth), nonlinear response coefficient representing the sea surface thermal anomaly, quantifying the asymmetry of the cold / warm eddy thermal response, resulting from the differential response of mixed layer depth to eddy polarity, thermal-dynamic coupling response coefficient, characterizing the thermal-dynamic synergistic effect, reflecting the modulation of sea surface thermal anomaly on the displacement efficiency of thermocline.

[0092] Step 204, the hierarchical least squares method is used to solve the regression coefficients in the mesoscale eddy three-dimensional temperature field inversion model. During the hierarchical least squares solving process, an independent training set is constructed for each standard layer, and the target is defined as minimizing the mean square error of the predicted temperature and the observed value. The formulas of each regression coefficient are introduced below.

[0093] (1) Exponential decay + asymptotic background model is adopted: .

[0094] where, is the surface response intensity, , N is the total number of observation samples for statistical calculation, which is a dimensionless quantity, is the temperature anomaly of the n th observation, with the unit of °C, is the average value of the temperature anomaly, is the sea surface temperature anomaly of the n th observation, with the unit of °C, is the average value of the sea surface temperature anomaly.

[0095] is the deep background response, with the unit of °C / °C, , is the temperature anomaly measured or calculated at a selected deep reference depth (z n ) in the th observation, should be selected at a depth where the influence of turbulent mixing is basically eliminated and large-scale processes dominate.

[0096] is the decay depth scale, with the unit of meters, , is the turbulent diffusion coefficient, which quantifies the efficiency of vertical mixing of heat and momentum in the ocean due to turbulence (eddy), with the unit of m² / s, The larger the value is, the stronger the turbulent mixing is, and the higher the efficiency of heat downward transport is, where .

[0097] is a velocity scale that measures the strength of the momentum flux at the air-sea interface (i.e., the wind stress), with unit m / s. The stronger the wind, the larger the wind stress, and the larger the value of . It is determined by the sea surface wind stress and the sea water density : .

[0098] h is the boundary layer depth, usually refers to the depth of the ocean mixed layer or the boundary layer, with unit meter. It is the depth of the upper ocean that is fully mixed by wind waves and other processes, and the properties are uniform. It is the main range of action of turbulent mixing.

[0099] is the Richardson number, a dimensionless number, which is a key parameter to judge the stability of the fluid. It compares the stabilizing factor (represented by the density gradient ) that suppresses turbulence and the destabilizing factor (represented by the velocity shear ) that promotes turbulence. Its calculation formula is , indicates that the flow is stable, and the stratification suppresses turbulent mixing. indicates that the flow is unstable, and promotes the development of turbulence. indicates that the stratification is neutral. In this model, a larger positive number will significantly suppress the turbulent diffusion coefficient , thus making the decay depth scale smaller. In the context of the Richardson number formula, specifically refers to the vertical gradient of density, that is, the speed and direction of the change of density in the vertical direction (depth direction), and its complete mathematical expression is , indicates the partial derivative. In the context of the Richardson number formula, specifically refers to the speed of change of horizontal velocity in the vertical direction (depth direction), and its complete mathematical expression is .

[0100] (2) .

[0101] where is the climatological temperature gradient, which is a background field, indicating the intensity of the temperature change in the vertical direction of the ocean itself, with unit °C / m, , is the continuous form of the climatological temperature gradient, which is the mathematical partial derivative, is the approximate calculation of the discrete form of the climatological temperature gradient, is the climatological mean temperature at depth z, characterizing the vertical temperature profile of a specific ocean region in a long-term statistical sense; is the climatological temperature difference, the change of the climatological mean temperature within the depth interval [z - Az, z + Az]; is the depth interval.

[0102] is the isotherm vertical displacement efficiency factor, a dimensionless scaling factor that represents the amount of isotherm vertical displacement (m) per unit sea level anomaly (SLA) (m) at depth z . That is η , is the isotherm vertical displacement at depth z , with unit meter.

[0103] is the decay scale depth, with unit meter, a larger value indicates that the influence of the dynamic process can reach a deeper depth, usually related to barotropic and first baroclinic mode dynamic processes.

[0104] is the decay of the first mode (or the primary mode), indicating that the influence of isotherm displacement decays exponentially with depth starting from the sea surface.

[0105] C is the amplitude constant, dimensionless, used to control the strength of the high-order mode contribution.

[0106] is the depth at which the high-order mode has the most influence, with unit meter, at which the isotherm displacement has the strongest response to SLA.

[0107] is the width of the Gaussian function, with unit meter, determining the range (thickness) of the maximum influence, a larger value indicates a wider range of influence.

[0108] (3) Based on the mixed layer heat budget equation: ; where, is the time tendency term of the mixed layer temperature anomaly, representing the rate of change of the mixed layer temperature anomaly with time, i.e. the rate of warming or cooling, with unit °C / s. is the vertical temperature gradient at the bottom of the mixed layer (the temperature gradient at the thermocline not far below the bottom of the mixed layer), usually (the temperature decreases with depth). is the net surface heat flux, the heat entering the ocean is positive (heating the ocean), and the heat leaving the ocean is negative (cooling the ocean), is the specific heat capacity of seawater, ​is the mixed layer depth, is the entrainment velocity, , k is a proportionality coefficient, usually positive.

[0109] The final form is a piecewise Gaussian function: .

[0110] where, is the non-linear strength, and is the center depth of the Gaussian function, representing the depth where the temperature anomaly response is the largest due to the entrainment process, corresponds to the cold eddy, corresponds to the warm eddy, in meters. and is the width (standard deviation) of the Gaussian function, controlling the spread of the temperature anomaly response in the vertical direction, corresponds to the cold eddy, corresponds to the warm eddy, in meters.

[0111] (4) The displacement efficiency function is extended as: .

[0112] Substituting into the expression: .

[0113] The final form is: .

[0114] where, is the isopycnal displacement efficiency factor (extended version) that depends on depth and sea surface temperature anomaly, representing how much isopycnal displacement a unit SLA can cause at depth z under a specific SST' condition, dimensionless. is the base displacement efficiency function, representing the thermocline response without SST perturbation. is the modulation coefficient function, representing how much proportion of change a unit SST' anomaly can cause to the displacement efficiency at depth z , in 1 / °C. is the peak amplitude of the coupling strength SST' modulation strength, constant / °C -1 . is the maximum coupling depth (core depth of the thermocline), constant / m. is the coupling thickness, constant / m.

[0115] This application uses gradient descent to solve each regression coefficient, with the following steps.

[0116] (1) Initialize each regression coefficient , b= 1, 2, 3, 4, is the initial guess of the regression coefficient, and the superscript (0) indicates the 0th iteration.

[0117] (2) Iterative update: ; where, is the step size , meaning that only very small adjustments are made to the coefficients each time, which helps the optimization process to converge stably to the minimum point, rather than oscillating around the minimum or even diverging. is the regression coefficient after the th iteration update, is the regression coefficient after the th iteration update. MSE is the mean squared error, which measures the average difference between the predicted temperature and the observed value , and the smaller the MSE, the better the model fits. is the partial derivative of the mean squared error with respect to the regression coefficient , which indicates the steepest upward direction of the MSE function in the direction under the current coefficient value.

[0118] (3) Regularization: Introduce an L2 penalty term (hyperparameter λ = 0.01) to constrain the coefficient magnitude to prevent overfitting.

[0119] (4) Convergence criteria: Relative error change or iteration step number q > 10 4 If the first criterion is still not met after 10000 iterations, the algorithm is forced to stop to prevent infinite loops or waste of computing resources due to slow convergence.

[0120] The objective function of the least squares optimization is: . is the L2 penalty term to punish large coefficient values. is the regularization strength hyperparameter, which controls the proportion of the penalty term in the total objective function, , meaning that the algorithm still aims to minimize the fitting error as the main goal, while moderately constraining the coefficient size to prevent them from becoming excessively large.

[0121] Step 205, according to the regression coefficient, using the mesoscale vortex three-dimensional temperature field inversion model to invert the mesoscale vortex three-dimensional temperature field, while embedding the dynamic constraint in the inversion process, to obtain the temperature of each standard layer of the mesoscale vortex.

[0122] Wherein, the dynamic constraint includes quasi-geostrophic equilibrium constraint and sea surface boundary condition.

[0123] Quasi-geostrophic balance constraint: Based on the quasi-geostrophic property of mesoscale eddies, a closed relationship among temperature, pressure, and velocity is established.

[0124] Pressure-temperature correlation: .

[0125] Geostrophic velocity: , .

[0126] Sea surface boundary condition: .

[0127] where the integration path is from depth z to the sea surface ( z = 0), embodying the vertical integration property of hydrostatic balance. is the pressure anomaly at radial distance r and depth z , representing the deviation of pressure from the climatological mean field due to oceanic eddies and other activities, with the unit of pascal (Pa) or decibar (dbar). It is noted that the pressure on the sea surface is usually assumed to be constant (atmospheric pressure), so here actually refers to the pressure change acting on the bottom pressure gauge due to the density anomaly in the ocean interior, rather than the real atmospheric pressure anomaly.

[0128] is the thermal expansion coefficient, representing the relative change rate of the volume of seawater when the temperature increases by 1 °C at constant salinity and pressure, for typical seawater, is a positive value (about 2 × 10 -4 / °C), meaning that warming will lead to a decrease in density (volume expansion).

[0129] is the reference density of seawater, taking a constant value (such as ~ 1025 kg / m³), i.e., the density anomaly itself is ignored in the formula. is the temperature anomaly at radial distance r and depth z , with the unit of degree Celsius (°C).

[0130] is the eastward (x-direction) component of the geostrophic velocity, is the northward (y-direction) component of the geostrophic velocity, with the unit of meters per second (m / s). The positive and negative signs in the geostrophic velocity represent the direction of the Coriolis force, and the flow direction is always such that the Coriolis force points to the left of the flow direction (in the Northern Hemisphere), to balance the pressure gradient force pointing from high pressure to low pressure. is the gradient of the pressure anomaly p in the horizontal direction, is the rate of change of the pressure anomaly in the east direction, with unit of Pascal per meter (Pa / m). Geostrophic balance requires the flow to follow the pressure contours, in the northern hemisphere: if the pressure increases in the y direction ( ), a flow in the -x direction will be generated ( ). If the pressure increases in the x direction ( ), a flow in the +y direction will be generated ( ).

[0131] Further vertical mode separation:

[0132] Pressure anomaly decomposition: The core assumption is: the three-dimensional pressure anomaly field of a mesoscale vortex (or other quasi-geostrophic motion) can be decomposed into a product of a function r only related to the radial distance and a function z only related to the depth , i.e. the horizontal structure of the vortex at all depths is "self-similar", only the amplitude changes with depth. is the horizontal structure function of the pressure anomaly, which describes how the pressure anomaly of the vortex is distributed and attenuated in the horizontal direction. For example, it determines the radius and central intensity of the vortex. is the vertical structure function of the pressure anomaly, also known as the vertical mode, which describes how the amplitude of the pressure anomaly of the vortex changes with depth, and also determines the vertical structure of the flow velocity, etc.

[0133] Radial equation: , solution: , i.e. the modified Bessel function. Where, is the Rossby deformation radius, is the radius, is the second kind of modified Bessel function (order 0), the core structure of the vortex, which is finite divergence at , reflecting the central intensity of the vortex, the attenuation characteristics of the vortex periphery, exponential decay, and the limited range of influence of the vortex.

[0134] Vertical equation (inhomogeneous boundary): . Where, is the square of the buoyancy frequency, which is a key parameter for measuring the stability of the ocean stratification, , with unit of per second square, , the larger the value, the stronger the stratification of the ocean, the more stable the stratification, representing the restoring force of buoyancy to the disturbance in the vertical direction. is the deformation radius or Rossby radius, which is the most important scale parameter in the stratified fluid on a rotating earth, representing the balance scale between the rotation effect and the stratification effect. The horizontal scale of the motion (such as vortex, planetary wave) is usually related to of the same order of magnitude. The larger, the stronger the stratification effect, the wider the horizontal range that the motion can affect. , is the reduced gravity, , is the representative density difference, and H is the depth of the entire water column or the equivalent depth of the stratification layer.

[0135] Boundary condition: G|z=0, p 0, .

[0136] β2physical constraints: where, is the strength of the forcing applied at the ocean surface (z=0), i.e., the strength of the surface forcing, which is the external source driving the fluid motion (such as surface pressure anomalies, heat flux), its value is fixed by the boundary condition p 0.

[0137] The effectiveness of the mesoscale vortex three-dimensional temperature field inversion method provided in the present application is verified below.

[0138] (1) Objective analysis and interpolation.

[0139] The quartile range is used to remove abnormal points (|deviation|>3×quartile range). The quartile range is a statistical quantity that measures the degree of data dispersion, and its calculation method is the difference between the 75th percentile (Q3) and the 25th percentile (Q1) of the data, which represents the distribution range of the middle 50% data and is not sensitive to extreme values.

[0140] Gaussian correlation function weighted interpolation: , where, denotes the temperature value at the target grid point . is the weight of the observation point assigned to the first n observation, and the size of the weight depends on the spatial distance between the observation point and the target grid point. is the actual temperature (anomaly) measurement value of the observation point of the first n observation. is the radial distance between the observation point of the first n observation and the target grid point. is the correlation scale, which is a key parameter in the Gaussian function, and it determines how fast the weight decays with distance.

[0141] (2) Synthetic analysis: an polar coordinate system is established with the vortex center as the origin, and is divided into r / R bins (0.1 RThe median of T' in each bin is calculated. R is the radius of the vortex. r / R is the dimensionless radial distance, which is used to normalize the vortexes of different sizes so that their structures can be compared and synthesized. R The interval represents the width of the bin, which divides the vortex from the center (r=0) to the periphery (r=2) into a series of continuous annular regions ("bins") with a width of 0.1 times the radius of the vortex. r The median of T' in each bin is calculated. r =2 R The median of T' in each bin is calculated.

[0142] Radial smoothing: Savitzky-Golay filter (window=5, order=2), formula: where, is the smoothed temperature anomaly at radial distance r, r is the index of the data points in the window, m =0 corresponds to the center point, m =-1 and m =-2 correspond to the left points, m =1 and m =2 correspond to the right points. m is the convolution coefficient of the Savitzky-Golay filter, which is uniquely calculated by the specified window size and polynomial order, and essentially serves as the weight for the weighted average of the data in the window. The design of these weights allows the smoothed curve to retain the characteristics of the original data trends (such as peaks and valleys) to the maximum extent. is the interval of the horizontal position, i.e. the radial distance between the centers of two adjacent bins.

[0143] (3) Algorithm verification: To verify the reliability of the algorithm and the accuracy of the inversion results, all data in January 2018 (0-60°N, 0-180°E in 2018) were selected for comparison (Argo data in 2018) as evaluation data, and the inversion temperature at the same time and latitude range was compared. The selection of all data in a year is to ensure that the data set covers different time periods and the corresponding ocean temperature conditions in different time periods, and to conduct a comprehensive evaluation.

[0144] BIAS is used to measure the deviation between the inversion results and the true value (Argo data in 2018), and the formula is: ​ ; wherein, is the temperature data of 2018 after inversion (0-60 °N, 0-180 °E), is the Argo observation data in 2018.

[0145] The correlation coefficient and error rate between the automatically extracted vortex equivalent radius and vortex moving speed and the corresponding physical quantities of the mesoscale vortex data in 2018 AVISO, and the correlation coefficient R1 and the error rate Error Rate : ; , is the mean value of , is the mean value of .

[0146] The present application is aimed at the actual demand of high accuracy and real-time of mesoscale vortex three-dimensional structure inversion, based on field, historical and reanalysis data, using regression analysis method to construct the cold / warm mesoscale vortex three-dimensional temperature field inversion model, comprehensively using field observation and satellite remote sensing, realizing real-time and accurate inversion of mesoscale vortex vertical structure, and carrying out field test verification, providing data guarantee for marine activities. Specifically, the OW method is used to identify the edge of mesoscale vortex on the sea surface, and the WA method is used to identify the edge of vortex through flow field in the 5m layer-200m layer below the sea surface. The vortex three-dimensional temperature inversion uses regression analysis method, and a mathematical model is constructed based on multiple historical data, so as to realize the inversion of vortex temperature. The historical data used include objective reanalysis data, Argo historical observation data and satellite data and other types of data. Through in-depth mining and integrated analysis of these rich and representative data resources, a mathematical model is constructed which can accurately represent the sea surface temperature, sea level, sea level anomaly between the sea surface and the three-dimensional temperature field of the sea. The model can accurately predict the three-dimensional temperature distribution of each level vortex. The results show that the constructed mesoscale vortex three-dimensional temperature field inversion model has high accuracy in multiple verification cases, the correlation coefficient is above 0.9, and the mean square error is less than 0.5 °C.

[0147] Based on the same inventive concept, the embodiments of the present application also provide a marine mesoscale vortex three-dimensional temperature field inversion device for implementing the marine mesoscale vortex three-dimensional temperature field inversion method described above. The implementation scheme of the problem solving provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more marine mesoscale vortex three-dimensional temperature field inversion device embodiments provided below can refer to the limitations of the marine mesoscale vortex three-dimensional temperature field inversion method in the above, which will not be repeated here.

[0148] In an exemplary embodiment, as Figure 3As shown, a marine mesoscale eddy three-dimensional temperature field inversion device is provided, which comprises a data acquisition module 301, a mesoscale eddy identification module 302, a model construction module 303, a coefficient solving module 304 and a temperature inversion module 305.

[0149] The data acquisition module 301 is configured to acquire multi-source marine observation data.

[0150] The mesoscale eddy identification module 302 is configured to identify a mesoscale eddy based on the multi-source marine observation data, and determine a three-dimensional structure of the mesoscale eddy.

[0151] The model construction module 303 is configured to construct a mesoscale eddy three-dimensional temperature field inversion model by using a regression analysis method based on the multi-source marine observation data and the three-dimensional structure of the mesoscale eddy.

[0152] The coefficient solving module 304 is configured to solve regression coefficients in the mesoscale eddy three-dimensional temperature field inversion model by using a hierarchical least squares method.

[0153] The temperature inversion module 305 is configured to invert a mesoscale eddy three-dimensional temperature field by using the mesoscale eddy three-dimensional temperature field inversion model according to the regression coefficients, and embed a dynamic constraint in the inversion process to obtain a temperature of each standard layer of the mesoscale eddy.

[0154] In an exemplary embodiment, a computer device can be provided, which can be a server or a terminal, and an internal structure diagram thereof can be as shown in Figure 4 The computer device comprises a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store multi-source marine observation data. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement a marine mesoscale eddy three-dimensional temperature field inversion method.

[0155] Those skilled in the art can understand that, Figure 4The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0156] In an example embodiment, a computer readable storage medium is provided, storing a computer program which, when executed by a processor, implements the steps of any of the above method embodiments.

[0157] In an example embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of any of the above method embodiments.

[0158] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0159] A person of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium and can include the processes of the above method embodiments when executed. Any reference to a memory, database or other medium used in the embodiments provided by the present application can include at least one of a non-volatile and volatile memory. The non-volatile memory can include a read-only memory (ROM), a magnetic tape, a floppy disk, a flash memory, an optical storage, a high-density embedded non-volatile memory, a resistive memory (ReRAM), a magnetoresistive memory (MRAM), a ferroelectric memory (FRAM), a phase change memory (PCM), a graphene memory, etc. The volatile memory can include a random access memory (RAM) or an external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms such as a static random access memory (SRAM) or a dynamic random access memory (DRAM), etc.

[0160] The database involved in each embodiment provided by the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, and the like, without being limited thereto. The processor involved in each embodiment provided by the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, and the like, without being limited thereto.

[0161] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present disclosure.

[0162] The principles and implementation modes of the present application are described by applying specific examples herein, and the above embodiments are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In conclusion, the content of the present specification should not be understood as a limitation of the present application.

Claims

1. A method for retrieving three-dimensional temperature field of mesoscale eddy in the ocean, characterized in that, The method comprises: acquiring multi-source marine observation data; based on the multi-source marine observation data, identifying mesoscale eddies to determine the three-dimensional structure of the mesoscale eddies; based on the multi-source marine observation data, according to the three-dimensional structure of the mesoscale eddies, constructing a three-dimensional temperature field inversion model of the mesoscale eddies by using a regression analysis method; solving the regression coefficients in the three-dimensional temperature field inversion model of the mesoscale eddies by using a hierarchical least squares method; according to the regression coefficients, inversing the three-dimensional temperature field of the mesoscale eddies by using the three-dimensional temperature field inversion model of the mesoscale eddies, and embedding a dynamic constraint in the inversion process to obtain the temperature of each standard layer of the mesoscale eddies; the multi-source marine observation data comprises a buoy temperature and salinity data set, satellite altimeter data, a marine climatic data set, and drift-type sea-air interface buoy observation data; based on the multi-source marine observation data, according to the three-dimensional structure of the mesoscale eddies, constructing a three-dimensional temperature field inversion model of the mesoscale eddies by using a regression analysis method, comprising: determining sea surface height anomaly data according to the satellite altimeter data; determining sea surface thermal anomaly data according to the buoy temperature and salinity data set, the marine climatic data set, and the drift-type sea-air interface buoy observation data; determining thermal-dynamic coupling data according to the sea surface thermal anomaly data and the sea surface height anomaly data; constructing a three-dimensional temperature field inversion model of the mesoscale eddies by using a regression analysis method, taking the sea surface height anomaly data, the sea surface thermal anomaly data, and the thermal-dynamic coupling data as independent variables, and taking the temperature of each standard layer of the mesoscale eddies as a dependent variable; the three-dimensional temperature field inversion model of the mesoscale eddies is: ； wherein, is the temperature at the standard layer of depth , is the sea surface thermal anomaly data, is the sea surface height anomaly data, is the thermal-dynamic coupling data, is a random error term, is the vertical transfer efficiency of the sea surface thermal anomaly to the deep water, , , , is a regression coefficient, represents a linear response coefficient of the sea surface thermal anomaly, represents a linear response coefficient of the sea surface height anomaly, represents a nonlinear response coefficient of the sea surface thermal anomaly, represents a thermal-dynamic coupling response coefficient.

2. The mesoscale oceanic eddy three-dimensional temperature field inversion method according to claim 1, characterized in that, based on the multi-source marine observation data, identifying mesoscale eddies to determine the three-dimensional structure of the mesoscale eddies, comprising: for the sea surface area, identifying mesoscale eddies by using a parameter method based on the satellite altimeter data to determine the three-dimensional structure of the mesoscale eddies in the sea surface area; for the area below the sea surface, identifying mesoscale eddies by using a wrapping angle method based on the satellite altimeter data to determine the three-dimensional structure of the mesoscale eddies in the area below the sea surface.

3. The mesoscale oceanic eddy three-dimensional temperature field inversion method according to claim 1, characterized in that, The thermal-dynamic coupling data is the product of the sea surface thermal anomaly data and the sea surface height anomaly data.

4. The mesoscale oceanic eddy three-dimensional temperature field inversion method according to claim 1, characterized in that, The dynamic constraint comprises a quasi-geostrophic equilibrium constraint and a sea surface boundary condition.

5. A device for retrieving three-dimensional temperature field of a mesoscale oceanic eddy, characterized in that, The device is applied to the method for inversing the three-dimensional temperature field of the marine mesoscale eddies according to any one of claims 1-4, and the device comprises: a data acquisition module for acquiring multi-source marine observation data; a mesoscale eddy identification module for identifying mesoscale eddies based on the multi-source marine observation data to determine the three-dimensional structure of the mesoscale eddies; a model construction module for constructing a three-dimensional temperature field inversion model of the mesoscale eddies by using a regression analysis method based on the multi-source marine observation data and according to the three-dimensional structure of the mesoscale eddies; a coefficient solving module for solving the regression coefficients in the three-dimensional temperature field inversion model of the mesoscale eddies by using a hierarchical least squares method; a temperature inversion module for inversing the three-dimensional temperature field of the mesoscale eddies by using the three-dimensional temperature field inversion model of the mesoscale eddies according to the regression coefficients, and embedding a dynamic constraint in the inversion process to obtain the temperature of each standard layer of the mesoscale eddies.

6. A computer device comprising: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes the computer program to implement the mesoscale eddy three-dimensional temperature field inversion method in any one of claims 1-4.

7. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the mesoscale eddy three-dimensional temperature field inversion method in any one of claims 1-4.

Images